Sounds dramatic? In this article, we are going to explore the modern day equivalent of the mythical Sirens: a strategy that wins 990 times out of 1000 with tiny drawdowns and feels like free money. The equity curve is smooth, the win rate is near perfect, and the backtest screams positive expectancy.

…yet if you only have ten losing trades, you know very little about how often you really lose and how big those losses can be. Your estimate of expected value is mostly wishful thinking, no matter how pretty the curve looks.

Problem outline

The problem is not “high win rate is bad”. The problem is that traders treat a high win rate with very few losses as hard evidence of edge, when in reality it means:

• You have very sparse data about the frequency of large losses.
• You have very sparse data about the magnitude and structure of those losses.
• You have almost no information about how those losses interact with path dependence, margin and volatility.

Expected value depends on both the probability of losing and how much you lose when you do. When the losing side of the distribution is barely sampled, your backtest is underpowered exactly where it matters most.

High win rate plus few losers is not a sign of robustness. It is a sign that the backtest tells you almost nothing about the only trades that can blow you up.

Expected value is a joint object, not a win rate

Expected value is just the probability weighted average of outcomes. For trading, think of it as:

• How often you win.
• How much you win on average.
• How often you lose.
• How much you lose on average.

The backtest is supposed to estimate all four. High win rate strategies tend to compress the first three and hide the fourth:

• Win rate looks very high.
• Average win is small and stable.
• Losses are infrequent, irregular and often state dependent.

That means your estimate of expected value is sitting on top of two fragile quantities:

1. The true probability of loss.
2. The true loss size distribution.

If you only have 10 losing trades in 1000, there is enormous uncertainty around both. You can pretend the “average loss” is stable, but statistically you are extrapolating from almost no data.

Two separate failure modes

There are two ways traders fool themselves.

1. They trust the win rate but ignore that the loss size is essentially unknown.
   This is the classic short volatility or naked options trap. The strategy wins constantly in normal markets. When volatility explodes, losses jump to levels that never appeared in the backtest. Margin models like SPAN explicitly warn about deep out of the money short options creating huge losses in extreme moves, yet these scenarios are rarely simulated in retail testing.
2. They trust the loss size but pretend the loss frequency is well known.
   This shows up with defined risk strategies or tight stops. Maximum loss per trade appears bounded, so traders assume the “bad case” is fully understood. In reality, a small change in loss frequency can flip expected value from positive to negative. Empirical work on edges in trading shows that even modest changes in win rate or payoff ratio can completely reverse profitability, especially after costs.

The point is simple. If you only understand one side of the losers, you do not understand expected value.

Sparse losers mean EV is structurally underidentified

With 10 losers in 1000 trades:

• Your estimate of how often you lose is based on 10 data points.
• Your estimate of how big you lose is based on those same 10 data points.
• Your estimate of the worst case is based on 1 data point.

Any serious expected value calculation that pretends those numbers are precise is lying to you. Practical guides to options EV make this clear. Expected value is sensitive to both probabilities and payoffs. When payoffs are fat tailed and probabilities are low, small errors in either completely dominate the calculation.

You are not just underestimating risk. You are overestimating how certain you are about your expected value.

Path dependence makes “we have a crash in the data” meaningless

Even if your test period includes 2008, 2020 or August 2024, you are not done. For path dependent structures like options portfolios, the outcome of a volatility spike depends heavily on:

• Days to expiration when the shock hits.
• How far the underlying is from your short strikes at that moment.
• How many positions you have on, and at what sizing, because high win rate encourages scaling.
• How your broker’s margin model responds to that exact combination.

A volatility expansion when your short puts are 60 days to expiry is a different event than the same move at 20 days. At higher DTE, vega dominates. At lower DTE, gamma and assignment risk explode. The same calendar crash can produce very different PnL paths depending on where your trades sit along the time axis.

So “the backtest includes a crash” means very little if it does not include crashes across many different DTE and position states. With few losers, you generally have one or two specific crash sequences that happen to line up with your rules. That is not robust information. It is a story about one particular path.

Example: 1–1–2

The 1–1–2 put ratio is a clean case study of how this all goes wrong.

Structure and payoff

The classic version on an index future or ETF is:

• Buy 1 out of the money put.
• Sell 1 lower strike put.
• Sell 2 even lower strike puts, usually far out of the money.

You receive a net credit. There is:

• No upside risk. If the market rallies, the structure expires worthless and you keep the credit.
• A wide zone where a gentle drift down yields profit.
• Undefined downside risk from the two extra short puts.

On historical data, especially in grindy bull markets, this looks fantastic:

• High win rate, often 95 to 100 percent winners in some backtests.
• Smooth monthly income.
• “Max loss” that rarely shows up in the sample.

The 1–1–2 is explicitly marketed as a high probability income strategy. Some practitioners report extended periods with no recorded losing trades in certain underlyings and DTE windows.

Where expected value uncertainty explodes

Look at it through the EV lens.

• Average win size is small and well sampled. You collect similar credits again and again.
• Loss frequency appears very low, but that estimate is based on few events.
• Loss magnitude is effectively unknown, because the combination of gap size, IV spike, DTE and margin response that produces a real disaster has not occurred often.

So your expected value estimate uses:

• A precise number for average win.
• A very optimistic estimate of how often you lose.
• A massively understated estimate of how much you lose when the path goes wrong.

That is how traders end up with a backtest that shows a steady equity curve and a “great” EV, then take a 30 to 100 percent hit in a single cluster of trades when volatility finally expands the wrong way. Public case studies of 1–1–2 traders turning mid five figure accounts into six figures and then vaporizing the gains in a few days during a volatility event are exactly this dynamic playing out.

DTE and path dependence

Now add your point about DTE.

Suppose you always enter 1–1–2s at 45 to 65 DTE, which is common. The behavior under stress depends a lot on when the shock hits:

• If volatility explodes early, when your short puts are still 40–60 DTE, the pain is mostly vega and margin expansion. The entire structure marks against you, but the long put spread has meaningful downside sensitivity and the short puts are usually far from the money. Margin increases, but you still have time, distance to strike, and a functioning bearish hedge via the long put’s delta. It is ugly, but usually survivable if you are not oversized.
• If volatility explodes late, when your short puts are 10–25 DTE, the profile is far more dangerous. Vega is mostly gone and the book is dominated by delta and gamma. A sharp move down can push the underlying quickly through your short strikes, so the single long put is fighting two short puts on a mostly intrinsic basis. The ratio and strike spacing, not theta, now define how much protection you really have. Margin requirements spike exactly as losses accelerate and assignment risk becomes real. This late-cycle vol expansion is structurally the worst case for a 1–1–2 and is what you should size and plan for.

A backtest that “contains” a particular crash only samples one of those combinations of DTE, distance to strike and open size. It does not tell you how the same shock would treat a similar book at different points in the cycle.

This is why simply saying “my backtest includes 2020, so I am safe” is delusional for path dependent short volatility structures. The state of your book at the moment of stress is as important as the stress itself.

Detection mechanisms

You cannot eliminate uncertainty, but you can quantify how unreliable your expected value estimate actually is.

1. Negative sample diagnostics

Treat the losing side of the distribution as a separate object:

• Count the number of losing trades.
• Count how many are “large” relative to average winners.
• Count how many occur in distinct volatility and trend regimes.

If you have single digit or low double digit losses, your estimates of loss frequency and loss magnitude are both extremely weak. That is your first red flag.

2. Bayesian intervals for loss frequency

Instead of saying “my loss rate is 1 percent”, treat it as a belief with uncertainty.

A simple Bayesian view:

• Start with a prior that does not assume absurdly small failure probability.
• Update with observed wins and losses.

With 990 winners and 10 losers, you will get a posterior range for the true loss rate that is wider than people like to admit. That range should feed directly into a family of expected value estimates:

• Best case EV if the true loss rate is at the low end.
• Worst case EV if it is at the high end.

If modest shifts in assumed loss probability flip your EV from strongly positive to clearly negative, your strategy is fragile. That fragility is essential information.

3. Bayesian or robust estimates of loss magnitude

Do the same for loss size:

• Do not assume the worst loss in your backtest is the worst loss possible.
• Model loss sizes as a distribution, not a single number.

A simple and practical approach:

• Treat observed losses as draws from a heavy tailed distribution.
• Use conservative estimates of the upper tail, for example by focusing on conditional average of the worst 5 or 10 percent of losses rather than the single worst.

Option risk guides and margin documents repeatedly highlight that deep out of the money short options can show minimal risk in normal scenarios but explode in extreme events. That is a hint that backtested loss sizes are systematically biased downward for those trades.

The takeaway: treat your estimate of the loss tail as “at least this bad, likely worse” unless you have a sufficient number of negative samples.

4. Regime and DTE conditioning

You also need to understand how losses depend on state:

• Break your trades by volatility regime.
• Break them by DTE buckets at entry and at exit.
• Look at loss frequency and magnitude within each bucket.

For a 1–1–2:

• Losses around big downside moves with 20 to 30 DTE remaining are not the same phenomenon as small marking losses during mild pullbacks at 60 DTE.
• If your big losses cluster in a specific corner of the state space, your EV is heavily exposed to how often you visit that corner in future.

This is where you see that “we survived 2020 in the backtest” might be because your rules were rarely loaded with short premium in the wrong DTE window when the crash hit. That is not a guarantee. That is luck.

5. EV sensitivity analysis

Finally, treat EV as a function of inputs and stress those inputs:

• Vary the loss probability within a realistic interval derived from your Bayesian posterior.
• Vary the loss magnitude by assuming fatter tails than observed.
• Include realistic transaction costs and slippage.

Research and practitioner writeups on option EV show that once you incorporate costs, many high win rate structures move from mildly positive to flat or negative EV.

If your strategy’s EV collapses under conservative input shifts, the backtest is not a foundation. It is marketing material. And if that’s the bar, a certain Mr. Madoff would be happy to lend you a hand ;)

Fixes you can apply

1. Upgrade your mental model of the strategy

You need structural understanding before statistics.

For a 1–1–2 that means knowing in advance:

• How PnL responds to price, volatility and time.
• How margin responds to different risk scenarios.
• How DTE changes your vega and gamma exposure.

SPAN style frameworks literally simulate a range of price and volatility changes to estimate one day worst loss. You do not need that exact engine, but you need your own simplified version of it for the positions you trade.

2. Incorporate scenario and path testing

Backtests that only replay historical daily closes are not enough.

Add scenario tests that:

• Shock price by more than anything seen in your sample.
• Shock implied volatility to crisis levels.
• Shift the timing of those shocks across different DTE.

Then:

• Reprice your positions under those paths.
• Recompute margin.
• See how many years of historical profits vanish in one sequence.

This directly addresses the joint uncertainty in loss frequency and magnitude. You are extending the data on the losing side in a structured way, rather than naively trusting the limited history.

3. Use conservative EV for sizing, not optimistic EV

Instead of sizing off the backtest’s point estimate of EV:

• Take a conservative combination of higher loss probability and larger loss size from your sensitivity analysis.
• Compute the implied EV under those assumptions.
• Size based on that conservative EV, not the best case.

This usually cuts your position size and leverage dramatically for high win rate structures. That is the point. You want to survive being wrong about your loss process.

4. Prefer defined or capped risk variants when possible

If the structural risk is too messy to model, cap it.

For 1–1–2 style trades, that can mean:

• Adding extra long puts to cap the downside.
• Using narrower widths or smaller ratios that reduce naked exposure.

You will sacrifice some income and win rate. You will often improve the robustness of expected value, because the loss magnitude distribution becomes far better behaved and easier to reason about.

5. Extend the dataset intelligently

You cannot manufacture crashes, but you can:

• Test across more underlyings with similar structure.
• Run walk forward tests over multiple non overlapping windows.
• Include recent stress periods explicitly and examine how your rules would have behaved given realistic DTE and exposure.

The objective is to build a larger, more diverse set of negative samples and stress behaviors. You are not trying to fit the past better. You are trying to see more ways the strategy can fail.

Best practices: when to trust and when not to trust a backtest

You should distrust a backtest, regardless of win rate, when:

• The number of losing trades is small relative to total trades.
• Losses show strong dependence on volatility and DTE, but you have not explicitly modeled this.
• The expected value estimate swings wildly when you nudge loss probability or loss size within reasonable ranges.
• Margin, slippage and liquidity are treated as afterthoughts instead of primary risk drivers.

You can place conditional trust in a backtest when:

• You have a substantial sample of losses across different regimes and DTE states.
• The loss distribution is structurally capped, or at least tightly constrained by design.
• Expected value remains positive across conservative assumptions about loss frequency and magnitude.
• Path and margin dependence are explored through scenario testing, not just hand waved.

Backtests are not proofs that a strategy is safe or profitable. At best, they are one noisy estimate of expected value and should be treated as such. When information about both the frequency and magnitude of losses is sparse, especially in path dependent setups like 1–1–2s, the backtest tells you almost nothing about the trades that can end your trading career.

The only rational response is to treat pretty equity curves and high win rates as a starting point, then attack the loser side of the distribution until you are convinced that your uncertainty about expected value is small enough to risk real money.

Just because you throw math at something, doesn’t mean you can rely on it. At the end of the day, there is no free lunch. And when a strategy seems to make you money but you don’t really understand why, you’re likely paying a price for it you’re not even conceiving.

Disclaimer: The information provided is for general informational and educational purposes only. It should not be considered financial, investment, or trading advice. Markets are unpredictable, and past performance is not indicative of future results. Always conduct your own research and consult with a qualified professional before making financial decisions.

References

OptionsTradingIQ, Tom King 1–1–2 Income Strategy, 2025.
DataDrivenOptions, 1–1–2 Put Ratio Trade, 2023.
ThetaProfits, 112 Options Trading Strategy: Profit Potential and Risks, 2025.
CME Group, CME SPAN Methodology Overview, 2025.
CME Group, SPAN Methodology, 2019.
NSE Clearing, NSCCL SPAN, 2025.
Option Alpha, How to Calculate Expected Value in Options Trading, 2023.
Quantpedia, Volatility Risk Premium Effect, 2025.
BuildAlpha, Edge in Trading: Professional Guide to Trading Edges, 2025.

Many traders begin their journey with moving average (MA) crossovers. The idea seems intuitive: when a short-term average crosses above a long-term average, it must signal an uptrend; when it crosses below, it must signal a downtrend. Yet from a signal-processing standpoint, this strategy has deep flaws. Unless you understand the mathematics behind moving averages and the nature of the market you are trading, crossovers are more likely to mislead than to generate reliable profits.

Let’s break down the problems, the math behind lag, and some possible refinements that make sense in certain markets.

The Problem of Lag

Every moving average is a smoothing filter. It reduces noise in the price but does so by looking back at a window of past data. That backward-looking nature introduces lag.

• For a simple moving average (SMA) with length N, the effective lag is about N/2 periods.
• A 20-period SMA lags price by roughly 10 bars.
• A 200-period SMA lags by about 100 bars.

When you compare two moving averages, for example a 20 versus 200, the crossover event happens only after both lagged signals finally converge. The trigger therefore lags real market turning points by nearly the full delay of the slower average. In our example, you are effectively trading a signal that occurs about 90 to 100 bars late.

What a Crossover Actually Implies

If you use a moving average crossover as both entry and exit, you are making an implicit assumption: that the market is moving in a cycle whose period is at least twice the lag of your signal.

Here is why:

• Your entry comes after a lag of about L.
• Your exit also comes after a lag of about L.
• To capture any part of the move, the cycle length of the market must be at least 2L.

For example, with a 20/200 crossover, if the slower leg lags by around 100 bars, then you need the underlying cycle of price swings to be at least 200 bars long. Otherwise, the trade will either capture too little of the move or miss it entirely.

That assumption is rarely justified if you only rely on the moving averages themselves. This is why pure crossover systems are unreliable: they presuppose structure in the market that may not be there.

SMA, EMA, and Filter Alternatives: A Lag Perspective

Traders often switch from SMAs to EMAs because EMAs react faster. While this is true, the improvement is limited.

• SMA: Uniform weights, lag ≈ N/2. Example: SMA(20) ≈ 10 bars, SMA(200) ≈ 100 bars.
• EMA: Exponentially decaying weights, lag ≈ N/3. Example: EMA(20) ≈ 6–7 bars.

Both are low-pass filters: they smooth data but always introduce lag. Other filters offer better trade-offs:

• Low-pass filters (Butterworth, SuperSmoother): Smoother than SMA/EMA with ~30–50% less lag at the same cutoff frequency.
• High-pass filters: React almost instantly (lag ≈ 0–1 bar), but are very noisy. Useful for cycle detection, not trend following.
• Roofing filters (high-pass + low-pass): Isolate tradable mid-frequency cycles. Typical lag is just 1–3 bars relative to the cycle, far less than SMA or EMA.

In short:

• SMA: blunt, heavy lag.
• EMA: faster but still delayed.
• Optimized low-pass: less lag for same smoothing.
• High-pass: near-zero lag, but noisy.
• Roofing: best balance, low lag with cycle focus.

This shows why simple moving averages are such poor tools: there are filter designs that deliver smoother outputs with far less delay.

When Crossovers Can Work: Cyclical Assets

The story changes if you trade markets with inherent cyclicality. Unlike broad stock indices or individual equities, many commodities display strong, regular cycles tied to physical supply and demand.

• Agricultural commodities such as corn, wheat, and soybeans have planting and harvest seasons.
• Energy commodities such as natural gas and heating oil follow weather-driven consumption cycles.
• Industrial commodities such as copper and aluminum often track inventory and manufacturing cycles.

Because these assets have recurring rhythms, it is more reasonable to expect dominant cycles of a certain length. If the cycle period is significantly larger than your crossover lag, the strategy can sometimes capture usable signals. In other words, context matters. What is nonsense on an equity index may be viable in soybeans.

How to Estimate Market Cycles

The key refinement is to only trade a crossover when the current dominant cycle of the market is longer than your signal’s round-trip lag. To do this, you need tools to estimate the period length of price oscillations.

Two common approaches:

1. Autocorrelation

• Measure how the price correlates with lagged versions of itself.
• Peaks in autocorrelation indicate repeating structures.
• From this, you can infer likely cycle lengths.

2. Hilbert Transform

• A signal-processing tool that extracts the instantaneous phase and frequency of a time series.
• By tracking the change in phase, you estimate the instantaneous period of the price cycle.
• John Ehlers pioneered its use in trading, showing how to build adaptive filters that respond to the market’s dominant cycle in real time.

With these methods, you can decide whether the crossover is worth acting on. If the estimated period is at least twice your lag, the trade has theoretical justification. If not, ignore the signal.

Adaptive Filtering: Beyond Static Averages

Ehlers’ work provides a more robust way to handle market signals. Instead of fixed-period averages, adaptive filters adjust to the dominant cycle of the data. Examples include the MESA Adaptive Moving Average (MAMA), the Autocorrelation Periodogram (ACP), and roofing filters.

The advantage:

• When the market cycle shortens, the filter speeds up.
• When the cycle lengthens, the filter slows down.

This adaptive approach minimizes lag while keeping noise reduction. It also directly addresses the flaw of arbitrary lookback periods.

Practical Takeaways

1. Pure crossovers are flawed. The lag is too large, and without context, you are trading ghosts.
2. EMA versus SMA matters, but only at the margin. EMAs lag less, but they do not fix the structural problem.
3. From a lag perspective: SMA lags by half its length, EMA by about one third, optimized low-pass filters by 30–50% less, high-pass by almost zero, and roofing filters by just 1–3 bars. High and low pass filters are objectively superior to SMA/EMA by design.
4. Know your asset. Commodities are more cyclical than equities. A soybean crossover may have justification where an S&P 500 crossover does not.
5. Measure cycles. Use autocorrelation or Hilbert transforms to check whether the dominant cycle period exceeds your crossover lag.
6. Consider adaptive methods. John Ehlers’ adaptive filters are worth studying if you want to keep lag under control while staying aligned with market structure.

Further Reading

For the interested, I can warmly recommend John Ehler’s books on the topic. They do require some enthusiasm for the math to fully appreciate the content but are absolutely worth it. He delivers a fresh, grounded and well explained perspective on the topic you will rarely find in the trading guru scene, for example. And for the dedicated searcher, they are available for free on the internet.

Conclusion

Moving average crossovers have become a rite of passage for new traders. They are simple, visual, and promise trend detection. But under the hood, they are crippled by lag. Trading them blindly assumes that markets cycle in predictable rhythms that are at least twice as long as your signal’s delay. In equities, this assumption rarely holds.

In commodities and other cyclical assets, however, the logic may have a foundation, provided that you confirm the dominant cycle length supports it. Adaptive cycle estimation tools such as autocorrelation, Hilbert transforms, and advanced filter designs allow you to separate valid signals from noise.

The lesson is clear: moving average crossovers are objectively inferior to other filters smoothing out lag. There is no good reason using them over something like a roofing filter. And even with those superior filters, you still need the right assets and the right context confirmed by other indicators.

If you liked this article, you are probably going to love this one. It’s about the most effective indicator I found in 5 years of full time trading, which is giving me the most reliable and high value signals.

If you found this information useful, feel free to leave a comment or clap for it! Thank you for your attention and have a great day.

Disclaimer: The information provided is for general informational and educational purposes only. It should not be considered financial, investment, or trading advice. Markets are unpredictable, and past performance is not indicative of future results. Always conduct your own research and consult with a qualified professional before making financial decisions. Links displayed might be affiliate links and I might earn a commission if you purchase through them.

It’s long known and well documented how flows in the options market contain information with predictive value for the underlying assets, but with the advent of 0 DTE trading, those flows are becoming a powerful tool, even for daytrading.

This is about seeing support and resistance BEFORE they form, before price has even reached the levels. Stay with me until the end of the article and I promise you won’t regret it.

Options? What’s that?

Options are my favourite derivative by far. They are like the son I always wanted, but never had. Not really, but you get the idea.

They are incredibly rich in value and versatility and can be used for detailed risk management. If you’re totally unfamiliar with them, I’d recommend diving deeper into the topic as it will make you a better trader overall. I will give a brief, surface level breakdown here, so you get the gist. If you’re already familiar with options as a derivative and their inner workings, feel free to skim ahead!

Options basically work like insurance: if you purchase an option, you have:

• the right, not the obligation (this is important!)
• to purchase (=Call options) or sell (=Put options) the underlying asset (e.g. SPY)
• at a defined price, called strike price (e.g. $600)
• up to a certain point in time (the expiration date).

Here, 1 options contract usually controls 100 shares of the underlying in the case of stocks, ETFs or indices. That means you purchase 1 option and gain the right to buy/sell 100 shares, not just 1.

You can now maybe already see how that is useful: if you think the market might collapse tomorrow but you are not sure, you might want to just purchase the OPTION to sell at the prices of today, even when the market drops tomorrow. In that case, you would want to buy a PUT option with a strike price around today’s asset price and an expiration date far enough out to cover the period you’re worried about.

Without going into too much detail here, there’s models that tell you how much an options contract should cost, based on factors like stock price, time to expiration and volatility expectations. Out of those models come the options price and some characteristic values for the options contract called ‘greeks’, namely delta, gamma, vega, theta and rho being the most popular ones.

To fully understand how and why the methods I am going to describe work, it’s useful to read up on those and understand where they are coming from.

A very brief overview over the 2 main ones we are going to be using would look something like this:

Delta

• the rate of change of the options price given a $1 change in the underlying,
• e.g. delta of +0.5 means if the underlying moves by $1, the option price changes by $0.50
• Think of this as equivalent in stock shares: a 0.50 delta option is equivalent to directional exposure of 50 shares of stock.

Gamma

• the rate of change of Delta (yes, the one we just talked about!) given a $1 change in the underlying
• if the underlying moves by $1, delta pushes its price and gamma pushes its delta,
• e.g. with a gamma of 0.03, it might now have a delta of only 0.47, reducing how much the option price reacts to price changes in the underlying (or increasing)
• Gamma is higher, the closer the strike price to the price of the underlying. This relation intensifies as the option gets closer to expiration.

For completeness sake: vega is the sensitivity towards volatility expectations, theta is the option price change depending on time value and rho is related to interest rates. The interplay of all those is defined in models like the Black-Scholes-Merton model via partial differential equations.

Understanding all that is NOT required to make use of the practical implications of it, though and is just provided for the interested as additional context.

One more thing that comes in handy to know: the further in-the-money an options is, the higher its delta. In-the-money simply means that the option is intrinsically worth something: a call option has intrinsic value when the strike price is lower than the current underlying price, since it gives you the option to buy the stock at that lower price. A put option has intrinsic value if its strike price is higher than the current underlying price since it gives you the option to sell at that higher price. The more intrinsic value an option has, the higher its delta.

Now you are equipped with the basics to get the gist of what we’re talking about and can officially call yourself a finance geek.

Delta-Gamma-Hedging

Market makers buy and sell all the time, providing liquidity for people to trade against. That’s why you don’t have to wait until someone somewhere is wiling to trade with you but you can trade against the market maker immediately instead.

While doing that, they don’t want exposure of their own to not be exposed to price risk. They make money off the bid/ask spread and not price moving. This is called ‘staying delta neutral’.

In order to achieve that, they need to hedge, aka balance out the exposure that they took on. If you buy a stock from a market maker, they are implicitly short the stock now, since they sold it to you. So they need to get long exposure from somewhere to balance things out. That might not always happen immediately and a small temporary exposure is fine. Maybe someone else wants to sell a stock now so they get rid of their implicit short like that! But if their inventory is becoming too imbalanced, they will hedge.

Imagine a market maker has a delta exposure of +0.5 that they want to get rid of and a gamma of +0.03. Since the multiplier of 100 applies, they sell 50 shares and are now delta neutral … right? Maybe! Maybe not.

What happens if price moves? We know gamma effects delta then, so the market maker is not fully hedged anymore. Now, their delta might be 0.53 (increased by the +0.03 gamma) and they are not fully hedged. They need to hedge again, because of the gamma. This is called delta-gamma-hedging.

Take note of the fact that their very selling of the 50 shares for hedging initially might also move price already, which can fuel a feedback loop.

Putting it to Practice

Now you know that holding an options inventory necessitates continuous hedging when price moves.

If we look at market makers as a whole and we see their inventories combined, we can also understand that depending on where price moves, it might trigger more or less directional hedging pressure, either bullish or bearish.

What you see above is the net delta exposure by strike. We arrive at it for a given strike by multiplying the put option’s delta by its open interest and the call option’s delta by its open interest respectively and adding them up.

A simple example: at strike S, there are 100 put options with a delta of -0.5 and 110 call options with a delta of +0.5. So the net exposure there is: (100*(-0.5)) + (110*0.5) = 5 delta.

The problem is … that this information is practically useless. Directional hedging is done fast and we can almost never make trades based on these flows.

What we can however use is the gamma exposure, which does not show directional exposure from delta but instead sensitivity to price changes from gamma, telling you about the intensity of the hedging flows to anticipate.

Because if price moves by just $1, ALL of the market maker’s options change in delta, requiring them to re-hedge.

We can simply build the gamma exposure by strike in the same manner as the delta ladder and it’s called the gamma ladder or convexity ladder:

Now imagine the majority of their inventory forces the market makers to SELL in order to hedge when price moves DOWN … a vicious cycle starts where they sell into dips and buy into rips! This is called negative gamma because it occurs when the net overall gamma exposure of the market maker is negative.

Conversely, in a positive gamma environment, market makers buy into weakness to hedge and sell into strength. This dampens volatility and stabiblises price.

This exact dynamic of:

• negative gamma causing hedging flows to reinforce price moves, making it gain momentum and
• positive gamma causing hedging flows to dampen price moves, causing it to lose momentum

is present intraday as well and THAT is where the value lies!

Since the reason for this is not so easy to grasp without a deeper understanding of options, I went ahead and built a little playground for you to experiment with and explore this visually.

You can set the price of the underlying, a put and a call option and visually understand how price changes effect trader’s exposure, market maker’s exposure and resulting hedging pressure:

Gamma playground 2

You can clearly see how gamma (dotted) is shaped like a mole hill and how delta exposure is sigmoid in shape relative to the underlying price.

On a side note: this dynamic of gamma related hedging flows is what majorly contributed to the gamma squeeze in GME (GameStop), so it’s great to appear smart with on parties!

What We Can’t See

While knowing the gamma exposure by strike from multiplying option gamma by open interest and netting out puts and calls is nice as a starting point, there is a problem with it. Open interest is only released once per day and not continuously updated. That means we know where we stand at the open, but from then on, we are blind.

What I did not tell you so far is that an options delta is more sensitive to changes (higher gamma) when it is closer to expiration. This makes sense intuitively if you consider that an option’s value is mainly composed of two factors: the underlying price right now and the expectation of volatility until it expires. If it has little time left to expire, volatility expectations tend to not matter as much anymore and price of the underlying becomes a much larger factor. Thus its sensitivity to price moves — aka gamma — increases.

This becomes even more important when we take into consideration that the share of 0DTE option trading volume (options that expire on the same day) has dramatically increased over the last years to well over 50%! That means there is a LOT happening intraday which we need to be aware of, otherwise we are trading on old information. Luckily, there is a solution.

Intraday Order Flow Classification

The question to solve is this:

How can we know which exposures market makers have at any given point in time?

This is where order flow classification comes in. See, we cannot directly observe the market maker’s inventory but we can infer it by what we can see: options trades taking place in real time.

There reasoning used is to look at the price at which the trade was made: if it’s closer to the bid, we assume it’s a sell, if it’s closer to the ask we assume it’s a buy. This is not perfect or correct 100% of the time but a reasonably good assumption to work with, with 80–90% accuracy.

So if we see a transaction for $1.00 when the option was quoted at $1.00/$1.01 before, we can infer the trader went short on the option and the market maker now implicitly is long. That way, we can update our open interest based gamma exposure dynamically throughout the trading day and work with up-to-date data!

Putting It All Together

You’ve made it far already. Let’s take a deep breath before we head into the final moments of your alpha-less life.

Inhale.

Exhale.

Good. Now, let’s continue.

For illustration and visual aid, let’s bring back the gamma ladder or convexity ladder from before:

How to make money with this now? It’s less about immediate direction but more so about the ‘slope’ of the playing field. Simply:

In a falling volatility environment (most common, calm markets):

• negative convexity attracts accelerates momentum when price is moving towards it
• positive convexity stalls price when it’s moving towards it

In a rising volatility environment (panic onset):

• negative convexity stalls
• positive convexity accelerates momentum

A simple look at VIX or net overall gamma exposure can tell you in which regime we are.

What Makes It Special

This is the single most powerful indicator I have come across in my trading career so far, which gives signals tradable manually by humans.

What is so special about it is that it is not derived from some technical analysis, moving average, Ley Line astrology mix — but on real, cold, hard, real-time cash flows. There is no subjectivity in the data, no random parameters to tune for. It simply shows you what will need to happen regarding market maker’s hedging flows, when price moves. That’s as pure of a signal as you can get.

As you can maybe see, I am a big fan of option flow based metrics. Personally, I have had great results with those insights. There are also plenty of scientific papers documenting the dynamics outlined in this article are playing out in practice. Few of them are focused on intraday trading so far though, as the huge increase in popularity of 0 DTE options is a more recent one. That’s why I’m working on a study showing the predictive power contained in those signals in particular for short time frames.

There’s More

So how do you get access to this information in real time now? Do you need to get a data feed, write optimised code to process the real time streaming and then aggregate information from index and index futures options to get useful results? Thankfully not.

There is a platform that does all this for you with exceptional grace: GEXBot. I have personally used them for almost a year now and couldn’t be happier with the results. They even integrate their information as indicators into popular charting platforms like SierraChart, MotiveWave, NinjaTrader and ATAS so you don’t have to switch contexts and have everything in one place. On top of the tools we discussed, they offer many more similar ones, all derived from real options flows, equipping you with a well rounded suitMy .

Beyond that, they offer a community Discord with many knowledgeable people helping you understand further and help with any questions. To me, it’s a really rare sanctuary of knowledge that has greatly improved my trading.

If you want to give them a try, I have even better news for you:

You get 50%(!!) off on your first payment by using code WEALTHACADEMY!

Neither the article or me are sponsored by them, so you can be certain that everything I said is my full and honest opinion.

Feel free to check them out here: gexbot.com/?via=wealthacademy

Closing Notes

I work on delivering the insights to you that I wish I had when I started trading, so you have it easier.

If you found value in this article, that makes me really happy. You can show me by leaving a clap or a comment! Also should you have any other topics you’d like me to write about, you can let me know too.

Thank you for your attention and have a great day!

Disclaimer: The information provided is for general informational and educational purposes only. It should not be considered financial, investment, or trading advice. Markets are unpredictable, and past performance is not indicative of future results. Always conduct your own research and consult with a qualified professional before making financial decisions. Links displayed might be affiliate links and I might earn a commission if you purchase through them.

Explore the science behind why making money in scalping feels so addictive — and how your brain’s reward chemicals can lead you to overconfidence, risk addiction, and ultimately, self-sabotage.

Trading isn’t just a mental game. It’s also a physiological one. Every decision you make during the day is filtered through your body’s chemistry. Most traders ignore this, but the hormones released during scalping can explain everything from hesitation at the open to reckless oversizing after a winning streak.

The Hidden Cycle: How Hormones Shape Your Edge — and Your Downfall

Every tick in your favor triggers a chemical cascade that most traders never even notice. Win → dopamine and testosterone spike → feels good → trade more. Trade more → cortisol drops, caution fades → risk increases. Lose → dopamine crashes → you chase losses. Repeat. Spiral.

To break this down, let’s look at the main players.

Dopamine: The Hit That Hooks You

Score a win, and dopamine floods your system. It’s the same neurotransmitter that powers gambling, social media, and drug addiction. In trading, this spike makes you crave more action, clouding judgment, fueling revenge trades, and making every tick feel personal.

Testosterone: Confidence, Momentum, and the Risk of Overreach

Testosterone tends to rise after successful scalps. Coates & Herbert (2008) found that winning streaks spiked testosterone on a London trading floor, boosting confidence and risk tolerance. That can help you hold through noise, but when testosterone spikes too high, especially later in the day when cortisol is lower, it pushes you toward bigger, bolder trades — and unnecessary risk.

After wins: you feel unstoppable, sizing up and skipping confirmations. After losses: testosterone dips, confidence falls, and hesitation creeps in. Either extreme can sabotage execution.

Cortisol: The Morning Spike and the Fade of Caution

Cortisol is your body’s main stress hormone. It spikes shortly after waking, making you alert and cautious — a phenomenon known as the Cortisol Awakening Response. Early in the day, that means you may exit winners too early or hesitate to re-enter after a stop-out.

Later in the day, cortisol falls. Stress drops, focus fades, and impulsivity increases. Profitable trades lower cortisol further, which feels good but erodes caution, setting you up to risk more than usual. Kudielka et al. (2004) showed that cortisol affects executive function and emotional reactivity, while Coates & Herbert (2008) observed traders pulling back from risk when cortisol was high.

Adrenaline: The Rush That Burns You Out

Scalping is fast and high stakes. Adrenaline sharpens your focus, helping you respond instantly to price changes. But too much over time is draining. The rush of a winning trade reinforces the cycle, hooking you on action instead of discipline, while chronic exposure leaves you burned out.

Don’t Fall Victim to Your Body

Daily fluctuations in cortisol and testosterone can change how you trade. High morning cortisol makes you defensive. Afternoon testosterone spikes drive overconfidence. Recognizing these rhythms lets you work with your biology instead of being blindsided by it.

Tools to Keep Your Hormones in Check

You can’t eliminate hormonal swings, but you can blunt their impact.

• Cool-down rituals: Pause after profits to reset. Journal feelings after each trade. If confidence or urgency spike, step away.
• Strict rules: Never adjust risk size after a win. Stick to pre-set plans.
• Diet and supplements: Green tea or black tea ease cortisol reactivity better than coffee (Steptoe et al., 2007). Ashwagandha reduces stress (Chandrasekhar et al., 2012). Zinc, magnesium, and vitamin D support testosterone balance.
• Cutting triggers: Alcohol lowers testosterone and disrupts cortisol. Nicotine worsens stress sensitivity. Both degrade performance.

A Daily Routine That Works With Your Hormones

• 07:00: Wake up. Hydrate with water and salt. No caffeine yet.
• 07:30: High-fat, high-protein breakfast to stabilize blood sugar and cortisol.
• 08:00: 5–10 minutes of breathing or stretching to buffer FOMO.
• 09:30–11:00: Trade high-conviction setups. You’re sharp, but watch defensiveness.
• 11:30–14:00: Cortisol falls, testosterone rises. Stick to rules, not impulses.
• 14:00–16:00: Fatigue sets in. Only execute pre-planned trades.
• Evening: Light movement, no alcohol, supplement with magnesium or zinc. Aim for 7+ hours of sleep.

Final Thoughts

Markets don’t just test your strategy. They test your state. Hormones like dopamine, testosterone, cortisol, and adrenaline shape your trading far more than most realize. If you learn to recognize when your biology is helping or hurting you, you don’t just trade better — you think clearer, react faster, and stay consistent.

Forget trying to become a machine. Learn to master the one you’re already in.

References

• Coates JM, Herbert J (2008). Endogenous steroids and financial risk taking on a London trading floor. PNAS 105(16):6167–6172.
• Kudielka BM, Schommer NC, Hellhammer DH, Kirschbaum C (2004). Acute HPA axis responses, heart rate, and mood changes to psychosocial stress in humans at different times of day. Psychoneuroendocrinology 29(8):983–992.
• Steptoe A, Gibson EL, Vuononvirta R, Williams ED, Hamer M, Rycroft JA, Wardle J (2007). The effects of tea on psychophysiological stress responsivity and post-stress recovery: a randomised double-blind trial. Psychopharmacology 190(3):381–389.
• Chandrasekhar K, Kapoor J, Anishetty S (2012). A prospective, randomized double-blind, placebo-controlled study of safety and efficacy of a high-concentration full-spectrum extract of Ashwagandha root in reducing stress and anxiety in adults. Indian J Psychol Med 34(3):255–262.

In this previous article we explored how order book imbalances can forecast price moves in the very short term.

The fundamental concept is:

If liquidity is vanishing from one side of the order book, price movement in that direction becomes statistically more likely.

This is used by market makers (as micro price) and widely documented in scientific research. But is the edge big enough to trade it after accounting for spread, slippage, fees, and commissions?

That’s what I want to find out, so I figured I’d take you with me on that journey and show what goes through my head on a normal Tuesday afternoon. So grab a cup of tea and let’s…

Build A Money Printer

This is going to be the first part of a series where we build a scalping bot based on statistically proven insights that carry alpha. We are aiming to combine such signals and use them as confirmation for each other to create the highest possible quality entry and exit signals to ultimately be profitable after accounting for costs. When that happens, I’m probably going to take the series offline and license the bot, so enjoy it while you can!

The challenge lies not even in finding alpha generating signals — there are plenty of those around—but rather in executing on them in a way that leaves you with net profit. Since academia in general doesn’t concern themselves with the pragmatic reality of our world (they’d not be called ‘researchers’ but ‘millionaires’ otherwise), it’s upon me to take on that task.

That can certainly mean that some research leads us astray, claiming to contain alpha when it actually only is valuable in a self-serving sense and to farm citations, nothing of use to those interested in making money. Nevertheless, there is research that does in fact contain practical value, one reason I know that is having conducted some myself. Which leads us to the most important foundational principle: never trust anyone’s results, create your own. Trusting on some paper to tell you what does and does not have edge would be delusional. Therefore, whenever possible and feasible, I conduct my own research, built from scratch and working with the raw data as input.

Now some might ask: why scalping? Why not swing trading or something else? In order to avoid repetition, let me point you to the answer to that question.

With that out of the way, let’s get started.

The Toolset

We are going to be building our bot on NinjaTrader 8.1 in C#, because it offers good enough speed and a platform that can handle high frequency execution decently well, at least in the dimensions realistically feasible for retail infrastructure. The data feed is going to be CQG: although feeds like Rithmic have higher tick fidelity, which would be desirable, the stability of CQG takes priority of this for the time being.

The Composition

In order for our bot to be functional, we need to account for several parts, which deserve special attention of their own each:

• The Signal. How the bot derives entry and exit signals is essential to its functioning for obvious reasons. This is the core piece that can be most tinkered with and optimised when we do find edge.
• The Entry & Exit Logic. As the name suggests, this is about when to enter and exit, meaning which signal thresholds to act on, whether to have take profit, stop loss, trailing stop, time based stop etc. Notably, risk management rules should not be the thing that makes the bot profitable but that enhances an existing edge. So we only optimise these when we have a statistically meaningful edge. The thresholds for entry/exit are the only parameters that allow discretion on the way to profitability here.
• The Order Management. Here I include which types of orders we use to get in and out of positions as well as their interplay. Think of limit orders on the active or passive side of the book, market orders, OCOs, conditions and timing of cancellations of each of those, and so forth. Managing those in a manner aligning with our intended logic can be challenging due to the fast paced environment.

As you can maybe see already, it’s not just a matter of ‘let’s code it up real quick’ but deserves some degree of management and planning — at least if you want to do it right and not fall flat on your face in a production environment. If you set rules for automation in stone, they better be good.

The Center Piece — The Signal

The heart of our bot is obviously the algorithm that drives the signal generation. For that we will revisit the idea explored in the article mentioned at the start: order book imbalances and their predictive value.

For this version, we will implement a methodology orienting itself on the one used by Cont. et al in their 2010 paper. In order to not overload with too many technical details, here is the rough outline:

• Over 10 seconds, we aggregate changes to the order book.
• We are considering +-5 levels of depth.
• Events are counted in the following manner:
  Ask side liquidity added & bid side liquidity removed contributes negatively to our value, ask side liquidity removed & bid side liquidity added contributes positively to our value. So to put it visually: liquidity added above price or removed below = bearish, vice versa bullish.
• Depending on where the event takes place in terms of distance from the best bid/ask, we weigh it so that further out levels have a smaller impact.
• Price moves count as a complete removal of all liquidity from the previous state and an adding of the liquidity around the new price.
• The resulting numerical aggregate is decayed over time with a half life of about 2 seconds, so more recent changes have a greater numerical impact.
• We normalise the resulting value by total liquidity observed, clamping it to [-1, +1], so it’s nice and interpretable.

This conceptual approach can be labeled GOBI — generalised order book imbalance since we generalise the top-of-book concept to a variable depth. Given we are dealing with liquidity levels, I’m sure there is a good pun hidden in that, but since I’m not interested enough to find it right now, that’s left as an exercise to the reader.

There is a great deal of parameters that can be optimised here and that deserve consideration. E.g. why 10 seconds aggregation, why 2 seconds half life time and so on…
We just go with those initial values for now and can come back to adjust them later if we see fit. Hyperparameter optimisation — and the associated risk of overfitting in the process — deserve their own consideration and would exceed the scope of this article for now.

Now we have a value that is >0 if we have a bullish signal, <0 for a bearish one, great!

The Entry & Exit Logic

How do we put our newly won wisdom into action now?

Picking a Threshold

The challenge is that a value greater or smaller than 0 means we should go long or short. Cont et al found (and it’s natural intuition too) that too small of an imbalance does not qualify for a proper signal and has little predictive value. So we want a threshold to look out for. When that is crossed, we enter. Else, we wait.

This is where academia leaves us to our own judgement already, since values like that are not widely provided. There are sources suggesting 1/3 as a value but ultimately, this will have to be left to our own discretion.

Notably, higher imbalances should correspond to a stronger effect which can either manifest in higher reliability of the signal or a greater expected pricing effect. Either way, it should improve the expected value of the trade taken as a result of following the signal. So this can be our first lever to tune: higher threshold = fewer, but (hopefully) better signals.

Configuring Risk Management

Usually I like to just let the signals run on their own first, no TP, no SL, no trailing stop. That way I can really see what the distribution of returns looks like in a vanilla setting, undistorted by arbitrary thresholds.

From there, assuming we have an edge, we can make adjustments to see which kind of risk management rules improve our edge, e.g. which losers don’t comeback, how far winners go into drawdown before materialising profit.

So this will be our first version: no safety net and no limit to the upside.

The Order Management

The novice does not even concern themselves with this. Market order in, market order out, right? What’s the big deal? That’s how you burn money, not print it. If you cross the spread on every trade twice, even before fees and commissions, you are making donations to the market maker, nothing more, nothing less. We want to consider alternatives to this:

• The first idea we should have to limit slippage at least is to use a limit order on the active side of the book. That means when we have a bid/ask quote of 100/101, instead of placing a market order and potentially getting filled at 102 or 105 or 110 on our buy, we submit a limit order at 101, signaling we don’t want to pay more than that to enter. This might mean we do not get filled and price runs up without us, but that’s the price we need to be willing to pay to avoid gruesome slippage consistently.
• To even further reduce costs, we want to avoid crossing the bid/ask spread as much as possible. Luckily, there is a way to do this. Instead of placing our order at the active side of the book, where our market order would be filled without slippage, we can place it either inside the spread or even on the passive side of the book. So instead of submitting our limit order at 101, we submit it at 100, hoping price will move towards our order. This again comes at the risk of not getting the fill and price moving up without having a position open but saves you the burden of having to cross the spread on entry. We can even do this on exits as well, although we do want to be much more selective with that here to not compromise our risk management here.

Let’s look at a simple example. We assume we are trading NQ with a spread of 2 ticks, fees of 1 tick worth and a slippage of 0.5 ticks on average.

• Market in, market out yields: 2 ticks from the spread, 0.5 ticks from slippage on entry and exit = 5 ticks, plus another 1 tick from fees works out to 6 ticks. So you need to make at least 7 ticks on average to even be profitable.
• Active limit in, market out. Same idea, but you save 0.5 ticks from slippage on entry on average, meaning you have costs of 5.5 ticks and only need 6 on average to be profitable.
• Passive limit in, market out. You save 2 more ticks by not having to cross the entry spread, if you don’t get the trade as a result, you don’t lose anything. Now you have costs of 3.5 ticks and need only 4 ticks on average to be profitable on the trades that you do catch.

As you can see, those tricks can help significantly reduce the actual edge your signals need to provide in order to derive value from them. Going from 7 ticks down to 4 is roughly a 43% reduction.

It should be said that in my view, those considerations only really make sense if the costs in relation to the trade result are meaningful. Logically, you don’t need to worry about this if you build a swing trading bot, making one trade a day since the magnitude of your PnL overshadows the cost vastly.

The Bot In Action

Now I gave you a great outline of what we’re building, how and why. But I lied to you. I said I’d be building the bot alongside writing the articles, but that’s not entirely true. Actually, as I’m writing this I already have built everything described here. It’s live right now. And I’m sure you are eager to know about the results… but you will have to exercise patience, for that will be revealed in the next article of the series!

To not leave you high and dry now (and to tease your curiosity even further), here’s a little treat for you nonetheless:

Thank you for your attention and see you next time!

Disclaimer: The information provided is for general informational and educational purposes only. It should not be considered financial, investment, or trading advice. Markets are unpredictable, and past performance is not indicative of future results. Always conduct your own research and consult with a qualified professional before making financial decisions.

Every week, the Commitments of Traders report lands with little fanfare. It is a dry snapshot of futures positioning. Yet across currencies and index futures, these snapshots contain a single, consistent predictive message. When speculative net exposure is elevated, future returns most often continue in the same direction over short horizons. There is no second, separate signal hiding on the hedger line. There is one net positioning signal. This article explains precisely what the studies find, how far those results extend, and how traders can use the information without importing myths.

What the CoT Report Is, and What It Is Not

The CoT report is a weekly breakdown of open interest by trader category. Commercial hedgers and large speculators are the main groups most traders watch.

If speculators are net long, hedgers must be net short by construction, aside from small traders and rounding. That means the hedger series and the speculator series carry the same information, just with opposite signs. When discussing predictability, you do not have two independent signals. You have one signal, which you can describe from either side.

For practical use, two operational choices make the CoT far more informative:

• First, work with net position levels rather than raw contracts. Normalize them against history using percentiles or z-scores so “extreme” means something objective.
• Second, align expectations with the correct horizon. The research points to short windows measured in weeks for most effects, with a longer tail in EUR/USD and a separate medium-term channel tied to hedging pressure.

Signal One: Continuation

The foreign exchange evidence is clear on three points.

• Levels matter. When speculative net exposure is long, subsequent FX returns tend to be positive over short horizons. When speculative net exposure is short, subsequent returns tend to be negative. The strongest forecasting power shows up over one to four weeks.
• Weekly changes don’t matter much. They describe what just happened more than what will happen. Their out-of-sample predictive content beyond about one week is weak. For traders, that means levels carry the information and flows should be treated with caution as a forecasting device.
• Turning points carry the information. Local extremes in speculative positioning tend to align with continuation of the prevailing move. Local troughs sometimes precede rebounds, but these reversal effects are weaker and less reliable. The turning-point window where this information shows up spans roughly one to twelve weeks, again strongest at the short end.

Taken together, the FX literature supports a disciplined continuation interpretation of CoT levels, while advising skepticism toward flow-based forecasting and toward broad claims about reversal at speculative extremes.

EUR/USD: A Longer Tail

EUR/USD is on of the the most studied FX pairs in the CoT literature, and it has shown a stronger and longer-lived continuation pattern than most currencies. There is an important modern nuance though.

Post-2008, some of these relationships weaken once you correct for multiple testing and account for the changing composition of trader categories. The signals have not vanished, but they are smaller and more dependent on which speculative group you examine.

The practical implication is to use the same continuation logic, but with tighter expectations for effect size and with a firmer emphasis on risk control and confirmation. For an educational takeaway, EUR/USD is the one currency where traders can reasonably test both short-horizon and medium-horizon implementations of the same core principle.

Elevated speculative net longs have historically supported continued appreciation beyond the next several weeks, while still fading in strength as the horizon extends.

Signal Two: Hedging Pressure

Beyond the level and turning-point effects, the hedging pressure literature examines shocks to commercial hedging demand. This is a different lens that still uses CoT-style information but constructs a series for changes in hedger demand rather than speculator levels.

Positive hedging pressure shocks in currencies, which you can think of as sudden and sizable increases in commercial demand to hedge against USD strength, are followed by broad USD depreciation. The effect is not immediate. It builds and peaks at roughly five months, then fades thereafter. This is not a reversal story in the short run, but a medium-horizon predictability channel that operates alongside the short-horizon continuation effects documented for speculative levels.

For a trader designing research, that two-speed structure suggests separating short-term continuation signals from medium-term hedging-pressure signals rather than blending them. The construction and estimation windows differ, and mixing them tends to blur the information.

Putting It To Work

A fact-only reading of the studies leads to a simple, implementable blueprint.

• Start with levels, not flows. Compute speculative net positioning and normalize it against an appropriate history so you can identify truly elevated or depressed readings. For FX outside EUR/USD, size the strategy for one to four week horizons and assume the edge decays quickly. For EUR/USD specifically, test an extended holding window out toward twelve to twenty weeks, but expect weaker effect sizes and more variability after 2008.
• Use turning points as context, not as automatic reversal flags. A local peak in speculative longs supports continuation, not contrarian bets. A local trough can be informative for rebound potential, but it is the weaker of the two and benefits from confirmation through other tools.
• Treat hedging pressure as a separate module. If you build a hedging-pressure shock series, evaluate it on medium horizons measured in months. Its best use is as a slow signal that can frame exposure for macro portfolios rather than as a trade trigger for weekly systems.

Scope, Limits, and Realistic Expectations

The CoT findings are consistent across several markets and time periods, but they have limits that matter to trading results.

The data is weekly and lagged. Tuesday positions are reported on Friday. That cadence favors slower, rules-based updates rather than intraday timing. The signals are statistical edges, not certainties. They shift with structural changes in market composition, regulation, and the behavior of dominant participants. Post-2008 EUR/USD results illustrate this point. Many relationships remain, but they are smaller and more selective.

Crucially, the strongest effects are horizon-specific. Short-term continuation in FX and futures does not imply medium-term predictability. The hedging-pressure channel does, but it is a different mechanism and must be tested as such. Mixing horizons muddles inference and often leads to overfitting.

The bottom line is straightforward:

1. CoT net positioning contains a reliable, short-horizon continuation signal in FX and futures, with EUR/USD exhibiting a longer tail.
2. A distinct medium-horizon effect arises from hedging-pressure shocks in currencies.

If you build around those facts and use appropriate horizons, the CoT can be a practical, evidence-based input to a robust trading process.

Disclaimer: The information provided is for general informational and educational purposes only. It should not be considered financial, investment, or trading advice. Markets are unpredictable, and past performance is not indicative of future results. Always conduct your own research and consult with a qualified professional before making financial decisions.

References

• Quantified Strategies, Commitments of Traders Trading Strategies: COT Report And Backtest, 2024.
• Wang, Futures Trading Activity and Predictable FX Movements, 2004.
• Klitgaard & Weir, Exchange Rate Changes and Net Speculator Positions, 2004.
• Tornell & Yuan, Speculation and Hedging in Currency Futures, 2009.
• Hossfeld, Do Speculative Traders Anticipate USD/EUR?, 2016.
• Röthig, Do Speculative Traders Anticipate USD/EUR?, 2023.
• Bräuer, Currency Hedging Pressure and Exchange Rate Dynamics, 2019.
• Chen, Chong, Duan, Large-Trader CoT and S&P 500 Predictability, 2013.

Most traders react to what has already printed. They wait for a candle to close or an indicator to confirm. The order book lets you work one step earlier. It shows where liquidity sits and how easily price can move before the move occurs. When one side of the book carries more resting volume than the other, that asymmetry changes the mechanics of price formation. The result is a short predictive window that human traders can use, typically from a few seconds to about a minute.

This article explains what an order book imbalance is, why it leads price, how to read it in real time, and how to incorporate it into a disciplined intraday process. It also anchors the claims in peer-reviewed research that examines predictability on human-tradable horizons.

What order book imbalance really means

The order book is a live ledger of limit orders. Bids rest below the mid price, offers rest above it. In a balanced book, depth on the bid and ask sides is roughly symmetric across the top few price levels. Imbalance appears when one side holds clearly more resting size than the other. That may occur at the top of book only, or across several levels.

Imbalance changes the ease of movement. A thin offer stack means a given burst of buy market orders can lift price through multiple levels with less traded volume. A thick bid stack means the same burst of selling will travel less far because it encounters more liquidity. Liquidity is the road surface under the wheels of price. Where the surface is smooth and thin, price rolls quickly. Where it is rough and deep, price slows.

The key is that imbalance is structural information. It is not a lag of past trades. It is a snapshot of potential future trades, which is why it can lead price for a short interval.

Evidence that imbalance leads price in human timeframes

Several lines of research support this idea.

• Cont, Kukanov, and Stoikov show that order flow imbalance has a near-linear relationship with short-horizon price changes. When an imbalance is aggregated over brief intervals, the link to subsequent returns remains strong, especially within tens of seconds. So using something like an exponential moving average on the the imbalance yields signals for scalping.
• Gould and Bonart study queue imbalance at the best bid and best offer. They show that queue imbalance predicts the next move in the mid price, with especially strong performance in large-tick instruments where discrete queue dynamics dominate. Although their prediction target is one-tick ahead, the mechanism they document is the same one intraday traders use over multi-second windows. The side with the thinner queue yields first.
• Deep learning studies reach the same conclusion from a different direction. Models trained on limit order book tensors consistently rank imbalance and queue features among the most informative inputs for short-horizon direction classification. Results are robust out of sample for horizons measured in seconds and tens of seconds.
• Agent-based work shows the mechanism is not an artifact of a particular dataset. In simulations where agents condition on imbalance, strategies that trade in the direction of the dominant side tend to generate positive expectancy when evaluated over seconds to about a minute. Profitability persists even with small execution delays, which brings the result into the realm of discretionary trading rather than pure high frequency.

Finally, research on price pressures from intermediaries helps explain why the effect persists beyond a single tick. Inventory imbalances and quote imbalances from liquidity providers create short-run pressure that does not vanish instantly. During session opens and macro releases, when participation concentrates, the short-term impact of imbalance on both direction and volatility becomes even more visible.

Taken together, these threads support a practical claim:

Imbalance is a forward signal with a short but usable half-life. It can tilt the odds for 5 to 30 seconds, and in some conditions for close to a minute.

How to detect imbalance in real time

To read the order book, you need a clean depth-of-market display, a fast enough feed, and a consistent way to quantify what you see.

A simple workflow for an indicator:

1. Focus on the top three to five price levels on both sides.
2. Sum resting size on the bid and on the ask. Weigh by distance in accordance with the levels explanatory value: [1.00, 0.50, 0.25, 0.125, 0.0625].
3. Compute a ratio. For example, bid depth divided by total depth.
4. Treat values above roughly 60 percent as directional.
5. Watch how that ratio evolves, e.g. via a moving average, rather than fixating on single snapshots.

Pair the ratio with a minimum absolute size threshold so that e.g. a 70 percent skew built on almost no contracts does not fire.

Why imbalance predicts price

Imbalance changes the cost of moving price. When one side of the book is thin, each incoming market order consumes more levels and produces a larger price change. This mechanical channel is what Cont, Kukanov, and Stoikov (2010) measured when they showed a nearly linear relationship between order flow imbalance and short-horizon price changes.

A useful way to think about this is friction. Price moves more easily toward the thin side of the book, because the same volume of market orders travels further. Just as a car rolls faster downhill than uphill, price accelerates where resistance is lower.

Using imbalance in a strategy

Imbalance is not a full system, but a filter and a timing aid. The highest value use cases are simple:

• Breakout confirmation. If price is coiling under a level and the liquidity above thins while it builds below, your breakout setup has real backing. The moment you see price start to uptick without a fresh surge in aggressive buying, it is often the tell that liquidity above is no longer refreshing. That is the ignition.
• Failure detection. If price tests a level repeatedly but the book tilts the other way and keeps replenishing, expect failure. E.g. price tries to move up but runs into fresh liquidity again and again. The same applies if cumulative delta turns in favor of buyers but the liquidity above grows heavier. That is absorption, not escape.
• Entry timing. The best entries in fast markets often occur when the skew first appears and refresh behavior changes. Waiting for the candle close may be too slow. Conditioning on the book lets you step in earlier with defined risk.

Three practical rules help:

1. Use imbalance with at least one other flow tool such as cumulative delta or a footprint view of absorption.
2. Prefer signals that align with the volatility regime and session timing.
3. Cut the signal at a fixed horizon. If a skew does not resolve within your window, treat it as expired.

Limitations and safeguards

Imbalance is powerful because it is visible. That visibility invites games. Spoofing, hidden liquidity, and iceberg orders can momentarily distort the picture. You will not eliminate this risk, but you can reduce it.

• Require persistence. A skew that lasts only a second is noise. You can achieve this by using a moving average over the imbalance.
• Watch for participants not coming back. If liquidity shrinks and stays thin after a test, the signal is stronger.
• Track correlated markets. In index futures, a skew that aligns across related contracts is harder to fake, think ES and NQ.
• Respect news. During unscheduled events, the book can disappear faster than you can react. These times are for bots, not humans.

The right mindset is probabilistic. You are not predicting with certainty. You are stacking the odds by trading where the road is easier to travel.

Putting it all together

A robust intraday process can be built around three pillars. First, define the context. Volatility regime, session, and upcoming catalysts frame the quality of any signal. Second, locate the structure. Identify levels where a change in refresh behavior would matter. Third, let imbalance and flow confirm or deny your idea. When the book thins on the path of least resistance and the tape agrees, take the trade with a firm horizon and a predefined stop.

Do this consistently and you will find yourself buying a touch earlier in real breakouts, standing aside in likely failures, and exiting sooner when the skew expires. That is the edge order book imbalance offers.

References

• Aït-Sahalia, Y.; et al. Predicting Returns with Order Book and Trade Imbalances. 2022.
• Cont, R.; Kukanov, A.; Stoikov, S. The Price Impact of Order Book Events. 2010.
• Gould, M. D.; Bonart, J. Queue Imbalance as a One-Tick-Ahead Price Predictor in a Limit Order Book. 2015.
• Hendershott, T.; Menkveld, A. J. Price Pressures. 2014.
• Zhang, Z.; Zohren, S.; Roberts, S. DeepLOB: Deep Convolutional Neural Networks for Limit Order Books. 2018.
• Yagi, R. Impact of Order Book Imbalance Strategies in Agent-Based Markets. 2023.

Introduction

Market watchers often treat the VIX as a barometer of fear. When volatility expectations rise, equity markets usually fall, and the relationship is assumed to be broad, slow-moving, and long-term. Yet a growing body of high-frequency research suggests something subtler is happening beneath the surface.

The VIX does not just mirror the S&P. It sometimes leads it, and the leadership emerges in windows measured not in days but in seconds. Understanding these fleeting intervals is critical for traders who operate on intraday horizons, because they reveal where price discovery really begins in moments of stress. This opens a window for statistical arbitrage opportunities.

Why Lead–Lag Matters

A lead–lag effect means that one instrument systematically moves before another. In efficient markets, prices should reflect information instantly across related assets. But frictions such as liquidity, hedging flows, and information channels create short gaps where one market reacts before the other. For index traders, the question is whether volatility products like VIX futures react first to shocks and whether those moves filter into equity index futures such as the E-mini S&P (ES).

If volatility leads equities, even by just 5 seconds, traders can extract predictive signals, while portfolio managers gain insight into where hedging demand originates during turbulence.

Evidence from High-Frequency Research

A key study by Bangsgaard and Kokholm (2021) examined tick-by-tick data for VIX futures and ES futures between 2013 and 2020. Using cross-correlation functions and lead–lag ratio metrics, they found:

• Time horizon: The lead–lag relationship is extremely short. It is strongest within the first 0–5 seconds after a move in VIX futures and decays to noise beyond 15–20 seconds
• Asymmetry: The correlation curve skews in one direction. VIX futures more often move first, while ES follows.
• Volatility dependence: During turbulent periods, VIX’s leadership strengthens. The lead–lag ratio approaches 0.8 in high-volatility regimes, compared to weaker effects in calm markets.
• Granularity: Even at millisecond resolution, the skewness remains, confirming that VIX price discovery consistently precedes equity futures.

Other researchers using five-minute data windows found similar patterns. VIX not only predicts the direction of the next ES move but does so with significant statistical power, especially when volatility is rising rapidly. This predictive content appears asymmetric: rising VIX is more informative about future downside in equities than falling VIX is about upside.

Why Does VIX Lead?

Several mechanisms explain why volatility can move before equities:

1. Hedging flows: Institutional investors often hedge equity exposure with S&P options. When stress builds, demand for protection pushes up implied volatility quickly, feeding into VIX futures.
2. Dealer positioning: Market makers adjusting their option books transmit pressure into VIX futures before index futures react.
3. Liquidity tiers: The VIX futures market is narrower and more sensitive to order flow, so marginal demand shows up faster there than in the broader ES book.
4. Information channel: In times of uncertainty, traders look to volatility markets as a first signal of risk, amplifying their role in price discovery.

Practical Takeaways for Traders

For intraday traders, the VIX–ES lead–lag relationship is not a crystal ball but a probabilistic edge:

• Monitor cross-asset order flow: When VIX futures tick sharply higher, especially in volatile sessions, expect ES to follow within seconds via a downward move.
• Condition on volatility regimes: The lead effect is strongest during high-volatility periods. In calm conditions, the signal is weaker.
• Beware noise: The window is small. Beyond 20 seconds, correlation fades, so lagging signals are unreliable.
• Extend cautiously to Nasdaq: Studies focus on ES because VIX is built from S&P options. Nasdaq futures (NQ) react almost simultaneously, but through correlation with ES rather than direct linkage to VIX.

Beyond the One-Day VIX

The CBOE has introduced ultra-short horizon indices such as VIX1D, based on options expiring the same day. This metric further sharpens sensitivity to intraday sentiment shifts, though at the cost of greater noise. As researchers explore these measures, the possibility of constructing volatility dashboards that track implied volatility across zero-day, one-day, and longer maturities offers a richer way to detect turning points in real time. Let me know if you are interested in tools like that!

Conclusion

The VIX is more than a fear gauge. At high frequency, it plays the role of scout, stepping into new ground before the equity market follows. Studies show that the leadership window is measured in seconds, most often between 0 and 5, rarely extending beyond 20. The effect intensifies when markets are unsettled, underscoring how risk perceptions crystallize in volatility products before they appear in stock prices.

For traders, the message is simple: watch the VIX tape. In the heat of intraday trading, when volatility jolts higher, the S&P often trails close behind. The gap may be small, but in markets where opportunity is counted in ticks and seconds, small can be decisive.

References

• Bangsgaard, J. & Kokholm, T. (2021). The Lead–Lag Relation Between VIX Futures and SPX Futures. Aarhus University Working Paper.
• Andersen, T., Bondarenko, O., & Gonzalez-Pedraza, F. (2020). Intraday Predictability of the Stock Market: Evidence from VIX and S&P 500 Futures.
• CBOE Global Markets. VIX and VIX1D Methodology Papers.

How to Spot and Trade Breakouts With Order Flow

Every trader knows the frustration of false breakouts. Price pokes above resistance or dips below support, only to whip back and trap anyone who chased the move. Chart patterns alone rarely tell you if a breakout will stick. That is where order flow comes in. By looking beneath the surface of price and studying how trades are executed, where liquidity sits, and how participants behave, traders can tell the difference between noise and genuine momentum.

Order flow analysis does not require complex mathematics. It comes down to watching who is aggressive, who is passive, and when that balance changes. Five signals in particular give early clues that a breakout is real rather than a trap.

Delta Divergence Resolving

The first signal comes from the relationship between price and cumulative delta, which tracks whether aggressive buyers or sellers dominate.

Imagine a session where cumulative delta climbs steadily. Buyers are repeatedly lifting offers, yet price remains stuck under resistance. This is absorption: passive sellers are taking everything that comes their way.

The turning point arrives when delta stalls, but price begins to tick higher anyway. That inflection means the offers are no longer refreshing. Buyers do not need to push harder; the mere absence of supply lets price slip through. Traders sometimes describe this as price “jumping through thin air.” It is the moment absorption ends and breakout momentum begins.

The practical takeaway is to watch for divergence between delta and price.

If price advances despite delta flattening, it signals passive sellers stepping back. That is often the earliest tell that the ceiling is about to give way.

Order Book Thinning at the Ceiling or Floor

Another way to spot breakouts is by monitoring the depth of market, or DOM. When price approaches a well-defined boundary, you often see large passive orders sitting there, acting like a wall. These orders can absorb a lot of flow and keep the market contained.

But if those orders stop refreshing, the wall suddenly looks thin. What had been a fortress becomes a doorway. Traders watching the order book will notice that the heavy blocks of offers at resistance or bids at support shrink instead of replenish.

That thinning is a precursor to a break. Aggressive orders that once bounced off a thick book now find little to stop them. Price does not need dramatically more demand to move higher — it simply needs the supply to vanish.

Seeing the book dry up at key levels is often enough to anticipate that the next wave of aggression will slip through.

Absorption Failure

Breakouts often show up clearly on footprint or volume profile charts. When price returns to the same level multiple times, you will usually notice heavy bursts of trading activity. Aggressive buyers may keep hitting the offers, or aggressive sellers may keep pressing into the bids. If each of those pushes is matched by strong opposing orders and the liquidity keep replenishing, the level continues to hold.

This matching of pressure on both sides is what traders call absorption.

Over time, however, the counterflow can weaken. Imagine three pushes into resistance:

• The first sees heavy buy volume matched by heavy sells.
• The second sees less of a sell response.
• By the third, the imbalance is mostly one-sided. Sellers are backing off.

This shrinking absorption is a reliable sign that the passive side is losing conviction. Once their willingness to defend the level collapses, the breakout is free to unfold. Observing the relative size of counterflow across multiple tests helps traders distinguish between ranges that are robust and those that are hollowing out from within.

Delta Flip at the Micro Level

Not all signals show up in big aggregates. Sometimes the most telling clues are micro patterns. Right before a bullish breakout, you may see a small pullback inside the range where delta flips negative — a cluster of selling pressure — yet price does not fall.

This asymmetry is important. It shows that even when sellers attack, strong hands are willing to absorb, and price refuses to move lower. Weak hands are flushed out, but the market does not give ground. When that happens near the ceiling of a range, it often precedes an upward ignition.

The same holds on the downside. If buyers push but price does not rise, it suggests the level is about to give way lower.

Watching for these micro flips where delta and price disconnect is a way to catch breakouts in their earliest moments.

Liquidity Provider Hedging Behavior

The final signal involves a broader perspective. Liquidity providers such as dealers and market makers manage inventory across correlated markets. In Nasdaq (NQ) and S&P (ES) futures especially, their hedging behavior can amplify breakouts.

If dealers have sold heavily into buying pressure, they accumulate short inventory. In calm conditions they may manage it without fuss. But when volatility is high, they cannot hold the risk. They hedge by buying correlated instruments or adjusting options positions. This hedging adds fuel to the move in the same direction as the breakout.

You can often see this when volatility spikes and correlated markets move together. A push in Nasdaq futures coincides with a pop in S&P, a shift in options implied volatility, or sudden strength in sector ETFs. That alignment is not random — it reflects hedging flows that turn local breakouts into broader momentum.

Recognizing these cross-market cues gives traders an extra layer of confidence that the move is real.

Bringing the Signals Together

Each of these signals tells the same story in a different language. Delta divergence resolving shows price outpacing flow. Order book thinning shows the passive side retreating. Absorption failure shows defense weakening across repeated tests. Delta flips highlight micro-level asymmetry. Hedging behavior shows how inventory management spills over into correlated markets.

Used alone, each signal can be noisy. Used together, filtering out each other’s noise, they form a more reliable map of breakout conditions. The skill lies in learning to read when aggression, passivity, and liquidity shift in ways that let price finally escape the cage it has been trapped in.

Why This Matters

For traders, the difference between chasing false moves and catching genuine breakouts can make or break performance. Price alone is a blunt instrument. Order flow adds texture, revealing the battle underneath.

Academic studies support these observations. Hasbrouck (1991) demonstrated that order flow shocks produce lasting effects, not just transitory blips. Easley, Kiefer, O’Hara, and Paperman (1996) argued that persistent imbalances reflect informed trading. Hendershott and Menkveld (2014) showed that inventory management by intermediaries creates temporary pressures but ultimately reinforces directional moves. Together, this research explains why the patterns described above have predictive power.

Key Lessons

Breakouts are not random. They emerge when aggressive traders keep pressing, passive participants lose their willingness to absorb, and liquidity providers reach their limits. The signals to watch are clear:

• Delta rising then stalling while price lifts.
• Order book walls that thin instead of refresh.
• Repeated tests where absorption weakens.
• Micro flips where delta turns but price refuses to follow.
• Cross-market moves that reveal hedging pressure.

For a novice, it is enough to start watching how delta and price interact. For a more advanced trader, combining all five signals builds a framework that transforms the breakout from a guess into a structured trade.

References

• Easley, D., Kiefer, N. M., O’Hara, M., & Paperman, J. B. (1996). Liquidity, Information, and Informed Trading. Journal of Finance, 51(2), 481–508.
• Hasbrouck, J. (1991). Measuring the Information Content of Stock Trades. Journal of Finance, 46(1), 179–207.
• Hendershott, T., & Menkveld, A. J. (2014). Price pressures. Journal of Finance, 69(6), 2713–2756.

Takeaway: Breakouts are the moments when markets shift from balance to imbalance. Order flow gives you the tools to spot those shifts before they become obvious on a chart.

At first glance, this looks like little more than a chart pattern. Yet a surprising amount of research shows why the opening range is important, why the breakout sometimes works, and why it often fails. More importantly, modern refinements can turn this old idea into a conditional but effective tool for today’s markets.

What Exactly Is the Opening Range?

When U.S. markets open at 9:30 a.m. Eastern Time, they immediately absorb the overnight buildup of orders, positioning, and news. In the first thirty to sixty minutes, prices often whip around as large players adjust exposure and liquidity providers set their books. The highest and lowest prices recorded in this period are known as the opening range.

The ORB strategy then uses this range as a decision framework. A push above the high is interpreted as bullish, while a break below the low suggests bearish momentum. Traders often place stops at the opposite side of the range, and positions are usually closed by the end of the day or managed with trailing exits.

The attraction is that the opening range compresses the day’s initial battle into a visible box. Whatever happens after can be judged against that early standard.

What the Evidence Says

Economists have long recognized that the opening session is special. Wood, McInish, and Ord (1985) showed that the first hour contains a disproportionate amount of price discovery on the New York Stock Exchange. Harris (1986) found that the early burst of activity shapes intraday return patterns in predictable ways.

Later work examined ORB as a strategy in its own right. Holmberg, Wang, and Wu (2013) tested intraday breakouts across futures contracts and found that unfiltered ORB rules often fail once realistic costs are considered. Crucially, they discovered that profitability clusters in periods of high volatility. Donninger (2014) confirmed similar patterns for S&P 500 futures: without filters, results hovered around breakeven, but once volatility regimes were considered, performance improved.

The message is straightforward. ORB is not a timeless rule that works in all markets. It is a conditional tool that performs well in turbulence and poorly in calm conditions.

Volatility as a Filter

The most effective filter for ORB is volatility. Traders commonly use the CBOE Volatility Index, or VIX, as a simple gauge.

• When VIX sits below 15, breakouts tend to fail and the strategy loses money.
• Between 15 and 20, results are inconsistent. Some days produce trends, others collapse back into the range.
• Above 20, breakouts are more likely to follow through, and the strategy can become meaningfully profitable.

This regime dependence matches trader intuition. Calm markets grind sideways and punish breakout attempts. Stormy conditions bring directional order flow, and ranges are left behind decisively. Academic research simply quantifies what seasoned traders already feel: the ORB is a storm-weather strategy.

Order Flow and Cumulative Delta

Another angle comes from market microstructure research. Every trade has both a buyer and a seller, but one of them initiates. If a trade executes at the ask, the buyer was aggressive. If it executes at the bid, the seller took the initiative. Summing these trades over time produces cumulative delta, a measure of whether aggressive buyers or sellers dominate.

Hasbrouck (1991) demonstrated that shocks in order flow have lasting impact on prices, not just transitory effects. Easley, Kiefer, O’Hara, and Paperman (1996) showed that persistent imbalances often reflect informed trading, and therefore have predictive power.

Applied to ORB, this means that the balance of aggressive activity during the opening range can tilt the odds:

• If cumulative delta is strongly positive but price refuses to rise, it signals that sellers are absorbing. When they finally step away, price often breaks upward quickly.
• If delta is deeply negative but price holds steady, it suggests hidden buying support that may give way to a downside break.

The key is not just the delta itself but how price behaves in relation to it. When price starts to move despite delta stalling or even flipping, it often marks the transition from absorption to breakout.

Liquidity and Absorption

Liquidity providers such as market makers sit on the other side of aggressive traders. They may sell repeatedly to eager buyers or buy from urgent sellers in order to capture spreads and rebates. In doing so they build inventory.

As long as they believe the flow is uninformed or as long as their risk limits allow, they are willing to absorb. But when that tolerance runs out, two things happen: they either pull their orders, leaving the book thin, or they hedge their inventory elsewhere, which itself reinforces the move.

Hendershott and Menkveld (2014) describe this as price pressure. Absorption can suppress moves for a time, but the eventual rebalancing often amplifies the very trend that was held back. For ORB traders, this means that a range with strong one-sided delta and stubbornly flat price is not boring at all. It is a coiled spring.

A Walkthrough Example

Consider a day in Nasdaq futures. From 9:30 to 10:30 a.m. the contract trades between 15,100 and 15,200. During that time cumulative delta climbs steadily as buyers lift the offer, yet price stays capped below 15,200.

This is absorption. Sellers are stepping in, confident enough to keep leaning on the offer. Then, around 11:00, delta stops climbing. If sellers truly had the upper hand, price would drift down. Instead it ticks upward and punches through 15,200. That move, occurring without fresh delta support, is the signal that the wall of offers has vanished. The breakout runs, helped along by liquidity providers now forced to hedge their accumulated short positions.

Beyond the Basic Setup

The ORB is most powerful when combined with other perspectives. Volatility filtering with VIX is the most straightforward. Order flow metrics such as cumulative delta add insight into whether a breakout is likely to stick. Options positioning, which can pin prices through dealer hedging, offers yet another dimension.

Exploring these combinations turns ORB from a blunt rule into a flexible framework. Traders can study not just successful breakouts but also failed ones, where price and delta diverged in the opposite way. These failures often contain as much information as the successes.

Key Lessons

The Opening Range Breakout has endured for a reason. It directs attention to the most informative part of the day, where order flow and liquidity collide with overnight positioning. Research confirms that:

• The opening session carries the most informational trading.
• ORB on its own is close to breakeven after costs.
• Profitability appears in high-volatility regimes, especially with VIX above 20.
• Order flow measures and liquidity dynamics explain why some breakouts fail and others explode.

For newcomers, ORB provides a simple way to frame the chaos of the open. For experienced traders, it becomes a conditional but robust tool when filtered by volatility and confirmed with order flow.

Where to Explore Further

If the idea of ORB sparks interest, the natural next steps are to explore how regime filters, order flow measures, and options positioning interact with it. Academic studies suggest that conditional strategies, not universal rules, offer the most durable edges. ORB fits neatly into that picture.

A series of experiments could include: testing ORB only on days with high volatility, layering in cumulative delta as a directional filter, and adding awareness of options dealer positioning. Each layer reduces false signals and builds a more complete understanding of why the strategy sometimes works.

References

• Easley, D., Kiefer, N. M., O’Hara, M., & Paperman, J. B. (1996). Liquidity, Information, and Informed Trading. Journal of Finance, 51(2), 481–508.
• Harris, L. (1986). A transaction data study of weekly and intradaily patterns in stock returns. Journal of Financial Economics, 16(1), 99–117.
• Hasbrouck, J. (1991). Measuring the Information Content of Stock Trades. Journal of Finance, 46(1), 179–207.
• Hendershott, T., & Menkveld, A. J. (2014). Price pressures. Journal of Finance, 69(6), 2713–2756.
• Holmberg, U., Wang, J., & Wu, J. (2013). Assessing the profitability of intraday opening range breakout strategies. Journal of Futures Markets, 33(6), 548–573.
• Wood, R. A., McInish, T. H., & Ord, J. K. (1985). An investigation of transactions data for NYSE stocks. Journal of Finance, 40(3), 723–739.
• Donninger, C. (2014). S&P 500 Intraday Trading: Opening Range Breakout. SSRN Working Paper.