<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Ethan Wang on Medium]]></title>
        <description><![CDATA[Stories by Ethan Wang on Medium]]></description>
        <link>https://medium.com/@oneforest?source=rss-6192963b940------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*BSwsgAoouFq-34YtC0T_Tg.png</url>
            <title>Stories by Ethan Wang on Medium</title>
            <link>https://medium.com/@oneforest?source=rss-6192963b940------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 00:46:46 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@oneforest/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[I Automated 4,000 Trades on Kalshi — Here’s the Results]]></title>
            <link>https://oneforest.medium.com/i-automated-4-000-trades-on-kalshi-heres-the-results-2f686c214e25?source=rss-6192963b940------2</link>
            <guid isPermaLink="false">https://medium.com/p/2f686c214e25</guid>
            <category><![CDATA[math]]></category>
            <category><![CDATA[algorithmic-trading]]></category>
            <category><![CDATA[side-hustle]]></category>
            <category><![CDATA[prediction-markets]]></category>
            <category><![CDATA[make-money-online]]></category>
            <dc:creator><![CDATA[Ethan Wang]]></dc:creator>
            <pubDate>Fri, 06 Mar 2026 11:24:28 GMT</pubDate>
            <atom:updated>2026-03-06T11:24:28.143Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/300/1*s3ujNXh8Ffro7PdauCXxuQ.jpeg" /></figure><p>Last month I built a trading bot for Kalshi — a prediction market where people bet yes or no on real-world events. In its first month, it executed roughly 4,000 trades across $600K in volume, completely autonomously.</p><p>Here’s what I built, what I learned, and what I wish someone had told me before I started. This is Part 1 of a series — future posts will go deeper on the math, engineering, and strategy.</p><p><strong>What is Kalshi?</strong></p><p>Kalshi is, at its core, pretty simple. Pick an event — “Will it rain in NYC tomorrow?” or “Will GDP come in above 3%?” — and buy a contract that pays out $1 if you’re right. The price of that contract is essentially the market’s implied probability. If you think something is more likely than the market does, you buy. If you think it’s less likely, you sell.</p><p>Think of it as a stock market, but instead of companies, you’re trading on outcomes.</p><p><strong>Why I Built a Bot</strong></p><p>I’ve always been interested in quant finance. Kalshi gave me something rare: a real market with real money where I could actually apply the theory. No need for a hedge fund job or a $25K minimum account. Just an API, some Python, and a willingness to lose money while learning.</p><p>So I did what any reasonable person would do — I built infrastructure.</p><p><strong>What the Bot Actually Does</strong></p><p>I want to be clear: this isn’t one strategy. It’s an infrastructure and data collection layer that supports multiple strategies — everything from arbitrage to longer-term directional plays. Think of it less like “a bot that trades” and more like “a platform I built so I can deploy and test strategies quickly.”</p><p>The tech stack is deliberately minimal. Python, my own re-implementation of Kalshi’s WebSocket APIs (for finer control over the way requests are sent and rate limits are handled), and standard libraries. Almost no dependencies. When you’re trading real money, you want to understand every line of code between your strategy and your orders.</p><p><strong>The Numbers</strong></p><p>In my first full month:<br>• ~4,000 trades across $600K in volume, fully automated<br>• Biggest single trade: a 9,100% return (yes, really — that one was luck)<br>• Worst day: a four-figure loss caused by a bug</p><p>That worst day deserves its own story. It came from a combination of not understanding when the Central Limit Theorem doesn’t apply and missing hard-coded stop losses in my code. Add in some bad luck and you get a very educational day. The lesson: your risk management can’t rely on statistical assumptions that don’t hold in your specific market. I’ll dig into this more in a future post on risk management.</p><p><strong>What Surprised Me</strong></p><p><strong>The Markets Are More Efficient Than You’d Think</strong></p><p>For their relatively low volume, Kalshi markets are surprisingly efficient. Don’t walk in thinking you’ll find easy mispricings everywhere. They exist, but they’re not free money — you’re competing with other bots, and the profitable ones are fast.</p><p><strong>Textbook Finance Breaks in Interesting Ways</strong></p><p>This was the most intellectually fun part. A lot of standard quant theory assumes continuous prices — Brownian motion, Black-Scholes, that whole world. On Kalshi, prices move in discrete 1-cent ticks, and contracts are bounded between $0 and $1. That combination breaks more approximations than you’d expect.</p><p>My workaround: logit-space transformations. Instead of modeling prices directly, you transform them into an unbounded space where the math behaves better, then transform back. It’s a small trick that makes a big difference.</p><p><strong>Engineering Matters More Than Math</strong></p><p>I genuinely don’t think you need to be a math wizard to trade profitably on Kalshi. But you absolutely need to be a competent engineer. Good engineering is the foundation on which your modeling and predictions actually live. Execution latency matters — your theoretical edge can evaporate if your orders are slow. Telemetry and portfolio tracking matter — you need to know what your bot is actually doing in real time. And if you’re running multiple strategies in the same market, managing their interactions is its own engineering challenge.</p><p>My worst day? The root cause was a math problem — but good engineering would have caught it before it ever mattered. Hard-coded stops, better telemetry, tighter guardrails. The math will be wrong sometimes. Your infrastructure needs to survive that.</p><p><strong>Advice If You Want to Try This</strong></p><p>Collect data. This is the single most important thing. Before you trade a dollar, collect data. Market prices, order books, event outcomes, timestamps — everything. Your strategies will come from patterns in the data, not from staring at a whiteboard.</p><p>Read the resolution rules. Every Kalshi contract has specific rules about how it resolves. The edge cases in those rules are where a lot of value hides — and a lot of losses come from not reading carefully enough.</p><p>Size up slowly. Backtest first. Then paper trade live. Then trade minimum sizes until you genuinely understand how your edge degrades with latency, and whether you’re facing adverse selection. Only then should you put real money behind it.</p><p>Respect the market. After almost 4,000 trades in a month, I’m now confident enough to let the bot run autonomously. But that confidence was earned, not assumed. My risk profile is well-understood — not because the risk is low, but because I’ve measured it.</p><p><strong>What’s Next</strong></p><p>I’m still running the bot, still iterating on strategies, still collecting data. Kalshi is a young market with a lot of room to grow, and I think the opportunity for quantitative traders is real — especially if you’re willing to do the engineering work that most people skip.</p><p>If you’re a CS student, a quant-curious developer, or just someone who thinks prediction markets are cool — I’d say start building. The barrier to entry is lower than traditional finance, the feedback loop is fast, and the lessons transfer everywhere.</p><p>—</p><p><em>I’m a CS and Math student who builds things. This is Part 1 of a series on building a Kalshi trading bot. Next up: the engineering and infrastructure that makes it all work.</em></p><p><em>If you want to follow along, subscribe.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2f686c214e25" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>