<?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[Unreasonable Effectiveness - Medium]]></title>
        <description><![CDATA[Summaries of key research in AI &amp; bioinformatics from AlphaFold author, plus ideas on quantitative venture investing. - Medium]]></description>
        <link>https://medium.com/unreasonable-effectiveness?source=rss----7df47b2673b7---4</link>
        <image>
            <url>https://cdn-images-1.medium.com/proxy/1*TGH72Nnw24QL3iV9IOm4VA.png</url>
            <title>Unreasonable Effectiveness - Medium</title>
            <link>https://medium.com/unreasonable-effectiveness?source=rss----7df47b2673b7---4</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 13:20:05 GMT</lastBuildDate>
        <atom:link href="https://medium.com/feed/unreasonable-effectiveness" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Abe Othman writes: should seed investors follow on?]]></title>
            <link>https://medium.com/unreasonable-effectiveness/abe-othman-writes-should-seed-investors-follow-on-a81eb8bc3af2?source=rss----7df47b2673b7---4</link>
            <guid isPermaLink="false">https://medium.com/p/a81eb8bc3af2</guid>
            <category><![CDATA[seed-investment]]></category>
            <category><![CDATA[data-driven-investing]]></category>
            <category><![CDATA[follow-ons]]></category>
            <category><![CDATA[angel-investing]]></category>
            <category><![CDATA[venture-capital]]></category>
            <dc:creator><![CDATA[Steve Crossan]]></dc:creator>
            <pubDate>Tue, 30 Jun 2020 11:30:36 GMT</pubDate>
            <atom:updated>2021-06-15T16:32:47.145Z</atom:updated>
            <content:encoded><![CDATA[<p>I’ve been thinking about rational follow-on strategy for a long time, and how to use <a href="https://medium.com/@stevecrossan/rational-venture-investors-should-have-bigger-portfolios-c9ea2ff4589">portfolio simulations</a> to decide when to make follow-on bets. Abe Othman at AngelList has just published some really interesting <a href="https://angel.co/blog/should-seed-investors-follow-on">work</a> that moves the debate forward. Here’s my brief explainer of what he did.</p><p>Abe looked at a ‘universe’ of 1218 seed stage convertible investments from 2014 to 2017. He then imagined a pool of 10 randomly selected investments each year for a total of 40 investments in the pool, and simulated 10,000 random pools this way.</p><p>For each pool of 40, he then looked at the performance of 3 strategies:</p><ol><li>Never follow-on. Invest an equal amount of capital into each of the 40 startups.</li><li>Always follow-on. For example, if a company you invested in in 2014 raises an upround in 2015, you invest the same amount of capital again in that company. Because he assumes a fixed amount of capital, each time you follow-on you forgo 1 opportunity to invest in a new company in 2015. The one you forgo is chosen at random from the 10 in the pool for that year.</li><li>As 2, but only when the valuation doubles or more at conversion into the upround.</li></ol><p>The final valuation of each portfolio is calculated 18 months after the end of the investment period, ie Summer 2019. It includes both realized and unrealized valuations.</p><p><strong>Results</strong></p><p>Cutting to the chase: Within this pool, never following is the best strategy most of the time, although the difference between the 3 strategies is not huge.</p><p>The exception is if you get into one of the top handful of companies in your ‘universe’ relatively early in your investing window.</p><p>Why? Because in the latter case the return on even the more expensive follow-ons can beat the expected return on the forgone new seed investment (at least for this dataset).</p><p>In a bit more detail:</p><ul><li>Never following had the highest median outcome across 10,000 pools.</li><li>Within each pool, Never beat the other 2 strategies most of the time.</li><li>But, Always following has a higher mean — that is it is more skewed by the occasional outlier.</li></ul><p><strong>Clusters</strong></p><p>The clustering diagram is really interesting, but took me a couple of passes to fathom:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*WqZJFUtsgnjYi_A3" /></figure><p>Each point on the plot represents 1 pool of 40 possible investments, and the performance of 2 strategies (Never, and Always) on that pool. So for example the rightmost point represents a pool in which Never achieved about 5.5x portfolio return but Always achieved almost 10x.</p><p>The clusters are very interesting. The 2 outlier clusters towards the upper left and the upper right, represent performance on pools which contain the <strong>single best investment</strong> from the universe of 1218 startups. This is ‘outlier strategy’ writ large — as I’ve written <a href="https://www.linkedin.com/pulse/power-laws-pooling-more-realistic-model-venture-returns-steve-crossan/?published=t">elsewhere</a> any portfolio that contains this investment will outperform every portfolio that does not.</p><p>But why does ‘Never’ sometimes outperform ‘Always’, even in portfolios containing that ‘universe winner’ (these portfolios are represented by the green circled cluster)? Surely if you follow on to that ‘universe winner’ you’ll always win.</p><p>The reason is that by having a strategy of always following on, you’ll sometimes miss the opportunity to invest in the ‘universe winner’. Remember that each dot represents a fixed pool of 40 companies, considered as 10 potential investments per year. The dots in the green circle represent times when the ‘Always’ strategy used up all its capital before it got to the ‘universe winner’, and thus missed out (presumably because a lot of the early bets converted). This is the ‘opportunity cost’ of doing follow on, and these are the cases when it didn’t work out.</p><p><em>Originally published at </em><a href="https://www.linkedin.com/pulse/abe-othman-writes-should-seed-investors-follow-steve-crossan/"><em>https://www.linkedin.com</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a81eb8bc3af2" width="1" height="1" alt=""><hr><p><a href="https://medium.com/unreasonable-effectiveness/abe-othman-writes-should-seed-investors-follow-on-a81eb8bc3af2">Abe Othman writes: should seed investors follow on?</a> was originally published in <a href="https://medium.com/unreasonable-effectiveness">Unreasonable Effectiveness</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Every great startup is a great product startup]]></title>
            <link>https://medium.com/unreasonable-effectiveness/every-great-startup-is-a-great-product-startup-ccd6411895db?source=rss----7df47b2673b7---4</link>
            <guid isPermaLink="false">https://medium.com/p/ccd6411895db</guid>
            <category><![CDATA[venture-capital]]></category>
            <category><![CDATA[product-management]]></category>
            <category><![CDATA[product]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Steve Crossan]]></dc:creator>
            <pubDate>Tue, 30 Jun 2020 11:30:02 GMT</pubDate>
            <atom:updated>2019-09-23T11:50:02.675Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0sHqCUsJ-ZQ1DcW6R1SKpw.jpeg" /></figure><p><em>In recent months I’ve had many conversations on product thinking with startups and with colleagues at Atomico and Firstminute. Last week </em><a href="https://www.atomico.com/team/siraj-khaliq/"><em>Siraj Khaliq</em></a><em> posted a </em><a href="https://medium.com/atomico/europes-product-management-problem-9061bc71dc99"><em>rallying cry</em></a><em> for European Product Management. And prompted in part by conversations with </em><a href="https://www.firstminute.capital/team"><em>Lina Wenner</em></a><em> I’ve been thinking about a related challenge: getting companies to think of themselves as product-first.</em></p><p>Since I’ve been spending most of my time investing in and advising startups, I’ve noticed a surprising pattern: many don’t deeply embrace the fact that they are <em>product</em><strong><em> </em></strong>businesses.</p><p>Of course most companies talk about the importance of product, but often product thinking isn’t embedded deeply in the company from top to bottom. I worry that not enough founders think about this when planning and building their company.</p><p>This may be a partial view (I spent a dozen years in product functions at Google and DeepMind), but to me the company <em>is</em> the product; it’s everything that matters. Whether you’re a consumer app or an API or a platform for computationally creating new widgets, you’re delivering value to a set of users. To do that in a scalable way you have to think of your business as a product.</p><p><em>Product is a process, not a static thing</em></p><p>Being great at product means being great at continuously making and improving products. The heart of it is a loop that many <a href="https://mikebracken.com/blog/the-strategy-is-delivery-again/">others</a> have described in different ways; this is my version:</p><ul><li>Make an informed guess about who the users are and how you can bring them value.</li><li>Build something as quickly as possible that brings them value, all the while trying to understand their world in as much detail as possible. Challenge yourselves to do this in weeks to months.</li><li>Get at least one user (preferably more) using what you’ve built, in their <em>real workflow</em>.</li><li>Measure how much value they are getting. This bit will take a lot of thought, and is likely to be very specific to the use case. That use case might be a messy workflow involving multiple people which is hard to measure. Nonetheless if you’re scrappy enough, you can find a way to get a handle on it. I know of one automation company that starts by videoing everything that’s happening on the factory floor and uses video analysis to measure how it can be improved.</li><li>Iterate until they are getting so much value that they can’t do without it. Sometimes this involves picking a different set of users, doing a related but different thing that is more valuable, or more scalable, or more impacted by your technology edge.</li><li>You know you’re getting there when your users are telling other people to use what you’ve built. You can also tell them (but wait until it’s awesome).</li><li>Keep making it better using the same process. Keep checking you’re measuring the right things. You want your users not be able to live without your product. If that’s not quite true yet, why not ?</li></ul><p>The holy grail of ‘product market fit’ is the (temporary) outcome of this process — <em>the process itself is much more important</em>. If the company lives the process, product market fit will come. But it’s not a one-time event: it needs to be continually nurtured.</p><p>In the end you should *only* care about scalably creating real user value. Your fab technology or algorithm or data or team is a means to that end: if something else (especially something simpler) does the job better, you should embrace it. Many companies begin with a technical insight or breakthrough — and that’s often what makes new value creation possible — but if you get too wedded to it without finding those users and that value you’ll sooner or later run into trouble.</p><p><em>Warning signs to look out for</em></p><ul><li>Your revenue comes mostly from proof-of-concept projects with potential customers. In fact, you probably want to avoid these kinds of projects entirely (especially with innovation departments of larger customers). I’ve lost count of the number of startups who have gone down this route and never emerged with a full contract, let alone a scalable product.</li><li>Your product sells well (you’ve identified a real need) but customers are not using it more over time. This can be an early warning for future churn. But if you identify this pattern quickly, you can find out why, and fix it.</li><li>Onboarding takes time and/or work from you and/or the customer. This can drift into customisation and service revenues. How could you make it so that the customer gets immediate, obvious value ?</li></ul><p><em>It has to be in the DNA of the company</em></p><p>“The DNA of the company” is a bit of a cliché but captures the idea that this product process thinking needs to be foundational, and to run throughout the company from top to bottom. That’s because it has to be something that the company does every day — that’s the only way to get to these rapid iteration cycles.</p><p>This doesn’t mean that founders always have to be experienced product managers. But they have to be willing to become great product <em>leaders.</em></p><p>This usually means that one of the founders needs to spend most of their time making the company great at product, and communicating its importance to everyone in the company <em>all the time</em>. Sometimes (especially early on) that means they themselves are doing product management, product design, or both. But they don’t have to be the ones doing that, as long as they are continually communicating who the users are, what the value is, and helping the company to shorten those learning cycles.</p><p><em>You can hire product managers, but you can’t hire DNA</em></p><p>Another common pattern, even for companies coming up to Series A or Series B rounds, is to hear ‘we need some help with product / we know we need to hire product / we know we need a user researcher’. That can sometimes be a warning sign; a lack of strong product market fit is one thing, but if the founders aren’t focused on product themselves that’s much more worrying (and will in turn make it hard to hire well in this area).</p><p>Above all, this is a plea for founders (at least one member of the founding team) to spend most of their time on product — including all of its different aspects:</p><ul><li>Spending significant time with users in their real workflow</li><li>Thinking about the way the market is structured, and where your product fits in. For example I’ve seen companies go from thinking they will sell warehouse automation software to thinking they will run their own warehouses, because they will capture more value that way if successful, and end up in a more dominant position.</li><li>Thinking about the other tools in the workflow (or more generally, in the users’ life), and how you fit in (or better, replace them).</li><li>Working out how to measure how successful you’re being (really successful, not <a href="https://techcrunch.com/2011/07/30/vanity-metrics/">vanity successful</a>), how to get that information as quickly as possible, and how to act on it as quickly as possible.</li><li>Frequently changing the product, testing how successful your changes have been, and pushing good changes to everyone.</li></ul><p><em>Sure, but that’s not really my company</em></p><p>Be wary of giving yourselves excuses for why this doesn’t apply to your company. “It takes a long time to get into our kinds of customers.” “We have to educate the market.” “We’re not that kind of business.” “We’re an enabling technology, not a product.” If you’re thinking this way, it’s eventually going to catch up with you. Sometimes you can keep raising on the promise of your tech, or the market, or your natural charm — but in the end you do really want to build a great business, don’t you? Challenge yourselves to find ways to get value to users quickly. Think about it: if what you’re doing has value, you should be able to get someone to use it.</p><p>I’ve seen a lot of ‘long sales cycle’ / POC-first companies in the last 2 years, especially in industrial automation. But the very best of them had moved quickly away from that mode to create an offering that was obvious to the customer, created great value for them immediately, and was soon impossible to imagine life without.</p><p>They did this by relentlessly focusing on removing friction to that ‘aha’ moment. Back and forth on legals? Do away with them; give the solution to the customer for free to begin with. Takes a long time to install and program to the customers’ environment? Make it so an operator can start using it straight away with no programming or integration. (This is a very common pattern in disruptive companies. I know of one company that avoided the need to integrate its image classification software with corporate image management solutions by distributing a smartphone version by word of mouth among operators, who simply took pictures of what they saw on their screens). Especially in older industries, there often seem to be a whole series of such blockers — but that’s your opportunity. Relentlessly eliminating them one by one (or even better, end-running around them all) is your path to greatness.</p><p><em>Hmm. Sounds complicated.</em></p><p>The good news is product is one of those things that sounds hard but isn’t really. What feels challenging is that it’s a lot of different things, and your company needs to do them all the time.If it feels daunting, you can find good product coaches these days who can be there when you need them to keep you on track.</p><p>But the good news is that if you adopt the <strong>mindset</strong>, you’ll find that your organisation starts doing those things. You as founder don’t have to do all of them yourself, but you have to make a company that wants to do all of them (because when you get into the flywheel, it’s exhilarating for everyone).</p><p>Together you’ll start to wrestle something new and impactful into being, and if you keep doing it, and keep getting faster at learning, acting and learning again, the flywheel will start to pick up speed. Just don’t hide from it.</p><p><em>Thank you to Siraj and Lina for discussion, and to Suryansh Chandra of </em><a href="https://automata.tech/"><em>Automata</em></a><em>, Jacob Haddad of </em><a href="http://accurx.com/"><em>AccuRx</em></a><em> and Eleanor Warnock at Atomico for reading and commenting on earlier drafts.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ccd6411895db" width="1" height="1" alt=""><hr><p><a href="https://medium.com/unreasonable-effectiveness/every-great-startup-is-a-great-product-startup-ccd6411895db">Every great startup is a great product startup</a> was originally published in <a href="https://medium.com/unreasonable-effectiveness">Unreasonable Effectiveness</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Power laws with pooling: a more realistic model of venture returns]]></title>
            <link>https://medium.com/unreasonable-effectiveness/power-laws-with-pooling-a-more-realistic-model-of-venture-returns-fbfb1029a62c?source=rss----7df47b2673b7---4</link>
            <guid isPermaLink="false">https://medium.com/p/fbfb1029a62c</guid>
            <category><![CDATA[seed-investment]]></category>
            <category><![CDATA[jupyter]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[venture-capital]]></category>
            <category><![CDATA[angel-investing]]></category>
            <dc:creator><![CDATA[Steve Crossan]]></dc:creator>
            <pubDate>Tue, 30 Jun 2020 11:29:05 GMT</pubDate>
            <atom:updated>2019-09-24T14:27:56.221Z</atom:updated>
            <cc:license>http://creativecommons.org/licenses/by/4.0/</cc:license>
            <content:encoded><![CDATA[<p><em>TL;DR: In the </em><a href="https://medium.com/@stevecrossan/rational-venture-investors-should-have-bigger-portfolios-c9ea2ff4589"><em>last post</em></a><em> I built some models of venture portfolios of different sizes based on the idea that venture outcomes are </em><a href="https://en.wikipedia.org/wiki/Power_law"><em>powerlaw</em></a><em> distributed. The conclusion was, other things being equal, bigger portfolios should do better, with around 150 investments being table stakes. Those models assumed each investment was independent, with exactly one investor. The pooled model presented here is more realistic; overall it dampens expected returns without changing the overall pattern that more is better. Interesting implications are that dealflow and brand are key for VC.</em></p><p>In the <a href="https://medium.com/@stevecrossan/rational-venture-investors-should-have-bigger-portfolios-c9ea2ff4589">last post</a> we simulated over 45M independent outcomes (sum(x=5-&gt;300) x*1000).</p><p>But in practice portfolios are not independent and the universe of investible businesses is much smaller. Our first model is more like a simulation of venture <em>builders</em>, where the fund creates its own independent businesses.</p><p>How about instead we create a <em>pool</em> of businesses that receive investment, and our portfolios sample from this pool? In other words, in the investing life of a fund, a finite number of fundable businesses will be created. During this period, every fund will be choosing from that finite set of businesses.</p><p>How big should that pool be ? Well according to <a href="https://news.crunchbase.com/news/q3-2017-global-report-vc-deal-dollar-volume-projected-reach-post-dot-com-highs/">crunchbase</a>, in 2016/2017 around 3,500 angel and seed funding rounds happened globally per quarter:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/320/0*5h-OdkayB82iHR7X.png" /></figure><p>Let’s say there’s some under-counting, and that the number is growing, and round up to 5,000 rounds per quarter.</p><p>So in the 3–5 year investing life of a typical fund, there’s a unverse of 60,000–100,000 companies that they theoretically could invest in (of course they wouldn’t see that many deals, but that doesn’t affect our model).</p><p>Let’s split the difference, and take a pool size of 80,000 companies. What I’m going to do next then is generate this pool from the same power law distribution as the previous post, and have each portfolio draw its investments at random from the pool.</p><p>Here’s a histogram of 10,000 companies drawn at random from this pool, compared to the correlation ventures data, and the independent draws data from the original post:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/488/1*UsbwWU6jz44qtYc2QlRhyQ.png" /><figcaption>Outcomes drawn from pool</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/488/1*SYA7VcDNOo7MyeLuQ3an7Q.png" /><figcaption>Original Post</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/838/1*OyezVIOcuvTI45r0Ox65gQ.png" /><figcaption>Correlation Ventures real-world data</figcaption></figure><p>Note this is only a gut-check since I don’t have access to the underlying data and so can’t fit the model, but it looks reasonable. I’ve also tried a variety of parameters for the power law distribution, and the overall trends are robust.</p><h4>Performance</h4><p>Let’s use the same method as before, sampling 1000 portfolios of each size between 5 and 300 investments, and looking at average statistics across each set of 1000. We’re interested in the triple rate (how often a random portfolio could be expected to triple the fund) and the failure rate (how often it could be expected to return &lt;1x).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/479/1*FNA-6zh23fD-Qa5pFhqpSw.png" /><figcaption>Portfolios drawn from a pool of 80,000 fundable companies</figcaption></figure><p>Compared to the original independent draws model:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/479/1*r-6og3o1cNKz9LGsDB79XA.png" /><figcaption>Portfolios drawn independently from a powerlaw</figcaption></figure><p>You can see that the pattern is very similar — it’s hard to tell the difference.</p><h4>Mean Performance</h4><p>The differences between the models come out when you look at the mean performance:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/481/1*hi8m22tGTxP0MnbIiTRNsA.png" /><figcaption>Portfolios drawn from a pool of 80,000</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/484/1*HndUsBn19XvgLrj03rBMsQ.png" /><figcaption>Portfolios drawn independently from distribution</figcaption></figure><p>This time the mean is behaving more as we would expect; it’s not affected by the occasional extreme outlier that we get in 45M+ independent draws. It still looks high however compared (anecdotally) to historical returns. This may again just be the effect of outlier portfolios — the small number that do really really well as you’d expect with samples from an underlying power law. Again, median may give a better sense of ‘typical’ results:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/481/1*COzhTt5VuiQ4nFMh4sTAKg.png" /><figcaption>Median outcomes by portfolio size with investments drawn from a pool of 80,000</figcaption></figure><p>This looks more in line with historical data (and again very similar to the unpooled model). Again, median performance continues to improve through this range.</p><h4>Conclusions</h4><p>The pooled model of venture returns is intuitively more realistic than independent draws from a power law distribution, but the underlying trends remain the same: at least within this range, more investments are better, and 100–150 seem to be ‘table stakes’ in the sense that after this, no portfolio in 1000 samples loses money.</p><p>Thinking about a cohort of ~80,000 investible companies within the investment lifetime of a fund leads to some interesting insights:</p><ul><li>Even at that scale, the biggest return within the cohort is a significant portion of the overall returns. In the pool that I generated in this post, the maximum return of any single investment is 18660x, comparable to Jerry Neumann’s estimate of 10000x for the return of Andy Bechtolstein’s $100K cheque into Google, or the return on YC’s investment in DropBox.</li><li>What’s especially interesting is that single return is <em>over 5% of the entire return</em> from those 80,000 companies. This implies that in any given cohort, fund performance will be dominated by whether you get into that 1 investment :). Any fund that does will tend to outperform every other fund in that cohort.</li><li>In fact 5% may be a significant underestimate. I simulated 1000 cohorts of 80,000 companies each; the average ratio of the single most successful company to the return of the entire pool was just over 17%.</li></ul><p>What would you conclude if you took this modelling seriously ? Firstly, I think, deal flow is key: can you build a machine that can see and reliably process a high number of deals, while providing a great service to entrepreneurs ?</p><p>Secondly, and related, <em>brand</em> is enormously important. Success breeds success. Brand improves deal flow, and helps a fund get into deals that it likes. So the early years of an investment company should try to build up the brand.</p><p>[If you want to read &amp; experiment with the code that generated these graphs, you can copy it from <a href="https://colab.research.google.com/drive/1EAFQxQ68FfsThpVcNU7m8vqt4UZL0Le1#scrollTo=lYtQwg3pZMKu">Google Drive</a> or grab it from <a href="https://github.com/stevecrossan/VCModel/blob/master/PortfolioSim0_5.ipynb">GitHub</a>. If you’d like to help me extend it or turn it into a more re-usable library, email me stevecrossan @ gmail or send me a CL :)]</p><p>Next post: <a href="https://www.linkedin.com/post/edit/6579368703962472448/">follow on strategy</a>.</p><p><em>Thanks to </em><a href="http://reactionwheel.net/2017/12/power-laws-in-venture-portfolio-construction.html"><em>Jerry Neumann</em></a><em> for original work that led me down this path, to </em><a href="https://www.sethlevine.com/archives/2014/08/venture-outcomes-are-even-more-skewed-than-you-think.html"><em>Seth Levine</em></a><em> for the same, and for comments on drafts, and to </em><a href="http://nathanbenaich.github.io/"><em>Nathan Benaich</em></a><em>, </em><a href="http://correlationvc.com/team/detail/david-coats"><em>David Coats</em></a><em>, </em><a href="https://www.linkedin.com/in/elliotoco/"><em>Elliot O’Connor</em></a><em> and </em><a href="https://www.linkedin.com/in/arekwylegalski/"><em>Arek Wylegalski</em></a><em> for comments and discussion.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fbfb1029a62c" width="1" height="1" alt=""><hr><p><a href="https://medium.com/unreasonable-effectiveness/power-laws-with-pooling-a-more-realistic-model-of-venture-returns-fbfb1029a62c">Power laws with pooling: a more realistic model of venture returns</a> was originally published in <a href="https://medium.com/unreasonable-effectiveness">Unreasonable Effectiveness</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Modelling suggests rational venture investors should have bigger portfolios]]></title>
            <link>https://medium.com/unreasonable-effectiveness/rational-venture-investors-should-have-bigger-portfolios-c9ea2ff4589?source=rss----7df47b2673b7---4</link>
            <guid isPermaLink="false">https://medium.com/p/c9ea2ff4589</guid>
            <category><![CDATA[jupyter]]></category>
            <category><![CDATA[data-modeling]]></category>
            <category><![CDATA[angel-investing]]></category>
            <category><![CDATA[seed-investment]]></category>
            <category><![CDATA[venture-capital]]></category>
            <dc:creator><![CDATA[Steve Crossan]]></dc:creator>
            <pubDate>Tue, 30 Jun 2020 11:28:38 GMT</pubDate>
            <atom:updated>2018-04-19T11:07:39.505Z</atom:updated>
            <content:encoded><![CDATA[<h4>A model of venture returns following a power law distribution, under reasonable assumptions, points in the direction of larger portfolios — at least 150 investments per fund.</h4><p>This post explores some ideas about venture capital returns and portfolio size. It builds on other explorations including <a href="http://reactionwheel.net/2017/12/power-laws-in-venture-portfolio-construction.html">Jerry Neumann’s</a> and <a href="https://www.sethlevine.com/archives/2014/08/venture-outcomes-are-even-more-skewed-than-you-think.html">Seth Levine’s</a>. I wrote a simple simulator in a <a href="https://colab.research.google.com/">Google Colab</a> notebook, which generated the graphs below. After the <a href="https://medium.com/@stevecrossan/power-laws-with-pooling-a-more-realistic-model-of-venture-returns-fbfb1029a62c">next post</a> I’ll post a link to the notebook which will allow you to run the code and experiment with different parameters.</p><p>Jerry Neumann’s posts explore the consequences of the idea that venture returns (the returns to investments in early stage, high growth potential companies) are not normally distributed, as public market returns are, but follow a power law distribution.</p><p>The first step is to look at the evidence that this might be so, and if so what are the approximate parameters of the distribution.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/838/0*02ZKXuMMfa9_V87e.png" /></figure><p>This graph is from Seth Levine’s blog; the underlying data comes from <a href="http://correlationvc.com/">Correlation Ventures</a>, but I don’t have access to the data itself.</p><p>A priori it looks as though it would be a good fit for a power law distribution.</p><blockquote><strong>Sidebar: What’s a power law distribution?</strong></blockquote><blockquote>In a <a href="https://en.wikipedia.org/wiki/Power_law">power law</a> distribution, an outcome’s likelihood is inversely proportional to its size:</blockquote><blockquote>p(outcome) = C*outcome^-alpha</blockquote><blockquote>alpha is a parameter which governs the <strong><em>skew</em></strong> of outcomes (smaller alpha -&gt; more skew -&gt; more big outcomes)</blockquote><blockquote>Some <a href="http://reactionwheel.net/2015/06/power-laws-in-venture.html">evidence</a> suggests alpha for venture investments clusters around 2.</blockquote><blockquote>C is a constant multiplier parameterized by alpha and the minimum outcome:</blockquote><blockquote>C = (alpha-1)* (min_outcome)^(alpha-1)</blockquote><blockquote>One wrinkle with using a power law is that the minimum outcome can’t be zero (even though it obviously can in venture). We’ll come back to this later but in the meantime a minimum return of 0.35 together with an alpha of 2.0 seems to approximate the data. What this roughly means is that at the low end of outcomes, I’m averaging out zeros and very low outcomes so the minimum is a return of 0.35. Later I’ll explore some alternatives such as having a fixed number of zeros, but the overall statistics will turn out quite similar.</blockquote><p>I wrote some simple code to sample outcomes (multiples on invested capital) from the distribution. Then a histogram of 10,000 outcomes looks like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/488/1*SYA7VcDNOo7MyeLuQ3an7Q.png" /></figure><p>Reading this you can see that from 10,000 simulated ‘companies’, just over 63% had a return of 0–1x, and 0.5% had a return &gt; 50x. Looks like it’s in the right ballpark compared to the histogram from Seth’s blog.</p><p>Then I ran monte carlo simulations of a large number of portfolios of different sizes, and looked at some of the metrics that come out. A “monte carlo simulation” is a fancy way of saying you take a lot of samples — in this case I created 1000 random portfolios each of size 5, 6, 7, 8, … investments all the way up to 1000 portfolios of size 300.</p><p>From the histogram above (and experience) we know the general pattern will be a large number of small outcomes and a very small number of very large outcomes. To examine what that means for portfolio size, I simulated 1000 portfolios at every portfolio size between 5 and 300, and then looked at some overall statistics.</p><h4>Chance of tripling the fund</h4><p>The core thing we’re interested in is, at each portfolio size, what are the chances of failing (returning &lt; 1x), tripling, 5x-ing and 10x-ing the fund ? The x-axis is portfolio size, the y-axis is the percentage time you see the outcome:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/479/1*r-6og3o1cNKz9LGsDB79XA.png" /></figure><ul><li>With just 5 investments in the portfolio, you have a more than 40% chance of losing money. This goes down rapidly until portfolios sizes between 100 and 150 when there’s very little or zero chance (what this means is that none of the 1000 random portfolios at that size lost money).</li><li>The chance of tripling the fund gently increases as you increase the portfolio size, though its rate of increase slows.</li></ul><p>That’s also reflected when you look at the <strong>minimum</strong> return of any of the 1000 portfolios at each size, which shows that after a size of about 150 investments per portfolio, no portfolio loses money, and the minimum keeps increasing. Here x-axis is portfolio size:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/481/1*iCmxtEyKY5jSsysgp4ksVA.png" /></figure><h4>Mean and median returns</h4><p>Let’s look at the mean returns at each portfolio size:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/484/1*HndUsBn19XvgLrj03rBMsQ.png" /></figure><p>Ouch. What’s going on here ? Well the thing about power law distributions is that they have no maximum value (though values become increasingly rare as they get larger). When we created our sample portfolios we independently sampled sum(x=5-&gt;300) x*1000 companies which is over 45M independent companies. Occasionally you get a crazily large outcome, and that obscures the results making the mean harder to discern (those crazy outcomes have more of an effect on the smaller portfolios). I’ll talk below about whether this is realistic, and what an alternative might look like.</p><p>Maybe the medians will give a better idea — 50% of the portfolios do as well or better than this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/481/1*_6H_u-Khk2AoJ7SCVi0vhQ.png" /></figure><p>Again, it looks like this continuously goes up, though the rate of improvement slows down, until at 300 investments the median portfolio is returning 2.6x.</p><p><strong>Some conclusions</strong></p><p>So all in all this model seems to show that the larger your portfolio, the better you’re likely to do. Why should that be so? With a power law distribution of returns, success is all about hitting that rare outlier. Other things being equal, the more chances you have, the greater the chance of a big hit.</p><p>I think we can do better than this simple model though.</p><ul><li>The model is very sensitive to changes in the value of alpha (lower alpha is better), though the overall trend that more investments are better is very consistent.</li><li>I’d like to get hold of an actual historical dataset (rather than the summary statistics) and try to fit it, to see if we can understand some realistic ranges for alpha.</li><li>The scenario is unrealistic in one important respect — it assumes each company in the 45M+ draws is completely standalone, with only 1 investor. It’s more realistic to say that there is a pool out there of funded companies, and investors are picking their investments from this pool. Separately I’ve done some modelling of that as well, and the trends are similar though the triple rate is a bit lower, depending on the size of the pool. I’ll save that for a follow-on post.</li><li>Anecdotal evidence suggest that the sector as a whole doesn’t do as well as this model suggests it should — I’d like to get some historical data to examine that also. If it’s true, it may be that alpha is too low, or the pooling effect may play a role.</li><li>I’ve also experimented with various ways of dealing with the unrealistic positive minimum (such as zero-ing out returns below a threshold). Again, this tends to lower returns as you’d expect, but doesn’t affect the trend.</li><li>Finally, the big issue: Can you manage the quality of your decisions (and the help you can provide) with such a big portfolio? I would argue, only with a lot of automation &amp; efficiency, and I’ll come back to that in a later post.</li></ul><p>This is part 1 of a series. <a href="https://medium.com/@stevecrossan/power-laws-with-pooling-a-more-realistic-model-of-venture-returns-fbfb1029a62c">Next post</a>: pools, sampling effects.</p><p><em>Thanks to </em><a href="http://reactionwheel.net/2017/12/power-laws-in-venture-portfolio-construction.html"><em>Jerry Neumann</em></a><em> for original work that led me down this path, to </em><a href="https://www.sethlevine.com/archives/2014/08/venture-outcomes-are-even-more-skewed-than-you-think.html"><em>Seth Levine</em></a><em> for the same, and for comments on drafts, and to </em><a href="http://nathanbenaich.github.io/"><em>Nathan Benaich</em></a><em>, </em><a href="https://www.linkedin.com/in/elliotoco/"><em>Elliot O’Connor</em></a><em> and </em><a href="https://www.linkedin.com/in/arekwylegalski/"><em>Arek Wylegalski</em></a><em> for comments and discussion.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c9ea2ff4589" width="1" height="1" alt=""><hr><p><a href="https://medium.com/unreasonable-effectiveness/rational-venture-investors-should-have-bigger-portfolios-c9ea2ff4589">Modelling suggests rational venture investors should have bigger portfolios</a> was originally published in <a href="https://medium.com/unreasonable-effectiveness">Unreasonable Effectiveness</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>