Turning $5M Into $100M
Balancing risk, reward, and conviction in venture capital portfolio construction.
In 2015, Collaborative Fund made an unusually bold bet—investing $5 million—20% of a $25 million fund—into a single startup. Less than five years later, that one investment returned over $100 million, singlehandedly quadrupling the entire fund.
Was this risk justified? Should a fund spread $25 million across twenty-five $1 million bets or five $5 million ones?
This wasn’t only a lucky outcome; it was also a deliberate choice rooted in conviction. Venture capital returns famously follow a power law—a tiny fraction of investments often generate the majority of returns, both for a given fund and the entire industry each year. The challenge is structuring a portfolio to catch at least one of these breakout outcomes without over-diversifying and diluting returns.
This post examines the math behind portfolio construction, the trade-offs between concentration and diversification, and why exceptional outcomes often come from high-conviction positions.
The flawed math behind venture portfolio construction
Venture investing is often called an art, but that doesn’t stop people from trying to turn it into a science. Many funds use math to guide portfolio construction, helping to estimate how many bets to place and how much to allocate per investment in a given fund. These range from probability-driven frameworks like the Kelly Criterion to simulation-based modeling like Monte Carlo.
No model perfectly captures the reality of venture capital. Each has flaws, assumptions, and blind spots. Some fail to account for the extreme variance in startup outcomes, while others assume investors can accurately predict success rates—something historical data suggests is extraordinarily difficult.
Still, these frameworks can help ballpark an approach. Below are some of the more common ones and their limitations.
Kelly criterion
Originally developed for gamblers, the Kelly Criterion calculates the optimal bet size to maximize long-term returns. In VC, it can be used to help determine the proportion of a fund to allocate per startup.
Why it isn’t a perfect fit for VC:
- You don’t know exact probabilities of success or payouts—Kelly assumes you do.
- It assumes you can reinvest winnings each round, but in VC, capital is typically locked up for a decade.
Back of the envelope power law math
Some VCs take a probability-driven approach to ensure their portfolio includes at least one breakout winner. The basic logic goes like this:
- Assume a startup has a 5% chance of delivering a 20x+ return.
- If you invest in just one company, there’s a 95% chance you miss a big winner.
- If you invest in two, the chance of missing a winner drops to 90% (95% × 95%).
- Keep investing, and the odds of missing a winner continue to shrink.
From here, one can set a threshold for failure—say, wanting a 20% or lower chance of having zero breakout winners—and backsolve to determine how many investments to make (32 in this case).
Why it isn’t a perfect fit for VC:
- The failure threshold is arbitrary—too low, and you dilute your best bets; too high, and you risk missing a winner altogether.
- It treats startup outcomes as binary (big winner or bust), ignoring the reality that many exits fall somewhere in between.
- A more nuanced approach would factor in different return profiles and optimize accordingly.
Monte Carlo simulations
This method runs thousands—sometimes millions—of randomized scenarios based on assumed success rates and exits, testing different allocation strategies to identify the best.
Why it isn’t a perfect fit for VC:
- Like all models, Monte Carlo is only as good as its inputs. Assumptions about success rates and exit values are just that—assumptions. Even using robust historical venture data doesn’t guarantee reliability.
- For example, AngelList ran Monte Carlo simulations using data from 3,000+ past investments. One might expect a dataset of that size to produce generalizable conclusions. However, adding just a few breakout investments drastically altered the projected returns:
- Adding Peter Thiel’s Facebook seed investment shifted the average return from 2.7x to 5.9x.
- Including Google’s and Uber’s seed rounds pushed it to 27.7x.
- Monte Carlo is useful for modeling potential outcomes and assessing risk, but it can’t predict the future. Even running simulations on a perfectly curated dataset of every VC deal from the last decade wouldn’t ensure reliable forecasts—past performance isn’t indicative of future results.
So… how should you construct your portfolio?
Each of these methods has flaws, but together, they offer a rough picture of effective venture portfolio construction.
They suggest that early-stage funds should aim for 25-40 investments per fund. This range balances the potential for capturing breakout winners while avoiding excessive dilution.
However, portfolio construction isn’t just math—it’s strategy. Models can give a broad range of appropriate portfolio concentrations, but they can’t tell you exactly how many bets to make or when to make a single more concentrated bet. Our decision to allocate $5M—20% of our fund—into a single company wasn’t model-driven. In fact, most models would have cautioned against it.
Below are some strategic reasons for varying degrees of portfolio concentration.
Arguments for less concentration
- Firm longevity: A broader portfolio increases the odds of landing a power-law outcome. An overconcentrated fund that strikes out may compromise future fundraises. LPs won’t tolerate a decade of locked up capital for a <1x MOIC. Diversification may sacrifice upside in a single fund, but it can keep a firm in the game. If our $5M bet had missed and the rest of the portfolio underperformed, our concentration strategy would have been scrutinized, potentially leading to some tough conversations with LPs.
- Leading vs. following: Larger checks often mean leading rounds, which comes with extra responsibilities—diligence, term negotiation, and board seats to name a few. Not all firms are structured and willing to lead.
- Poker analogy: Fred Wilson compares early-stage investing to poker. Players put in a small ante to see their cards before deciding whether to bet a larger amount. Similarly, a small early check secures a seat at the table, giving investors time to assess execution from the inside. A diversified approach in early stages increases exposure to potentially strong performers, enabling investors to deploy follow-on capital with greater conviction in subsequent rounds.
Arguments for more concentration
- Conviction: When a standout team, market tailwinds, and early traction align, a concentrated investment can be the right move. While conviction should underpin every investment (ideally), some opportunities inspire greater confidence than others. When that happens, check size should reflect it.
- Deeper involvement and better follow-on decisions: A concentrated portfolio lets investors be more hands-on, building stronger relationships with founders and offering targeted support. With fewer companies to track, follow-on decisions are also more informed.
- Quality over quantity: More deals can mean spreading capital too thin or investing in lower-quality startups just to fill a portfolio. If a fund can reliably identify its strongest investments at the time of investment, concentrating more on those maximizes returns.
- Top-tier funds tend to concentrate capital: Data shows higher-performing funds tend to be more concentrated. Whether through larger initial checks or follow-on investment, they focus on companies with the highest potential for outsized returns.
Conviction and check size dynamacy
Our $5M bet wasn’t one of five evenly sized investments within our $25M fund—it was an outlier. The fund made 20 investments, keeping it relatively concentrated, but this check was unusually large. It was placed with conviction, and in hindsight, it paid off.
Had this bet failed, we would have reassessed our approach to concentration. Big swings carry big risks, and a miss would have been a tough reminder of the balance between focus and diversification.
While we haven’t committed 20% of a fund to one company since, this experience reinforced conviction’s role in our portfolio construction. Since then, we’ve leaned on it more heavily, ensuring we’re positioned to bet big when the right opportunity presents itself.
Venture capital is as much an art as a science. The real challenge isn’t just portfolio sizing—it’s knowing when to take a concentrated risk and having the discipline to live with the outcome, whether it’s a hard-earned lesson or a $100M success.