Silent Evidence in Investing: The VC Graveyard and Mutual Fund Survivorship

When you hear about venture capital, you hear success stories.

Sequoia's role in Apple and Google. Kleiner Perkins's role in Amazon and Netscape. Benchmark's early investment in eBay. These are the public narratives.

The narratives are true. These firms did invest in legendary companies. But the truth is heavily selected.

The Invisible Graveyard

Here's what you don't hear:

A top-tier VC firm might invest in one hundred companies in a decade. Of those hundred:

Maybe three to five return a meaningful multiple on the fund's capital — enough to generate strong returns for the fund overall.
The rest return zero, return small multiples, or lose money.
Some companies are early failures. The business model doesn't work. The team doesn't execute.
Some companies seem to work for years, then fail spectacularly.
Some just slowly wind down.

The business works because the winners are so extreme — a single Apple-like exit can return many multiples on the entire fund. But it works only because the failures are massive in number and are silent.

You will never read a press release celebrating a failed investment. You will never see a profile of the VC who picked the ninety-five companies that returned zero. The ninety-five are invisible.

The public narrative is "VCs are brilliant talent-spotters." The private reality is "VCs are running an extreme portfolio strategy where most picks fail, but the few that work return so much that the overall fund still succeeds."

Both narratives are true. But only one is visible.

The Hedge Fund Database Problem

Here's another version of the same bias in hedge funds:

Academic researchers and investors use "hedge fund databases" to study hedge fund performance. The data shows impressive returns — often 10%+ per year with low volatility.

But these databases have a systematic problem: they exclude defunct funds.

Hedge funds that blow up and disappear are not in the database. Hedge funds that underperform and shut down are not in the data. Only surviving funds are represented.

This creates massive survivorship bias. A researcher who looks at the database might conclude hedge funds deliver 10%+ annual returns. But that conclusion is based on a sample that excludes all the failures.

When researchers have gone back and tried to estimate the impact of the missing data, they've found that the true average return of hedge funds, including the dead ones, is much closer to the market average or worse.

The 10% performance was a mirage created by survivorship bias.

Mutual Fund Selection Bias

The same bias affects mutual fund performance data.

Mutual fund databases publish returns for all surviving funds. But when a fund closes because of poor performance, it disappears from the database. Bad funds that underperform are delisted.

This makes the average mutual fund performance look better than it actually is, because the worst performers are no longer in the data.

A researcher looking at historical mutual fund returns will see that the average fund beat the market. But that's partly because the funds that underperformed by the most are no longer in the database.

If you could include the dead funds, the average performance would look much worse.

What This Means for Your Own Investing

When you evaluate an investment manager, be skeptical of track records.

A good track record might mean the manager is skilled. Or it might mean the manager is alive — still running, still reporting.

The best managers, by definition, are the ones still running and still reporting. The worst managers have disappeared. This naturally creates a bias toward the surviving managers looking good.

This is not to say track records are useless. But they are a ceiling, not a typical case. The typical case includes all the managers who failed.

The Survivorship Bias in Investment Strategy

The same logic applies to investment strategies that are "published" or "researched."

Academic papers describe strategies that have worked. Investors read the papers and try to replicate the strategies.

But the papers were published because the strategy worked in the past. The strategies that failed are less likely to be published or researched. They are silent.

This creates a bias where you read about all the strategies that worked and miss the strategies that looked similar but failed.

The Practical Implication

When evaluating an investment manager or a fund:

Ask about the graveyard. How many funds or strategies has this firm tried that no longer exist? What happened to them?
Check for selection bias in the data. Is the track record based on all funds ever managed, or just the surviving ones?
Assume the visible success includes luck. The managers who survived might be skilled, but they also were lucky. The counterfactual — equally skilled managers who happened to face different market conditions — is not visible.
Weight recent performance less. A manager who has beaten the market for 10 years might still just be lucky. A manager who has beaten the market for 20 years is more likely to be skilled, but even then, some luck is involved.
Consider that the best strategy might be invisible. The strategy that would work best might be one that no one has tried yet, or one that tried and failed and is no longer visible.

The Meta-Point

The core insight is this: the sample of visible investment successes is heavily biased toward the survivors.

You cannot infer the true return distribution of a strategy by observing only the ones that worked.

You need the data on the ones that didn't. And that data is systematically missing.