Survivorship bias

Survivorship bias: the short answer

Survivorship bias is the error of drawing conclusions from only the things that survived — funds still open, stocks still listed — while silently ignoring the ones that failed, closed, or were delisted. Because losers drop out of the dataset, the survivors that remain look far more successful than the full population ever was.

Where the name comes from

The best-known illustration comes from World War II. Military analysts studying bombers that came home wanted to bolt extra armor onto the spots that showed the most bullet holes. The statistician Abraham Wald spotted the flaw in their reasoning: the only planes they could examine were the ones that made it back. A hit to the areas that looked clean on survivors was exactly the kind of hit that stopped a plane from returning at all. The armor belonged where the survivors had no holes.

The same trap shows up everywhere in markets. When the failures have already been pulled out of view, you are measuring the survivors rather than the whole population — and your conclusions quietly inherit a systematic optimism you never asked for.

How it inflates fund and index returns

Mutual fund databases are a textbook case. Funds that perform poorly tend to get liquidated or folded into other funds, and their track records often disappear along with them. Average the returns of only the funds that still exist today and you are averaging a group pre-selected for having survived — which usually meant performing well enough to stay open in the first place.

Researchers have tried to size the effect. Studies of U.S. mutual funds have generally found that survivorship bias flatters average annual returns by roughly a percentage point or so per year — modest in a single year, but meaningful once it compounds across a decade of a supposed track record.

Stock indices carry the same problem. An index constantly drops companies that go bankrupt, get delisted, or shrink into irrelevance, and swaps in healthier ones. Study today's index membership as if it always looked that way and you are studying a roster curated, in hindsight, to leave out the dead.

Why it quietly wrecks backtests

This is where survivorship bias does the most damage to everyday traders. Suppose you backtest a strategy on "the S&P 500" using only the roughly 500 companies that sit in the index today. Every one of those names is, by construction, a business that grew, stayed listed, and earned its place. The Enrons, Lehmans, Blockbusters, and hundreds of quieter delistings never make it into your test — even though a real trader living through those years would have held some of them and worn the losses.

The failures are the part that matters most, because they are the losses your risk controls were supposed to handle. Strip them out and you have removed the hardest question from the exam. The strategy did not get better — the test got easier.

Everyday examples beyond markets

• "Successful founders dropped out of college." You hear about the dropouts who became billionaires; you never hear about the far larger crowd who dropped out and simply didn't make it. The winners are loud, the failures invisible.
• "Old buildings were built better." The flimsy old buildings already fell down. Only the well-built ones lasted long enough to be admired.
• "This fund family has a great lineup." The weak funds were quietly shuttered, and the brochure shows you the ones that lived.

The mistake is the same every time: judging a process by its survivors while the failures sit outside the frame.

How to guard against it

• Use a survivorship-free dataset. A point-in-time universe includes every name that was genuinely tradeable on each historical date — delisted and failed companies and all.
• Ask what's missing, not just what's shown. When you see a track record, the first honest question is: who isn't in this sample, and why did they drop out?
• Model realistic fills and costs. A backtest that assumes perfect entries on a clean roster of survivors flatters the result twice over.
• Distrust unusually smooth results. Real strategies lose sometimes. A curve with no scars often means the scarred cases were quietly removed.

How Coil reads it

Survivorship bias is the single easiest way to make a trading backtest lie, so Coil treats avoiding it as a matter of honesty rather than a technicality. Coil's research runs against a survivorship-free universe — delisted and failed names are included at each point in time, so the test faces the same losers a real trader would have owned, not a hindsight roster of winners. None of this is investment advice; it is educational background on how the research is built.

That discipline is exactly why Coil's published numbers sit below the too-good figures you see advertised elsewhere — and why they hold up. For context, the leadership-rotation backbone backtested +638% for 2017 to 2026 H1 versus SPY's +282% (survivorship-free, delisted names included, next-open fills, costs modeled), with a worst drawdown of -23% versus SPY's -32%, and it was positive in 9 of 10 years. The honest rider matters: through end-2025 it ran roughly even with SPY at about one-third less drawdown, and the outperformance concentrates in leadership regimes. These are research backtests, not live or client returns, and the engine is newly live. If you want the methodology spelled out rather than just claimed, the how it works page walks through it.

People also ask

What is survivorship bias in simple terms?

It is the mistake of judging a group by only the members that lasted, while ignoring the ones that failed and dropped out of view. Because the failures are missing, the survivors look far more successful than the whole group ever was.

How does survivorship bias affect backtesting?

If you test a strategy only on stocks still listed today, you exclude every company that went bankrupt or was delisted — exactly the losses a real trader would have taken. The result is an inflated, unrealistically smooth return the same rules would never produce on the true historical universe.

How much does survivorship bias inflate mutual fund returns?

Academic studies of U.S. mutual funds generally estimate the bias flatters average annual returns by roughly a percentage point or so per year, because liquidated and merged-away funds vanish from the databases. Compounded across a decade, even that modest gap adds up.

What is the airplane example of survivorship bias?

In WWII, analysts wanted to armor bombers where returning planes showed the most bullet holes. Statistician Abraham Wald noted they only saw planes that survived — the armor actually belonged where survivors had no holes, since planes hit there never came back.

How do you avoid survivorship bias?

Use a point-in-time, survivorship-free dataset that includes delisted and failed names, model realistic fills and trading costs, and always ask which cases are missing from any track record you are shown. Be suspicious of results that look too clean.

Related terms

Maximum drawdown · Market leadership · Momentum investing · Trend following · full glossary →