// research
Survivorship bias in crypto backtests, measured on 30 coins
Backtest buy & hold from 2021 on today's coins and the median coin made +29.61%, the average +420.12%. On the 30 coins an investor could actually have picked on 2021-01-01 — six of them delisted since — the median coin returned -24.92% and the average +117.04%, less than bitcoin alone (+176.08%).
7 min read
A backtest that starts in the past but takes its list of coins from today has already seen the ending: the coins that collapsed and were delisted are missing, and the coins that became big are in. This is survivorship bias, one of the traps listed in what backtesting is. Here it is measured: the same buy & hold on two lists of coins over the same five and three-quarter years.
Two lists of coins
- The 2021 list — what you could have bought. The 30 Binance USDT-M perpetuals with the highest turnover in the 30 days before 2021-01-01, picked by the platform as of that date (the "Take the symbols as of a past date" option of a new backtest), delisted coins included. Six of the 30 have been delisted since: EOS, WAVES, YFII, MKR, SXP and OMG.
- Today's list — what a backtest built today sees. Every perpetual that already traded on 2021-01-01 and that the platform still keeps current in October 2026. There are exactly 30: the 24 survivors of the 2021 list plus SOL, DOGE, NEAR, AXS, ATOM and RUNE — coins an investor could have bought in 2021, but that were not among its 30 most traded.
Why 30 and not the usual top 20: the 2021 top 20 lost only one coin to delisting (EOS), too few to measure anything, and 30 is the most a point-in-time list on the platform takes. A wider net catches more dead coins: of the 78 USDT-M perpetuals that traded on 2021-01-01, the platform's symbol registry shows 22 delisted since — 6 of the top 30, 16 of the other 48.
The test: buy once with 95% of a 10,000 USDT account at the open of the second 4-hour bar of 2021 and hold until 2026-10-01, or until the last bar before a delisting. Perpetual futures at 1x, isolated margin, 0.05% fee and 0.05% slippage per side, funding switched off so the run follows the coin's price as a spot holder would see it — the delisted coins exist on the platform only as perpetuals (the callout below explains why funding is off). One strategy, nothing tuned.
The average return across a list is also what splitting the money equally across its coins on day one would have made, so the mean below reads as "an equal-weight basket, never rebalanced".
The result
| Coins | Median return | Average return | Coins that lost money | Lost 80% or more | |
|---|---|---|---|---|---|
| 2021 list, as of 2021-01-01 | 30 | -24.92% | +117.04% | 16 | 8 |
| 2021 list without its six delisted coins | 24 | +6.25% | +151.95% | 12 | 6 |
| Today's list | 30 | +29.61% | +420.12% | 14 | 6 |
The same strategy over the same window looks very different depending on the list. On the coins you could have picked, the typical coin lost a quarter of its value and the equal-weight basket made less than bitcoin alone. On today's list the typical coin was up, and the basket made more than three and a half times as much as on the 2021 list.
Backtest
Buy & hold 1x perp, funding off — survivorship studyBTCUSDT · 4h · Futures · 2021-01-01 → 2026-10-01
- Return
- +176.08%
- Max drawdown
- 75.42%
- Sharpe
- 0.60
- Win rate
- 100.0%
- Trades
- 1
Where the gap comes from
Dropping the dead coins moves the median from a loss to a gain. Removing the six delisted coins alone lifts the median from -24.92% to +6.25% and the average from +117.04% to +151.95%. Four of the six had lost money; together they had a median of -71.44%.
Refilling the list from today's winners does the rest. The six coins that today's list adds were chosen, in effect, by what happened after 2021: SOL made +6,971.27% and DOGE +1,725.20%. Without its single best coin, today's list still averages +194.22% — the 2021 list without its best coin, BNB, averages +58.03%.
Backtest
Buy & hold 1x perp, funding off — survivorship studySOLUSDT · 4h · Futures · 2021-01-01 → 2026-10-01
- Return
- +6,971.27%
- Max drawdown
- 96.57%
- Sharpe
- 1.20
- Win rate
- 100.0%
- Trades
- 1
Backtest
Buy & hold 1x perp, funding off — survivorship studyBNBUSDT · 4h · Futures · 2021-01-01 → 2026-10-01
- Return
- +1,828.21%
- Max drawdown
- 72.28%
- Sharpe
- 1.03
- Win rate
- 100.0%
- Trades
- 1
Backtest
Buy & hold 1x perp, funding off — survivorship studyDOGEUSDT · 4h · Futures · 2021-01-01 → 2026-10-01
- Return
- +1,725.20%
- Max drawdown
- 92.93%
- Sharpe
- 0.78
- Win rate
- 100.0%
- Trades
- 1
How a delisted coin's run ends
A backtest cannot hold a coin that no longer trades. On the platform the run of a delisted perpetual ends at the last bar of its archive, and the position is closed there at the market as end of data — at whatever price the coin had left. The six delisted runs ended between 2022-04-12 (YFII) and 2025-12-05 (SXP).
Delisted does not always mean worthless: MKR ended at +167.10% and YFII at +11.21%. A token can leave an exchange because it is swapped for a new one or simply because of thin trading, and the backtest records only the price on the way out, not what a holder did next.
Backtest
Buy & hold 1x perp, funding off — survivorship studyOMGUSDT · 4h · Futures · 2021-01-01 → 2026-10-01
- Return
- -84.19%
- Max drawdown
- 98.38%
- Sharpe
- 0.14
- Win rate
- 0.0%
- Trades
- 1
Backtest
Buy & hold 1x perp, funding off — survivorship studyWAVESUSDT · 4h · Futures · 2021-01-01 → 2026-10-01
- Return
- -74.60%
- Max drawdown
- 97.32%
- Sharpe
- 0.30
- Win rate
- 0.0%
- Trades
- 1
Backtest
Buy & hold 1x perp, funding off — survivorship studyMKRUSDT · 4h · Futures · 2021-01-01 → 2026-10-01
- Return
- +167.10%
- Max drawdown
- 91.12%
- Sharpe
- 0.70
- Win rate
- 100.0%
- Trades
- 1
Surviving is not the same as doing well
The worst run of both lists belongs to a survivor: SNX, still trading, -91.95%. Of the eight coins of the 2021 list that lost 80% or more, only two were delisted — FIL, GRT, THETA, DOT, COMP and SNX are still listed and are in today's list too. Every one of the 36 runs, bitcoin included, fell more than 70% from a peak at some point of the window.
Backtest
Buy & hold 1x perp, funding off — survivorship studySNXUSDT · 4h · Futures · 2021-01-01 → 2026-10-01
- Return
- -91.95%
- Max drawdown
- 97.95%
- Sharpe
- 0.07
- Win rate
- 0.0%
- Trades
- 1
All 36 coins
| Coin | 2021 list | Today's list | Return | Max drawdown |
|---|---|---|---|---|
| SOL | — | yes | +6,971.27% | 96.57% |
| BNB | yes | yes | +1,828.21% | 72.28% |
| DOGE | — | yes | +1,725.20% | 92.93% |
| TRX | yes | yes | +1,088.86% | 72.41% |
| XRP | yes | yes | +531.05% | 84.01% |
| NEAR | — | yes | +258.23% | 94.96% |
| ETH | yes | yes | +247.83% | 80.52% |
| AVAX | yes | yes | +243.04% | 95.67% |
| BTC | yes | yes | +176.08% | 75.42% |
| MKR | yes | delisted 2025-09-08 | +167.10% | 91.12% |
| AXS | — | yes | +108.66% | 99.50% |
| AAVE | yes | yes | +82.59% | 92.22% |
| XLM | yes | yes | +70.33% | 90.16% |
| UNI | yes | yes | +68.22% | 94.06% |
| ETC | yes | yes | +53.66% | 95.74% |
| ADA | yes | yes | +34.82% | 95.09% |
| LINK | yes | yes | +24.40% | 89.40% |
| YFII | yes | delisted 2022-04-12 | +11.21% | 73.76% |
| BCH | yes | yes | -11.90% | 93.35% |
| CRV | yes | yes | -37.94% | 96.75% |
| RUNE | — | yes | -38.82% | 98.14% |
| LTC | yes | yes | -46.52% | 88.73% |
| VET | yes | yes | -51.46% | 97.96% |
| ALGO | yes | yes | -60.76% | 96.11% |
| ATOM | — | yes | -67.77% | 96.54% |
| EOS | yes | delisted 2025-05-21 | -68.28% | 96.28% |
| WAVES | yes | delisted 2024-06-11 | -74.60% | 97.32% |
| EGLD | yes | yes | -77.29% | 99.29% |
| COMP | yes | yes | -80.09% | 97.42% |
| DOT | yes | yes | -82.07% | 97.80% |
| THETA | yes | yes | -83.99% | 98.47% |
| OMG | yes | delisted 2025-01-31 | -84.19% | 98.38% |
| SXP | yes | delisted 2025-12-05 | -86.89% | 98.50% |
| GRT | yes | yes | -87.41% | 98.82% |
| FIL | yes | yes | -90.95% | 99.19% |
| SNX | yes | yes | -91.95% | 97.95% |
Linked returns are saved runs. The others come from the same strategy re-run on the platform without saving it, with the same settings — the eight saved runs match those re-runs to the cent. A delisted coin's date is its last bar. Medians, averages and counts in this article are computed from these 36 returns.
Conclusion
- On a list of coins taken from today, buy & hold since 2021 shows a median of +29.61% and an equal-weight average of +420.12%. On the list that existed on 2021-01-01 the same test shows -24.92% and +117.04%.
- More than half of the gap in the median comes from leaving out the six coins that were delisted; most of the gap in the average comes from adding the coins that only hindsight puts in the list, SOL and DOGE above all.
- Survivors fail too: the worst run of the study, SNX at -91.95%, is a coin that still trades.
To test without the bias, build the list as of the backtest's start date — the "Take the symbols as of a past date" option does it — as in the golden cross study on the top 20 of 2021. More on putting a test together: how to backtest a crypto strategy.
Test your own idea
Create a free account, start from a template or from your own rules, and read your first report.
More articles
- Does high funding predict crypto crashes? 8 coins, 2020–2026BIS economists found that high crypto carry predicts crashes. On 8 Binance perpetuals, 2020–2026, crashes came more often — but after bigger gains.
- Crypto momentum after the paper: a weekly test on 8 coinsLiu and Tsyvinski found weekly momentum in bitcoin up to 2018. On 8 coins in 2020–2026 the one-week signal lost to holding; four weeks cut drawdowns.
- Does price move the hour before funding? 8 coins, 2020–2026Do traders sell before paying funding? The hour before each settlement on 8 Binance perpetuals, 2020–2026: no reliable move, and fees took what was left.
- Bitcoin after the halving: 2020 and 2024 backtestedDid buying BTC at the halving pay? A year after 2020: +554.04%, top 2.5% of start days. After 2024: +30.45%, below the median. Two cases, no pattern.
- Weekends and time of day: when does bitcoin move?Do bitcoin gains come on weekends or outside US trading hours? Hourly BTC and ETH data since 2020, and session backtests: with fees, -99.87%.