A strategy claims to have "beaten BTC." The conclusion itself may be true, but whether the comparison is fair depends on a few conditions that are often left out. The three most common sources of distortion are: not deducting the strategy's trading costs, comparing a carefully selected version of the strategy's performance against a simple buy-and-hold BTC position, and using a sample period that happens to favor that particular strategy. If a comparison does not address these three points, it is describing a piece of history rather than proving a lasting edge.
A leading global cryptocurrency platform,suitable for both beginners and experienced traders.
New user benefit: 20% off trading fees upon registration!！
Cost basis: the more frequently a strategy trades, the more unfair the comparison becomes
Buying and holding BTC has extremely low trading costs: you buy once, may not sell for years, and slippage and fees occur only once or twice. Most active strategies, however, enter and exit repeatedly. Every trade generates fees and slippage, and derivatives strategies may also involve funding rates. A backtest experiment by CoinPaprika provides a reproducible example: a strategy had a pre-cost edge of +0.23 percentage points. After adding a 1% round-trip cost, the edge turned into -0.77 percentage points—the result flipped completely. The same experiment also found that delaying execution by one candle, changing from an idealized fill to opening at the next candle, consumed 0.295 percentage points, nearly one third of the strategy's edge.
A fair comparison needs to include every trade cost in the strategy's net return, then compare that net return against BTC's holding return using the same net-return standard. If the BTC side only counts price appreciation while the strategy side uses gross returns or returns after only some costs are deducted, the two sides are not measuring the same thing. Static cost assumptions common in academic backtests, such as a fixed 0.1% slippage plus 0.1% fee, can underestimate real costs during volatile or illiquid periods. But that is still a relatively honest approach. The more common problem is that strategy descriptions simply never mention costs at all.
Benchmark selection: beating "which BTC" counts
"Beating BTC" requires specifying which time period of BTC is being used. If a strategy's backtest period happens to cover a crypto bull market such as 2017–2021, BTC's unconditional 30-day return was 17.92%. Over 2022–2026, the same figure was only 2.40%, a difference of about 7.5 times. A strategy that "beat BTC" on 2017–2021 data may simply have participated in that rally. It would not necessarily do the same after 2022. CoinPaprika tested 88 RSI parameter configurations and found that 78 of them were profitable during the tuning period, but only 2 still had a positive edge after subtracting the benchmark return—the part you would have earned by doing nothing. This shows how large the gap can be between "profitable" and "beating the benchmark."
The fair approach is to use exactly the same start and end dates for the strategy and BTC, and to report the strategy's excess return relative to BTC over that same period, rather than only reporting the strategy's absolute return. If the backtest period spans multiple bull and bear phases, results should also be shown separately for each phase, not just as a single full-period total.
Out-of-sample validation: only unseen data counts as evidence
Strategy parameters are tuned on historical data. The tuning process itself makes the strategy adapt to that history, including its noise. If a strategy claims to beat BTC but only shows results on the same data used for tuning, the result has limited persuasive power. CoinPaprika's summary is blunt: after careful testing, most results are negative. The common failure pattern is not that the edge decayed, but that the edge never existed in the first place and was simply hidden by a rising market.
A stricter test is out-of-sample validation: running the same set of rules on data that was never touched during strategy development to see whether the excess return still exists. If a strategy only presents in-sample results, or if the out-of-sample window is very short, such as only a few months, that is not enough to support the claim of "beating BTC." A short out-of-sample result may just be luck, especially in the highly volatile crypto market.
Survivorship bias: comparing only the survivors is incomplete
If a strategy involves multiple coins or multiple parameter configurations, showing only the best performers creates survivorship bias. CoinPaprika data shows that among the 61,463 coins it tracks, 78.6% are no longer active. If a strategy is backtested only on a pool of coins that are still alive today, its performance will be overstated because the coins that went to zero were not included in the simulation of trading costs, slippage, and failed trades. BTC as a benchmark does not have this problem: it has always existed and remains highly liquid. If the strategy side involves smaller coins, the description should explain whether the coin pool excluded delisted or worthless assets.
A checklist you can verify yourself
When you see a comparison claiming that a strategy beat BTC, check these items in order: whether the strategy's net return deducts fees and slippage for every trade; whether BTC and the strategy use exactly the same start and end dates; whether the strategy parameters were tuned on the same data used to present results; if there is out-of-sample testing, how long the window is; and whether the strategy's coin pool includes assets that have already disappeared. If any of these five items is not disclosed, the reliability of the conclusion weakens. If the strategy's edge after costs is less than one percentage point, or if it only exists during one particular market regime, then a more reasonable interpretation of "beating BTC" is this: the strategy happened to be on the right side during that period of history, but whether it deserves real money requires a different set of evidence.
A leading global cryptocurrency platform,suitable for both beginners and experienced traders.
New user benefit: 20% off trading fees upon registration!！
References
- CoinPaprika·How to Backtest a Crypto Strategy: Data, Pitfalls & Overfitting, 2026-10-29
- Zenodo·BOS Alpha Estructural Whitepaper, 2026-10-29
- Gate·Why do some trading strategies make money in backtests but fail to cover trading costs in live trading?, 2026-10-29
- ScienceDirect·AutoQuant: An auditable expert-system framework for execution-constrained auto-tuning in cryptocurrency perpetual futures, 2026-10-29
