Six ideas that sounded profitable and failed our test
We have tested more than 40 ideas on five years of crypto data. Most of them sounded better than what we run today. Most of them were worse. Publishing the failures is part of the point: a system is defined as much by what it refuses to do as by what it does.
The test every idea has to pass
- The whole account, not one trade. An idea is replayed day by day with real fees, compounding and the drawdown brake switched on, the way it would actually run.
- Both halves of history. It must beat the current rules in 2022–mid-2024 and mid-2024–2026. Winning in only one half usually means luck.
- The worst drop stays inside 15%. More profit with a deeper hole is a fail, not a trade-off.
- Built twice. Promising results are rebuilt separately from scratch. If the two versions disagree, we don't trust either.
Six that failed
| The idea | Why it sounds good | What happened |
|---|---|---|
| Let winners run, no fixed exit | "Never cap your upside" | Returns fell by more than two-thirds. Most big moves give a lot back before you sell. |
| Trade more often | More chances to win | Each trade earned less and the account fell deeper. Faster signals catch more spikes that fade. |
| Bet against falling prices | Make money both ways | Lost money on average, per trade. Falling coins bounced too often. |
| Trailing stop-losses | Lock in gains, cut losses | Deeper drops, not shallower. Normal swings kept triggering sales near the bottom. |
| Add riskier, hype-driven coins | Bigger moves, bigger wins | The worst drop went past 15%. Fail. |
| 5× leverage "for more profit" | Same idea, five times the result | Wiped out. Twice. |
What this taught us
The ideas that fail have a pattern: they trade a small, frequent gain for a rare, large loss. That trade looks great for months, which is exactly why it's dangerous. A backtest that only reports the average return will happily recommend it. One that reports the worst drop first won't.
A caveat we always state: our coin history only includes coins that still trade today, so absolute backtest returns are optimistic. Comparisons between ideas, run on the same data, still hold, and that is how we use them.
We'll keep publishing what we test, including the ideas that make it through.
Risk notice: this article is for information only. It is not financial, investment or tax advice, and nothing here is an offer or solicitation. Cryptoassets are highly volatile and you can lose all of your money. Backtested figures are hypothetical and optimistic; live figures cover a short period. Past results never promise future ones.