Backtest Crypto Candle by Candle

Seven pairs is not seven samples

This is the mistake that quietly ruins most crypto backtests. Altcoins are, most of the time, leveraged expressions of Bitcoin. When BTC runs, SOL, AVAX and ADA usually run harder; when it drops, they drop harder. So a setup that produced thirty winners across seven pairs may have produced far fewer independent results than the number suggests.

The damage is not to your win rate, it is to your drawdown estimate. Correlated positions lose on the same day, and a backtest that counted them separately never showed you that day at full size. If you intend to trade several pairs at once, test the basket as one position, not seven.

The honest way to use seven pairs is as an out-of-sample check, not a sample multiplier: build the rules on one, then see whether they survive untouched on another. That is a different and much harder question — and it is the one that actually predicts live performance.

There is no daily close, so your daily candle is a decision

Crypto never closes. The daily candle you are reading is a UTC convention, not a session — nothing meaningful happens at midnight UTC, no auction sets a price, no desk squares up. A daily-bias rule imported from indices or forex is quietly assuming a structure that does not exist here.

That does not make higher-timeframe levels useless, it makes the boundary arbitrary — so test whether your setup still works if the day is cut somewhere else. If your edge disappears when you shift the daily open by a few hours, it was in the convention rather than the market.

Weekends are thin, and thin is where the wicks live

Traditional markets close and reopen with a gap. Crypto stays open into a much thinner book, and thin books produce the long wicks that spear resting stops before price returns to where it was. If your backtest quietly skips weekends because that is when you are not at the desk, you have removed exactly the conditions your stop is most likely to meet.

Replay them anyway and log them separately. Two useful outcomes: either your setup is fine across weekends, or you have found a genuine filter — and a filter you discovered by measuring is worth far more than one you inherited from a video.

The pairs are not interchangeable in volatility

A fixed percentage stop behaves completely differently on BTCUSD than on DOGEUSD or AVAXUSD. What is a normal hour of noise on a high-beta alt is a significant move on Bitcoin, so a stop that is sensible on one is either constantly clipped or absurdly wide on another.

Size the buffer to the instrument's own noise rather than carrying one number across all seven, and let the position size calculator keep the risk in money constant while the stop distance changes. For Bitcoin specifically, the deeper walkthrough is on the BTCUSD page.

How to backtest crypto in CRTLAB

Pick a pair, jump to a date, and step forward one candle at a time with the future hidden. Mark entry, stop and target as you would live — the trade is logged and scored automatically, including the ones a weekend sweep stopped you out of, which are the trades most people quietly leave out.

Free gives you all seven pairs on the most recent two weeks, no card. Pro opens the full history: BTC, ETH, SOL, DOGE, ADA and AVAX from mid-2022, and XRP from May 2023 — enough to cover a bear market, a full recovery and the long flat stretches in between, which is where most setups actually fail.