Part 2: the test
Time-series momentum: we tested 468 versions
Posted 2026-10-31
Time-series momentum: 468 versions tested on 16 crypto perpetual futures, 2 stock indices and 3 commodities over 2.7 years, fees included. 53% made money.
Watch it on TikTok · Instagram · YouTube
Full method, every setting and every trade →
The other part: What is momentum trading? (part 1)
What the video says
Time-series momentum. 53% of the 468 versions we tested made money. Too good to be true? And the best one hides a trap.
Yesterday: the rule. Today: the proof. Before you risk a cent on it, here's how it really did. We tested it on 16 crypto perpetual futures, 2 stock indices and 3 commodities over 2.7 years, fees included.
Quick reminder of the rule. Go long after a coin rose over the last weeks, short after it fell.
A version is the same rule with different settings, like where the stop or the target goes. Here is the best of all 468 versions. 353 trades, 20% winners. All markets together. Each square is one trade: green won, red lost.
R is what you risk on one trade: minus 1 R means you lost your stake once. We split the test in two periods. From November 2022 to March 2024: plus 0.48 R per trade. From March 2024 to July 2025: plus 0.23. Skill, or luck?
Now kind by kind, the typical version. Crypto perpetual futures: plus 0.01 R per trade. Stock indices: minus 0.04. Commodities: minus 0.09. Versions that made money: 58% on crypto perpetual futures, 20% on stock indices, 2% on commodities.
Against simply holding the market over the same days, it adds plus 0.02 R per trade, but that could still be luck. Then we ran that exact version on older crypto data it had never seen, 2020 to 2022. It still beat the market by plus 0.16 R per trade, but not reliably. On a full year it had never seen, it made minus 0.52 R per trade. Chance alone could explain it.
The trap here is luck. A small edge over a few hundred trades can appear by chance alone.
The answer. No reliable edge.
What this means for you: ask how many trades it took, and whether chance alone could explain it. Which strategy should we test next? Write it in the comments.
Every number and the full method are on backtestd dot com. Follow for the next test.
Follow the tests: TikTok · Instagram · YouTube · Contact
Backtest results, not financial advice.