We backtested RSI honestly — it beat buy & hold on only 8 of 47 famous assets
I ran the classic RSI mean-reversion strategy — buy when the 14-period RSI drops into oversold territory (below 30), step back out once it recovers past the midline (above 55) — the same honest way as everything else on this site: fees and slippage charged on every fill, only ever acting on a bar that has closed, always measured against simply buying and holding. No cherry-picked chart, no favourable window. Here is the whole glass — every famous asset, every winner and every loser.
So — does RSI work?
On the honest evidence: mostly, no. Across the 47 famous assets with enough trades and history to judge, the RSI rule beat buy & hold on 8 of them — 17%. That's below the 25% we measure across every strategy on every asset — so RSI does worse than a rule picked at random, not better. The typical RSI run didn't just lose narrowly either: its median result came in at -74 pts versus buy & hold over the window. A handful of real winners exist — that's exactly why the question is worth asking honestly — but the base rate says the burden of proof is on RSI, and “it looked great on one chart” is not that proof.
These are famous names that still trade today. Names that delisted, went bankrupt or were wound down aren't in the set (survivorship bias) — and that history isn't available from the free public data this tool runs on — so this rate isn't a representative base rate for every stock or ETF that has ever traded.
Where it shone, where it died
The most useful thing about the whole glass isn't the average — it's the spread. The exact same RSI rule, unchanged, swings from a big beat to a catastrophe depending purely on which asset you point it at.
Cardano — RSI beat buy & hold by +174 pts here. Where the timing genuinely added value it was usually by sitting out a decline the buy-and-holder rode down, then buying back after the drop — as much a statement about that asset's fall as about RSI's skill.
Nvidia — RSI came in -919 pts behind buy & hold. On a strong, sustained trend the rule sells the winner early and sits out the run — the single worst thing you can do to a trending asset.
That inconsistency is the honest answer to “does RSI work”: it isn't a property of the indicator, it's a property of the asset and the regime. RSI can help on something range-bound and hurt badly on something that trends — and you can't know which future you're in from the past.
Why RSI looks like it always works
- • You get shown the one chart where it nailed the tops and bottoms. Picking the asset and the window after you know RSI worked there is hindsight, not evidence. The whole glass above is what you get when you can't choose.
- • The costs are usually hidden. RSI trades a lot. Fees and slippage on every entry and exit quietly turn many “profitable” RSI rules negative — the friction was always real; most demos just leave it out.
- • Oversold can stay oversold. RSI's core assumption — that a stretched move reverts — simply fails in a strong trend, which is precisely when buy & hold does best. That's the same failure showing up on every worst-case asset.
Every RSI backtest is public — winners and losers, no cherry-picking. Read the whole RSI scoreboard, then run RSI on the asset you care about and see honestly which side of the line it lands on.
Figures are computed live from simulated backtests over roughly the last five years and update as new bars close. “RSI” here is the standard 14-period oscillator, buying an oversold reading (below 30) and exiting on the recovery back through the midline (above 55); other settings behave differently, which is the point — you can test any of them yourself. Hypothetical / simulated results — past performance is not a reliable indicator of future results. This is an educational, analytical tool, not investment advice and not a recommendation to buy or sell anything. You make all decisions and execute on your own broker.