What's a good Sharpe ratio? And why a very high one is a red flag.
The Sharpe ratio is return above cash divided by how much the returns swung around — reward per unit of risk. Higher looks better, and up to a point it is. But it's the one number people most often use to talk themselves into an overfit strategy, so the honest guide has two halves: the rough bands, and the reason a sky-high figure on a backtest should make you more suspicious, not less.
Put in an annual return, the risk-free rate and the volatility and get the Sharpe in one step — free, no sign-up.
The rough bands
Treat these as a sense of scale, not hard cut-offs — a Sharpe depends on the period, the asset and how it was measured. Historical ranges, not a promise about the future.
The return didn't even clear the risk-free rate — you took risk and got paid less than a savings account.
Where a lot of honest, long-run investing actually lives. A broad stock index has historically sat under ~0.5 over full cycles.
A real, respectable risk-adjusted return — and hard to sustain over a long, out-of-sample stretch.
Elite territory for a live track record. On a backtest, start asking how many variants were tried before this one.
On a backtest this is almost always overfitting, a look-ahead bug, a too-short sample, or leverage — not a durable edge. The higher it climbs, the less you should trust it.
Why a very high backtest Sharpe is a warning sign
A backtest Sharpe isn't the number you'll live — it's the number you found, and three things inflate it:
- • The variants you tried. Test one rule and read its Sharpe honestly. Test a hundred and keep the best, and the winner's Sharpe is lifted by pure luck — the more you searched, the higher the bar a real edge has to clear. This is the big one, and it's invisible in the single number.
- • A short sample. Over a handful of trades or a couple of years a flattering Sharpe is mostly noise; it needs enough independent trades before it means anything.
- • Fat tails and leverage. Sharpe assumes tidy, bell-curve returns and counts upside swings as “risk” too. A strategy that sells options or runs leverage can post a lovely Sharpe right up until the one day it doesn't.
That's why this site reports a deflated Sharpe: it discounts the raw figure for how many variants were in play and how noisy the sample is, so what you see is closer to the number that survives out-of-sample. A “good” Sharpe is one that stays good after that haircut — and rests on enough trades to trust. Chase a raw Sharpe of 4 and you're usually just optimising toward the luckiest overfit.
Compute one from figures you type — or batch-scan a strategy across many famous assets and get a deflated Sharpe on the winner: the figure discounted for how many you tried (deflation needs to know that count, so it lives in the scan, not a single run), so what survives is closer to what you'd actually live — not the flattering raw one.
The bands above are rough historical rules of thumb, not thresholds to trade on — a Sharpe ratio depends entirely on the period, asset and method used to compute it. Hypothetical / simulated backtest results, where mentioned, are 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.