What's a good max drawdown? There isn't a good number in the abstract.
A max drawdown is the worst peak-to-trough fall a strategy put you through — the deepest hole between a high and the low that followed. Everyone wants a target: is under 20% good? Under 30%? But the question has no honest answer on its own, because a drawdown isn't a cost you pay for nothing. It's the price of a return. The only honest way to read one is against what it bought — and whether you could have sat through it.
Put in any loss and see the gain it takes to climb back to even — free, no sign-up. A 50% fall needs a 100% gain, not 50%.
Why there's no good number
Take two strategies — an illustrative example, not our data. One went through a −40% drawdown and compounded to +300% over five years. The other never fell more than −15% and finished at −5%. Which had the “better” drawdown? The −40% one, easily — it was the price of a real return, while the shallow one was pure downside with nothing to show. A drawdown number in isolation tells you almost nothing. Pair it with the return it bought and it starts to mean something; ask whether you could have held through it and it means everything.
So “is X% a good drawdown?” is the wrong question. The honest ones are: what did this drawdown pay for, and could you have sat through it without selling? Those depend on the return and on you — not on a threshold anyone can hand you.
Our own drawdown reality
Here is the whole glass, not a rule of thumb. Across the 1096 honest backtests on this site with enough trades and history to judge — every strategy on every famous asset, costs on, no hindsight — the typical worst drop a run went through was −33%. Deep holes are the norm, not the exception: 77% fell more than 20%, 56% more than 30%, and 28% more than half their value, out to an extreme of about −98%. These are the simulated strategies' drawdowns over the window — not a promise about the future.
These drawdowns come from famous names that still trade today. Names that delisted, went bankrupt or (for crypto) died aren't in the set — and those include the deepest holes of all — so the full real-world range of drawdowns is wider than the slice shown here. And these are the strategies' drawdowns, not the assets': an exit-based rule can sidestep part of a fall a buy-and-holder would have sat through.
Palantir went through a −41% drawdown and still finished at +1178% over the window. Here the hole was the price of a real return — if you could have held through it.
Avalanche went through a −98% drawdown and finished at only -97% — deep pain, with little or nothing to show for it.
Two deep drawdowns, opposite outcomes. That is the whole answer to “what's a good max drawdown” in one picture: the size of the hole alone never told you which was worth living through — it depends entirely on the return it paid for.
And a deep drawdown costs more to undo than it looks
A loss and the gain needed to erase it aren't symmetric — you're compounding back off a smaller base, so the deeper the hole the more brutally the recovery turns against you. This is the gain required to get back to even, not one that's promised:
The typical −33% above needs about +49% just to get back to where it started.
Why a good drawdown is finally about you
Everything up to here has been about the return a drawdown bought. The last piece is whether you could actually hold on to collect it. A hole you sell out of pays you nothing, however handsome the number waiting at the far side — and the deep ones in the distribution above played out over months or years of real time, long enough that most people quit near the bottom. So when you read a backtest, don't ask whether its worst drawdown clears some threshold. Ask the honest question: could you have lived through that stretch, in real money, without selling? If the answer is no, the return it earned was never really available to you — and that, not a number, is what makes a drawdown “good” or not for you.
The recovery figures are an arithmetic identity — gain to recover = drawdown ÷ (1 − drawdown) — true of any market. Any live figures are computed from simulated backtests over roughly the last five years and update as new bars close. 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.