The indicator everyone recommends
Supertrend draws a line under price in an uptrend and above it in a downtrend, flipping when price crosses. It looks clean on a chart, it’s built into TradingView, and “best Supertrend settings” videos rack up millions of views. Almost none of them show a full backtest with costs. We did.
What we tested
The same Supertrend entry rules, run over our full backtest window across the top 50 coins by market cap, with fees and slippage applied to every trade (the exact window is shown on the settings page). Then we swept the exit grid: 5 stop-loss levels × 6 take-profit levels = 30 full backtests, each settling between roughly 3,900 and 5,600 trades — sample sizes large enough that luck is not the explanation.
The complete grid is published as a live page that refreshes with our data pipeline: Supertrend — best stop-loss & take-profit settings. We publish it whether it flatters the strategy or not.
The result: 0 profitable configurations out of 30
As of July 2026, every one of the 30 combinations lost money. The best configuration — a tight 4% take-profit against a wide 12% stop-loss — still lost about 40% over the window. The worst lost about 50%. There was no clever setting hiding in the grid.
The detail worth staring at: that best cell had a 56.8% win rate. It won more often than it lost — and still bled out. How?
Why a 57% win rate still loses
Look at the shape of that “best” configuration: take profit at 4%, stop loss at 12%. Each win banks a small gain; each loss gives back three times as much. Win 57 trades out of 100 at +4%, lose 43 at −12%, and you are deeply negative before costs — and with four to five thousand trades in the window, fees and slippage grind the result down further on every flip.
This is the standard indicator-strategy trap: the configurations that maximize win rate do it by cutting winners short and letting losers run — the exact opposite of what makes a trading system survive. A high win rate is a marketing number, not an edge.
Is it the settings? No — we checked all of them
The usual defense is “you used the wrong settings.” That defense doesn’t survive a grid sweep. We tested every stop-loss/take-profit combination our engine supports, and the entire surface is negative. If a strategy only works in the one cell you haven’t tested yet, it doesn’t work.
This matches what we found across our broader strategy research: simple directional signals on crypto OHLCV data — Supertrend included — have not survived honest out-of-sample testing with costs.
The honest takeaway
Supertrend is a readable way to describe what price already did. As a standalone crypto trading strategy with fixed exits, it lost money in every configuration we measured over this window. If someone sells you “the best Supertrend settings,” ask for the full grid, the costs, and the losing cells — or run it yourself in the free simulator and check the live numbers on the settings grid, which update automatically as new data comes in.
So what does work?
Same answer we give everywhere on this site: after testing strategy families across our research campaigns, what survived wasn’t a better entry signal — it was risk control: position sizing, defined risk, and surviving drawdowns. That’s less exciting than a magic indicator. It’s also the only thing our own data supports.