Supporting data
SuperTrend Parameter Backtest — Complete Results Grid
SuperTrend is an ATR-based trend indicator — it draws a line above or below price using an ATR multiplier, and flips sides when price crosses it, giving a long/short signal. The question most people land here with is simple: which ATR period and multiplier actually work? This page does not guess at an answer. It is the full output behind the summary tables in the SuperTrend indicator settings guide — every setting we tested, measured, not argued for. Every one of the 2,160 rows is here — 1,080 with transaction costs applied and 1,080 at zero cost. Nothing is filtered out: losing settings, wiped-out settings and winners all appear side by side. Sort any column by clicking its header, and narrow the grid with the filters.
Why this page exists
Most "best SuperTrend settings" pages give you one number and no proof. We ran the opposite of that: every ATR period and multiplier combination we had a reason to test, on real price data, with real trading costs included, and we are publishing the whole grid — the losers next to the winners — so you can check our claims instead of trusting them. That is the point of a backtest table like this one: it replaces an opinion about "the best setting" with a measurement you can re-run the filters on yourself.
What is in this grid
ATR period ∈ {7,10,14,20,30} × multiplier ∈ {1.0..5.0 step 0.5} (45 combos) × timeframes 1m/3m/5m/15m/30m/1h/4h/1d × 3 instruments × 2 cost scenarios. Plain SuperTrend trend-following: long/short, flip on signal, always in market, no pyramiding. Fills at the next bar's open after the flip bar's close confirms the signal, so there is no lookahead. Engine: vectorbt Portfolio.from_signals with full equity reinvested each trade. The full methodology, the cost assumptions and the per-instrument winners are in the guide.
⚠️ Wiped-out rows are flagged — do not read their drawdown
391 of the 1,080 real-cost combos (36%) lose the entire notional under full reinvestment — total return at -100%. They cluster on 1m/3m/5m, where trade count is highest and per-trade cost compounds fastest. A further 14 zero-cost rows hit the same wall.
When equity reaches or crosses zero mid-run, vectorbt computes drawdown % against a near-zero equity base, so the figure explodes to a nonsensical magnitude — the worst row in this grid reads -1.6×1014%. Max DD % on a wiped-out row is a divide-by-near-zero artifact, not a real number. Total return % (floored at -100%) is the trustworthy figure on those rows. Every one of them carries a WIPED badge and its drawdown cell is marked, so sorting by Max DD % cannot mislead you.
⚠️ Zero cost is a diagnostic baseline, not an achievable result
Nobody trades at zero cost — there is no broker, exchange or venue where the round trip is free, so no zero-cost row here is reachable. The scenario exists only to separate how much of a setting's result came from the signal and how much was handed over in transaction cost. Read the real-cost rows for what a setting did; read the zero-cost rows only to size the cost drag. A zero-cost number is never a projection and never what a cheaper broker would get you.
Data sources
- Gold and Silver: Dukascopy XAUUSD and XAGUSD 1-minute data, 2020-01-01 to 2026-10-20 (~6.8 years).
- BTC: Delta Exchange BTCUSD 1-minute data from 2023-12-29 — Delta has no earlier history, so every BTC row covers a ~2.8-year window, not the ~6.8 years Gold and Silver get. BTC rows are not directly comparable with the other two.
Supertrend indicator best settings — is 10, 3 actually the best?
"10, 3" — ATR period 10, multiplier 3.0 — is the most commonly quoted SuperTrend setting online, and it is one of the 45 combos in this grid, so we can check it directly instead of repeating the claim. Filter the table below to period = 10, mult = 3.0, real cost and this is what comes back: on Silver's 4-hour chart it returns about +215% over the full window, and on Gold's 4-hour chart about +92% — both real, both profitable. But the actual best setting for Silver on that same 4-hour chart is period 30, multiplier 3.5, which returns roughly +1,208% over the same data — more than five times what 10, 3 does. And "10, 3" is not a universal setting: drop it onto a 1-minute, 3-minute or 5-minute chart on any of the three instruments and it is wiped out (-100% total return) every time, because ATR period 10 on a fast chart generates far too many whipsaw flips for the trading cost to survive.
So the honest answer: 10, 3 is a real, working setting on slow charts (1h, 4h, 1d) for Gold and Silver — it is not a myth. But it is not the best setting in this grid on any instrument or timeframe, and on fast intraday charts it is one of the settings that fails hardest. "Best settings" depends on which instrument and which timeframe you mean — there is no single period/multiplier pair that wins everywhere. Search supertrend indicator settings or best supertrend settings and most pages stop at a recommendation; this grid lets you see the actual number behind each one and the number behind every alternative.
How to read this table
- Each row is one fully independent backtest run: one instrument, one timeframe, one ATR period, one multiplier, one cost scenario (real cost or zero cost).
- Use the Instrument and Timeframe filters first — settings that win on Silver's 4-hour chart can lose money on BTC's 5-minute chart, so never compare rows across instruments or timeframes as if they were the same test.
- Always read the Real cost rows when judging whether a setting is usable. The Zero cost rows exist only to show how much of the result was eaten by trading costs — see the warning below.
- Check the WIPED badge before trusting any other number on that row — once a setting is wiped out, its Max DD % cell is a math artifact, not a real drawdown (explained below).
- Sort by Total return % or Sharpe to find a timeframe's best setting; sort by Trades to see which settings generate so many flips that cost alone kills them.
- Use the search box for a quick combo lookup, e.g. type "Gold 4h 30" to jump straight to every Gold, 4-hour row using ATR period 30.
The full grid
Loading grid…
| Instrument | Timeframe | ATR period | Mult | Cost scenario | Trades | Win % | Total profit | Total return % | CAGR % | Max DD % | PF | Sharpe | Expectancy | Avg $/day |
|---|
Legend. WIPED = total return at -100%; the notional is gone and the Max DD % cell on that row (marked †) is a divide-by-near-zero artifact rather than a real drawdown. An em dash (—) means the cell is empty in the source data — 4 CAGR values, 1 max drawdown and 1 Sharpe could not be computed on fully wiped runs. All figures are stored at full source precision and rounded for display only.
What this grid does NOT prove
- It is a measurement of the past, not a forecast. Every number here comes from historical price data for the dates and instruments stated above. It tells you what a setting did on that data — it does not tell you what it will do next.
- A winning setting here is not a guarantee it keeps winning. Market conditions change. A setting that led the 4-hour Silver results in this window can underperform in a different window.
- This is a backtest table, not advice. This is analytical software output for research and education — it is not a recommendation to trade any instrument, and not a signal to act on.
- Small changes move the numbers a lot. Swap the instrument, the timeframe, or the cost assumption and the ranking can change completely — that is exactly why the full grid is published instead of one "best" number.
Back to the guide
SuperTrend Indicator Best Settings & Strategy Guide
The methodology, the cost assumptions, the per-instrument winners and what the backtest actually says about fast timeframes are all explained there.
Read the SuperTrend settings guide →