How to read TradingView's Strategy Tester report
TradingView throws a dozen metrics at you the moment a strategy compiles. Most of them are noise; a few are the difference between a winning system and an expensive lesson. Here's what each number is for — and which ones to trust.
The four panels
- Performance Summary — headline numbers (net profit, PF, max DD, win rate).
- List of Trades — every trade, entry/exit price, P&L — the only objective record.
- Properties — settings (capital, commission, slippage, order size). Garbage in, garbage out — see realistic backtest costs.
- Trades Analysis — distributions (best/worst trade, run-ups/drawdowns, time-in-market).
The metrics that actually matter
| Metric | What it tells you | Trust level |
|---|---|---|
| Net profit | Total $ result over the sample | Low — depends entirely on sample period |
| Profit factor (PF) | Gross profit / gross loss | High if PF 1.4–3; suspicious above 5 |
| Max drawdown | Worst peak-to-trough on the sample | High — the most useful single number |
| Win rate | Share of trades that won | Medium — meaningless without avg win/loss |
| Avg win / avg loss (R) | Payoff ratio | High — combines with WR for expectancy |
| Sharpe / Sortino | Risk-adjusted return | Medium — see trade-offs |
| # of trades | Sample size | Critical — below 100 is mostly noise |
The red flags that scream “overfit”
- Profit factor above ~5 on a non-trivial sample — almost always curve-fit.
- Win rate above ~85% with no explanation (e.g. obvious tight-stop mean reversion) — likely look-ahead or repaint.
- Equity curve a straight line — no drawdowns — suspicious; real systems breathe.
- Few trades (under ~50–100) — sample too small for any conclusion. See statistical significance.
- Net profit huge, max DD near it — one or two outlier trades carrying the system. Calmar will be ugly.
The numbers most traders ignore (and shouldn't)
Max drawdown duration — how long the equity curve stayed underwater. A 20% DD that recovers in two weeks is psychologically very different from one that takes a year. Largest losing trade — tells you the worst single hit; check whether it would have breached the prop firm's drawdown limit. Trade frequency — affects whether the modeled blow rate is even measurable in your evaluation period.
What the report can't tell you
Whether the result holds out-of-sample. Whether live execution will match the backtest. The distribution of possible futures, which only Monte Carlo reveals. Use the Strategy Tester as a first pass, then stress-test in the Monte Carlo Simulator before trusting any number. Before acting on any of these numbers, run the trade list through the backtest quality checker — it flags the sample-size, cost and repainting problems this report can hide.
Why backtests and live results differ: the full map →
The systems the alerts come from.
Every strategy and portfolio that runs on TradingView, with the preset per account tier. One PDF, then our letters with what the systems did.
By email, from Puerto Viejo, Costa Rica. Unsubscribe anytime.
Related
- Invite-only TradingView strategies: how access actually works
- What is MACD? The indicator explained
- Pine Script repainting: what it is and how to detect it
FAQ
What's the most important metric in TradingView's Strategy Tester?
Max drawdown combined with number of trades. Net profit and win rate look impressive but mean little without a meaningful sample size (100+ trades) and a drawdown number that fits your account's risk tolerance.
What does a high profit factor mean?
Profit factor in the 1.4–3 range is usually a real edge. Above 5 on a non-trivial sample is almost always a sign of overfitting or look-ahead bias, not skill.
Can I trust a TradingView backtest?
As a first pass, yes. As proof of an edge, no — the report tells you what happened on one specific sample with one specific set of parameters. Real validation needs out-of-sample testing and Monte Carlo stress.
Not financial advice. Performance figures referenced are hypothetical, modeled outputs (1,500-path Monte Carlo on a backtest + live sample). Past performance does not guarantee future results. Tool names are referenced for education; verify current features and prop-firm rules directly.