Backtesting & robustness analysis
Prove your strategy isn't curve-fit.
A backtest that looks perfect usually is β perfectly fitted to the past. Noon Barbari is a research tool for stress-testing your own strategies, so you can tell a genuine edge from a curve you accidentally drew through noise.
Run honest out-of-sample tests, score how robust a rule set really is, and watch how it behaves on data it was never tuned on β before you ever risk capital. This is methodology, not predictions.
Free, no card required. A tool for testing your own ideas β not buy/sell calls or advice.
The 4-second version
Same tools. Two verdicts.
A beautiful backtest means nothing until data it never saw agrees. Watch what a walk-forward pass does to each of these.
Tuned until it looked perfect
Kept honest with walk-forward
Numbers are illustrative. The methodology is the product: every strategy here gets this exact treatment before you risk anything.
Why your backtest is probably lying to you
Curve-fitting hides behind a beautiful equity curve. These are the failure modes a single in-sample backtest can never reveal.
It looks too good
A flawless historical curve is the classic warning sign, not the goal. The more a strategy is bent to fit the past, the less it tends to survive the future. A tool can only help if it tries to break the result, not flatter it.
You tuned it to the answer
Tweaking parameters until the chart looks great is curve-fitting by another name. Every knob you turn while staring at the same data quietly memorises that data instead of learning anything general.
Lookahead crept in
Using information that wouldn't have existed yet β a closing price, a future bar, a revised dataset β makes any backtest look smarter than it could ever be live. It is one of the easiest mistakes to make and the hardest to spot by eye.
Survivorship skewed it
Test only on the coins and periods that happened to do well and you measure luck, not method. Honest evaluation has to include the runs you'd rather forget.
How we try to break it instead
Every tool here exists to attack your strategy from a new angle. Surviving them is the closest thing to evidence of a real, repeatable edge.
Robustness scoring
Instead of a single backtest number, you get a read on how stable a rule set is across conditions β a plain-language read on whether the result looks structural or fragile and curve-fit.
Walk-forward analysis
Tune on one slice of history, then test on the next slice the strategy has never seen β rolling forward through time. It mimics how a strategy would actually have been deployed, exposing parameters that only worked in hindsight.
Monte Carlo & overfit probability
Reshuffle and resample the trade history thousands of times to estimate how much of the outcome was process versus luck, and how likely the chosen settings are to be overfit rather than sound.
Honest out-of-sample
The decisive test is data the strategy was never tuned on. We keep that separation explicit so a result you trust on paper has actually earned it, not borrowed it from the in-sample fit.
TradingView bridge
Import a Pine strategy and stress-test it honestly
Saw a viral TradingView strategy with a perfect-looking chart? Import the Pine script and put it through the same out-of-sample and walk-forward checks β the parts the screenshot never shows.
It is the fastest way to find out whether a popular strategy holds up on data it wasn't demonstrated on, or whether it only ever looked good on one cherry-picked window.
Your data. Your server. Your rules.
Every strategy is an explicit, readable rule set β never an opaque black box you have to take on faith. You can read it, version it, and audit exactly why a trade would have happened.
Run the engine yourself if you want full sovereignty over your research. No vendor sees your edge, and nothing about how a result was produced is hidden from you.
Stress-test a strategy in the next five minutes
Pick a strategy, a starting balance and a date, and the free backtester replays history honestly β no login, no card. When you want to save and iterate on your own rule sets, create an account.
A research tool for validating your own strategies. No buy/sell calls, no investment advice.