Walk-forward · Deflated Sharpe · real L2 spread · live paper-traded

The backtest looked great.
The money didn't survive it.

Your own idea, a signal service, or a post that went viral yesterday. We test it against 3 years of real Binance data with honest fees. If the edge isn't real, we tell you. No signup.

Under 2 minutes. No signup. No credit card. Most strategies fail. We will say so.

Test a strategy now
10/10 free tests remaining

Type your strategy in plain English. We test it against 3 years of real Binance data with honest fees, tell you exactly what broke it, and suggest one-click tweaks to try next.

Or pick:

No signup. No credit card. Real Binance fees + L2 spread on every fill.

How it works

Think of us as the x-ray for your strategy. Diagnostic, not judgmental.

1

Describe your strategy

Plain English. No code. "Buy when RSI drops below 30" works fine.

2

We run it honestly

Real Binance fees. Real per-coin spreads. Walk-forward validation. 3 years of data.

3

Get your diagnosis

What’s working, what needs tuning, and which 2 tweaks would flip it. Usually 60-90 seconds.

Why we built this

Most backtests lie about real-world fees. We publish where ours does.

Real Binance fees. Real per-coin spreads. Three years of walk-forward data so nothing looks good by accident. Every night we measure where our backtest predictions disagree with our forward-test reality, and publish the gap at /calibration. Most strategies don't survive. The ones that do, and the 2-tweak fixes that save the borderline ones, are worth finding.

We tested 20 popular strategies with real fees.

Here's the gap between what YouTube claims and what your account would actually look like. Same strategies. $1,000. Two years.

EMA Cross (13/80)
Headline
+$457
Real
+$94
Triple EMA (9/21/55)
Headline
+$382
Real
$-47
RSI Scalper (14)
Headline
+$221
Real
$-156
Supertrend (10/3)
Headline
+$314
Real
$-22
Bollinger + RSI
Headline
+$153
Real
$-178
The methodology

80% of retail algo traders lose money to overfit backtests. Six gates we use to catch it.

Most retail backtesters skip the validation stack that actually catches overfit. Pine Script has no built-in walk-forward. Freqtrade leaves it DIY. The marketplace bots run on selection bias. We use the same six gates a quant fund would use, applied to every strategy you submit:

Gate 1
Walk-forward train/test split
Train on 70%, validate on 30%, no peeking. Catches time-period overfit.
Gate 2
Deflated Sharpe Ratio
Adjusted for the full testing universe. The math that says “is this a fluke given how many candidates we tested?”
Gate 3
Per-fill L2 cost modeling
Real Binance fees, real per-coin spreads sampled from order books. Not OHLCV-only.
Gate 4
Cross-coin transfer test
An edge that only works on one coin probably is not an edge. Validates generalization across BTC/ETH/SOL and 7 others.
Gate 5
Frozen holdout window
A specific recent window that our lab has never trained on. Genes pass walk-forward then face this last test.
Gate 6
Regime-stratified consistency
No single market regime can blow the strategy up worse than 1% per trade. Catches the “wins one regime, dies in another” pattern.

Every gate is in the public audit trail at /architecture and the nightly calibration log at /calibration publishes where our backtest predictions disagree with our forward-test reality. Most platforms don’t measure that gap. None publish it.

...
live paper trades
3 years
of Binance data
multi-coin
traded live · multi-exchange priced
real L2
per-coin spread samples

Built to catch the backtests that look bulletproof and aren't. Independent, no VCs, honest about the losers.

Email me when a new strategy passes all 4 gates.

Of 9,628 strategies tested in our lab, 20 have passed walk-forward + Deflated Sharpe + cost modeling + regime-conditional edge. Get notified the moment number 21 lands. Roughly one email a month, sometimes less. No newsletter. No course pitches.

Free forever to test any strategy. Want unlimited tests and the survivor library ( 21 of 9,615 strategies that passed all four gates)? $10 one-time, lifetime access →

Every strategy has a truth. We just find it faster.

Under 2 minutes. No credit card. You'll know exactly where your edge is and isn't.

Test a strategy free →