AI backtest · paper trade · MT5 deploy honest by default

Backtest before
you trust.

Most tools stop at the pretty green line. We run 9 anti-bias rules and tell you when a strategy is actually overfit — before you risk a cent. Flip the card. ↓

Free forever tier · no credit card
// same-looking edge, opposite verdict — tap a tab
studio — backtest enginelive
you › long SOL 1h when EMA20 crosses EMA50, TP 6% SL 4%
Compiled ✓ — 3-year backtest complete
Real Bybit data · walk-forward · costs applied. This edge holds up on data it never saw.
EQUITY · $10k → $14.2k+41.8%
Win rate57%
Profit factor1.9
Max drawdown-12%
// no look-ahead · costs applied · force-close 168hPASSED ANTI-BIAS ✓
Deploy anywhere
BybitBinanceBitgetOKXMetaTrader 5
one workflow
From idea to validated live signal.
Without leaving the app. Describe it, test it honestly, paper-trade it, deploy it.
01

Chat-built strategies

Plain English → compiled strategy. 11 indicators, leverage & liquidation model.

02

Honest backtest

Up to 16 years of real market data — gold, FX, crypto — on 6 timeframes. 9-rule anti-bias gate. Warnings first.

03

Live paper trading

Push notifications on fills. Full trade log with charts. No money at risk.

04

Deploy when ready

Fan out to Bybit, Binance, Bitget, OKX or MetaTrader 5. Cancel anytime.

why trust our numbers
9 rules that kill
fake backtests.
Every result runs this gauntlet. If a strategy only worked because of a bias, we flag it in red — at the top, impossible to miss.
01
No look-aheadfuture data can't leak in
02
Walk-forwardtested on unseen data
03
Costs appliedslippage + fees, 0.2% RT
04
Sample sizen too small = flagged
05
Force-closeno infinite open bags
06
Overfit checkparam-tuning detector
// + no-survivorship · regime awareness · warnings-first — 9 checks total
⚠ REJECTED · rule 6/9

“It made +124% though.”

In-sample, sure. On data it never saw: −38%. That's the strategy most tools would have sold you — and exactly what our engine stops at the door. We'd rather show you the loss now than let you find it live.

inside the studio
Type it. Watch it compile.
No spreadsheets, no Pine Script. Just describe the idea.
you
long BTC 1h when EMA20 crosses EMA50, exit on RSI > 70
T
Compiled ✓ 3-year backtest · Bybit BTC/USDT · force-close 168h
412Trades
54%Win rate
1.7Profit factor
-14%Max drawdown
Heads up: the edge is concentrated in the 2023 bull run — regime-locked. We flag this at the top of every result, before the profit number.
Ready — paper-trade it forward, or deploy to Bybit.
failed a prop challenge?
60–70% of failures are a drawdown violation — not a bad strategy.

A stop that cannot fire before liquidation. A bet that does not shrink as the account does. A timer doing the work your take-profit was credited with. One good year carrying a nine-year total. All four are measurable before you pay a fee — and our engine reports them above the profit number, not in a footnote.

Read the post-mortem →

pricing
Start free.
Free
$0
no card, no expiry
  • 25 backtests / day
  • AI builds with the 7-day trial
  • Manual builder
  • 6 months of 1h history
Try free
Starter
$12.50/mo
billed yearly · $15 monthly
  • 500 backtests / day
  • 60 AI builds / month on our keys
  • Deploy to real exchanges
  • 3 live paper traders
  • MetaTrader bridge (self-hosted)
Try 7 days free →
POPULAR
Pro
$24/mo
billed yearly · $29 monthly
  • 3,000 backtests / day
  • Discovery Agent (100 runs)
  • 150 AI builds / month · 15 paper traders
  • 9 years of history
Try 7 days free →
Elite
$49/mo
billed yearly · $59 monthly
  • Unlimited backtests / day
  • Discovery Agent (500 runs)
  • All available history (gold back to 2010)
  • Cloud MetaTrader — no PC needed
Try 7 days free →

Stop trusting screenshots.

Backtest your first strategy in the next 60 seconds. Free, no credit card.

Try free →