Write it the way you'd say it
"Buy when the 20-day average crosses above the 50-day, sell when it crosses back." That is a complete, runnable strategy. No syntax, no parameters panel, no indicator library to learn first.
Free · no account needed
Say the strategy.
See how it held up.
One plain-English sentence. AlphaBack writes the Python, runs it over twenty years of real market history, and shows you exactly what would have happened — drawdowns and all. Roughly five seconds.
No card. No install. Five free runs before you sign up for anything.
20yrs
of daily market history
~5sec
sentence to full report
0
lines of code you write
100%
of the Python is yours to read
How it works
Three steps. None of them are
"learn pandas".
We turn it into Python
A model writes a real backtesting.py strategy, and we execute it in a locked-down sandbox against actual daily bars from Yahoo Finance. You can read every line it wrote.
You get the whole story
Return, worst drawdown, win rate, every individual trade, month-by-month heatmap, and how the same rules held up through the 2020 crash and the 2022 bear market.
What you get
A report, not a number
One return figure tells you almost nothing. These are the things that actually decide whether an idea survives contact with a real account.
The equity curve
Not just where you ended up — the shape of getting there, with the drawdown shaded underneath so you can see what you'd have had to sit through.
Every trade, listed
Entry date, exit date, price in, price out, profit and loss. The full log, so a suspiciously good result is something you can go and check rather than trust.
Month-by-month heatmap
Steady, or one lucky quarter carrying five flat years? A grid of monthly returns answers that in about a second, which no headline figure ever does.
Stress tests
The same rules re-run through the COVID crash, the 2022 bear market and a flat sideways stretch. Most strategies look excellent until one of those three.
The Python it ran
Collapsed by default, one click away, yours to copy. It is proof rather than product — but a result you cannot audit is a result you should not act on.
Side-by-side comparison
Put up to five versions of an idea on one chart, aligned on a shared calendar so a shorter run can't fake an advantage. Change one rule, see what it cost you.
The honest part
What a backtest cannot tell you
If a tool in this category isn't saying this out loud, ask why.
- Whether it will work next year. A backtest is a rehearsal against one recording of the past. Markets are not obliged to repeat it.
- Whether you'd have held on. A 13.8% drawdown is a number on a chart here. In a live account, in the fourth month of it, it is a decision you make at 2am.
- What you're really measuring. Test forty variations and one of them will look brilliant by luck alone. That's not an edge, it's arithmetic.
- The fills you'd actually get. We model commission and fill on the next bar. We do not model your broker being slow on the worst day of the year.
You already have the idea.
Find out if it held.
Five free runs, no card, no install. About five seconds each.