Write a quant strategy, see how it played out.
Write the Python yourself, or describe the idea in one sentence and let AI produce the code — both run the same backtest. Metrics, curves and a trade-by-trade log. The code AI writes is code you can read, edit and re-run, not a black box.
Real output from the workspace. Strategy and buy-and-hold drawn as return %, so neither line is squashed flat.
Write code, run a backtest, read the result
Two entry points, one destination: write the strategy yourself, or let AI turn a sentence into one. Either way the code runs on the same historical data.
01Hand-write a strategy in real Python
Subclass a base class, declare parameters, return a signal series. Indicators are hand-written (SMA, EMA, RSI, ATR, Bollinger, ADX) so nothing is a black box, and there is no drag-and-drop builder in the way.
The source you see is the source that run actually executed.
02Describe it, let AI write it, backtest it
Say "long above the 20-day average, flat once it breaks" in plain language. You get back a commented strategy class that runs as-is, plus a note on its logic and its known weak points. Do not like it? Edit it and run it again.
03Parameter scan and out-of-sample check
One good parameter set proves nothing. The heatmap shows which results are fragile; the out-of-sample check splits the history and runs both halves — profitable early and losing late means you fitted the past, not a pattern.
04Metrics, curves and a trade log you can export
Win rate, profit factor, expectancy, max drawdown — each with a note on what it hides. Plus the equity curve against buy-and-hold, the drawdown chart and every fill, exportable as an HTML report or CSV.
Write, run, read, change one thing
Get the strategy code in place
Edit a built-in template, write your own, or have AI generate it from a sentence. All three land on the same readable Python file.
Set symbol, timeframe, costs and risk
Fees, slippage, risk per trade and the daily loss cap all live in the sidebar. Those assumptions decide how good the result looks, so they have to be yours.
Run it and read the audit, not just the return
Metric cards, equity curve against buy-and-hold, drawdown, and a list of the engine's own limitations. If the books do not balance, the run is flagged.
Change one variable and run it again
Loosen the filter, shorten the stop, drop the trend guard — one at a time, and watch which way the number moves. Stop when the result stops surprising you.
$19 a month. That is the whole price list.
No annual lock-in, no credit packs, no "pro" tier hiding the useful parts. Cancel on the same page you subscribed on.
Everything below, for as long as you stay subscribed.
- Every strategy in the workspace, fully commented and editable
- Unlimited backtests on 5 years of daily history
- Out-of-sample check on every strategy
- Parameter scan with heatmap, to see how fragile a result is
- AI helper that writes a strategy from your description, with an explanation
- Export reports and trade logs for your own notes
Subscribing means you agree to the Terms, Refund Policy and Privacy Policy.
Billed monthly by card. Cancel any time; access continues to the end of the paid month. First purchase: 7-day no-questions refund.
Access starts immediately after payment. EU consumers: the checkout asks you to confirm immediate access and to acknowledge that you lose the 14-day right of withdrawal; without that confirmation, access starts when the withdrawal period ends.
Not sure yet? Try it free — 5 backtests a day, no card. Create an account to start; it never turns into a paid subscription on its own.
No card — just create an account
The trial needs an account (email and password) so the daily allowance is per person rather than per browser. No card, and no hidden "trial ends and we start charging" step. When the allowance is used up, it stops.
✓What the trial gives you
5 backtests, 1 parameter scan, 2 out-of-sample checks and 2 AI generations per day. That is enough to run five backtests, change three parameters, and watch a strategy fail its split test once.
✓What it does not give you
Exported reports carry a trial watermark and saved presets are per session. Subscribing removes both, and your past runs are kept.
No credit card. You can come back and look at the price later.
Questions that decide it
Do I need to know Python already?
if, lists,
and what an import does. If you have finished any beginner Python course you are
ready. You do not need pandas — the workspace and its comments cover the parts that matter.Does the AI-written code actually run a backtest?
Do I need a trading account, or money at risk?
Will it tell me what to buy?
Can I download the source code?
What data does it use?
What happens if I cancel?
Is a positive backtest a reason to trade?
Important
This is an educational product. It teaches programming and the statistical interpretation of historical simulations. It does not execute orders, does not connect to any broker or exchange, does not predict prices, and does not provide investment, legal, tax or accounting advice.
Trading involves substantial risk of loss. Backtested results are simulations produced by a model with documented limitations — no intrabar path modelling, no gap fill modelling, survivorship bias in the symbol list. Simulated results routinely differ from live results and past performance does not indicate future results. Nothing here is a recommendation to trade. If you later choose to trade, never risk money you cannot afford to lose.