Prove the edge.
Then trade it.
Backtest.ai gives systematic options traders what the fixed-menu tools can't: the ability to express an original idea, the statistics to know if it's real, and the loop to trade it live with confidence.
Idea in. Evidence out.
Three stages between a hunch and a live account.
Open data, a simple DSL.
Other testers hand you a menu of indicators. We hand you the data. Write any signal you can express, or start from 100+ templates.
Tell a real edge from noise.
Scatter, distribution, and decile plots render on every signal automatically. Walk-forward models catch the curve-fit before your capital does.
What you validate is what you trade.
Backtest at broker fidelity. Real margin, market accurate fill and slippage model. Then promote the exact tested strategy to paper or live.
Built for the question the others skip:
is it real?
Research the idea, not just the trade.
Every backtester answers "what did this trade return?" We answer the more important question first: does your signal predict anything at all? Correlation, distribution, and decile evidence before you risk a single run.
Validation that survives contact with the future.
Models train walk-forward: fold tables, coefficient stability, out-of-sample scatter. Memorizing the past doesn't pass. Use model outputs as entry gates or sizing functions; a construct no closed platform has.
No fantasy fills. No margin surprises.
The backtest can't take a position your broker would reject, and can't exit at a price the market never showed. Market accurate fill and slippage model, margin simulation, and sizing that survives the broker's math.
The backtest is the deployment.
Promote a backtest, or an allocation-weighted portfolio, and the platform snapshots every signal, model, and parameter. Backtest, paper, and live run the identical engine, and every event lands in an append-only ledger.
If your current tool can do all this, keep it.
Signal construction
Statistical validation
Models
Execution realism
Research → live
Data depth
Infrastructure
Honest caveat: if a preset menu covers your ideas and a green equity curve is proof enough, keep your tool. Switch when you have an edge to prove, or an idea the menu can't hold.
An AI that computes, not chats.
An AI quant assistant, not a chatbot: it writes DSL, wires signals, and runs real computations you can inspect. Ask it to build. Ask it to explain. Never ask it to guess, it won't.
Straight Answers.
Good, skepticism is the point of the product.
When does early access open?
We're targeting September 2026. Subscribers get in first, the earlier you join the list, the earlier your invite.
Is the engine a black box?
No. The engine is deterministic. The same inputs produce the same fills, every run. And "open" isn't just the DSL and open data: the whole engine is documented. The fill model, the methodology behind every metric, the assumptions the engine makes, it's all written down and inspectable. You can build any signal you want through the DSL without writing engine code or running your own infrastructure, and see exactly how every result was produced.
How is this different from OptionOmega or OptionAlpha?
They're strategy testers: pick from a menu of filters, get an equity curve. We're a quant platform: write any signal on the raw data, validate it statistically, and deploy the exact tested strategy live. Their workflow is a subset of ours.
What data do you have?
Minute-resolution stock and etf data, options chains, quotes, greeks, IV, back to 2012, plus point-in-time earnings and macro calendars. 10+ liquid symbols at launch (SPX, SPY, QQQ…), expanding by vote of the early cohort.
Do I need to know how to code?
If you can read a few lines of a formula, you can write a signal. The DSL is simple and the AI assistant explains, extends, and debugs it with you. There's nothing to host or maintain.
How realistic are the backtests?
Dedicated slippage, market impact, and fill models reproduce real execution conditions instead of idealized fills, so results survive contact with live markets. What you validate is what you trade.
What will it cost?
Pricing lands with the beta. Waitlist members see it first, and the founding cohort locks a rate that never goes up while subscribed. There will be a free tier generous enough to prove the engine is real.
Your best idea deserves a real test.
Join the waitlist, bring the strategy your current tool can't express.