Got a trading idea?
Say it in plain English.
AI codes it, backtests it,
and trades it FOR REAL.
You don't need to know Python. You don't need a finance degree. Describe your idea the way you'd say it to a friend — we test it against 20 years of real market data, and when you're ready, connect your broker and let it trade live.
YOU HAVE TRADING IDEAS.
THAT'S ENOUGH TO START.
If you've ever thought "I wonder if that would actually work" — this is where you find out.
You trade on gut feeling.
Time to check the gut.
You buy dips, follow news, maybe copy ideas from Twitter. Some trades work, some don't — and you honestly can't tell if you're good or lucky.
Your system works.
You just can't prove it.
Your rules live in your head, a spreadsheet, or a checklist — not in code. You can't test them. You don't know if the last three years were skill or luck.
You kept growing.
Your tools didn't.
PineScript hit its ceiling. ChatGPT writes code that doesn't understand look-ahead bias. You're copy-pasting between four tools every time you want to test one idea.
REAL PROMPTS.
REAL CODE.
REAL RESULTS.
Every one of these started as a single sentence typed in chat — some by people who can't code at all. Click through to see the prompt, the code the AI wrote, and the results. Remix any of them.
IDEA IN.
REAL TRADES OUT.
Honest data, the AI of your choice doing the technical work, and a one-click path from backtest to live trading.
you can trust.
We test your idea against 20 years of market history, exactly as it looked on each day — no peeking into the future, no forgetting the stocks that went bust. If the result looks good, it's because the idea was good.
codes for you.
Run your strategies on whichever model you trust — Claude, GPT, Gemini, DeepSeek. It writes the code, picks the right data, and avoids the classic mistakes that make beginner backtests lie. You talk strategy — it handles the math.
money — instantly.
The strategy you backtested is the exact same code that trades live. One click flips it from history to your real broker account — no rewriting, no re-testing, no surprises.
START BY CHATTING.
GO AS DEEP AS YOU WANT.
The platform has three gears. Most people start in first — and nobody has hit the ceiling of third yet.
to a backtest.
Describe ideas in plain English, and the AI builds, tests, and refines them with you. This is where everyone starts — and honestly, where the first limit shows up: the chat is only as good as the ideas in your head. Sooner or later, you'll test them all.
mom = pct_change(data, 21)
top5 = mom.nlargest(5)
for s in top5:
order_target_pct(s, 0.2)
set_stop_loss(s, 0.08)
Browse for new ones.
A visual library of ready-made building blocks — indicators, risk controls, entry and exit rules, instruments. Scroll through it like a catalog, snap pieces together in the interface, and discover combinations you'd never have typed into a chat. Still no code.
Zero limits.
Open the IDE and write classic Zipline-style Python — the same workflow quants have used since Quantopian. Every parameter, every edge case, every custom idea is yours to build.
Explore docs →The levels aren't a ladder you're forced to climb. Mix them: rough out an idea in chat, enrich it with blocks, fine-tune it in code. The platform meets you where you are — and doesn't run out of room when you grow.
THE ENGINE
THE PROS
TRUSTED.
Never heard of Quantopian? It was the platform where a generation of quants learned to trade, until it shut down in 2020. Its engine, Zipline, was the gold standard for honest backtesting — and we kept it alive.
We adopted Zipline, kept it open source, and rebuilt everything around it: institutional data, AI, and live brokers. If you were on Quantopian — same engine, better everything. If you weren't — you're getting the tool the professionals grew up on, minus the learning curve.
Today the engine keeps moving forward under the ZipLime name — same lineage, actively maintained. We ship upgrades to the core regularly, and the whole thing stays open source. Curious how the backtester actually works under the hood? We record ZipLimeTech, a video series walking through the internals, in public.
Zipline — open-sourced 2012 · adopted by Limex 2024 · evolving as ZipLime
▶ ZIPLIMETECHYour Zipline Backtest Was Lying — Commission Calculated After the Trade
▶ ZIPLIMETECHNo Live Trading in Zipline? We Built It In From the Start
▶ ZIPLIMETECHStop Hardcoding Your Strategy — Parameter Management in ZipLime
▶ ZIPLIMETECHPolars + Parquet in a Quant Engine — The Real Performance Difference
▶ ZIPLIME INTERVIEWWhat Zipline Should Have Become — Rust, Async & AI-Powered Backtesting
▶ ZIPLIMETECHBlocking Calls Cost You Money in Live Trading — Why ZipLime Uses AsyncGOOD ANSWERS NEED
GOOD DATA.
Free data from Yahoo quietly lies to you — it knows the future and forgets the losers. We buy the same feeds hedge funds buy, so your results mean something.
hedge funds use —
accurate to the tick.
50,000+ signals
a day.
machine-read within a
minute of filing.
FROM A SENTENCE TO
REAL TRADES IN FOUR MOVES.
Test it — watch it trade with fake money — go live when you're ready. Each step is one click, and you can stop at any of them.
BEFORE YOU ASK.
ANSWERED.
Straight answers — not the ones marketing wishes you'd ask.
How much does this cost?
Start on Free — no card required. It includes a limited number of backtest runs and a usage limit on the AI models (with a starter set of LLMs available). Pro and Max raise those limits and unlock the full model lineup — including the heavyweight models for code generation. See exact limits on Pricing →
Where does my algorithm run? Do I need to keep my computer on?
No. Everything — backtests, paper trading, and live strategies — runs in our cloud on AWS. Close your laptop, go to sleep; your strategy keeps watching the market 24/7. You get notifications when it trades.
Do you hold my money?
Never. Your money stays at your broker — we never touch it. The platform connects to your brokerage account through SnapTrade and sends orders on your strategy's behalf. You can pause or disconnect any strategy in one click, any time.
What is strategy optimization?
Say your idea is "buy when RSI < 30." Why 30? Why not 25 or 35? Optimization runs your strategy across a whole grid of parameter values and shows which ranges hold up — across different market regimes, not just overall. The goal isn't to find one magic number (that's how you overfit) — it's to see whether your idea is robust or fragile.
How do you make sure a strategy isn't just overfit to the past?
Chasing one perfect 20-year curve is a trap. Markets change character: the calm grind of 2017, the COVID crash, the 2022 rate shock, the AI rally — each is a different regime, and a strategy that shines in one can bleed in another. Tweak an idea until the whole curve looks perfect, and you've usually just built something that only works on the past.
So instead of one 20-year number, we give you a ready list of named market regimes — calm bull runs, crashes, rate shocks, meme-stock manias — and let you run your strategy against the ones that matter. You choose what to test; we make sure you're testing regimes that actually happened, not just the overall average.
Version control keeps you honest along the way. Every run gets a version, nothing is quietly deleted, and each tweak shows what it changed regime by regime — so you always know whether you improved the strategy or just repainted history.
Do I need to know how to code?
No. Chat and the block library cover most of what you'll ever want to test. Python is there when you outgrow them — and the AI will happily explain any line of code it writes.
The backtest showed a profit — will I actually make money?
Honest answer: not necessarily. A backtest shows how an idea would have performed — it's the best filter we know, but the future isn't the past, and trading always carries risk of loss. That's exactly why the path here goes backtest → paper trading → live, and why results are broken down by market regime. Never trade money you can't afford to lose.
How safe is connecting my broker?
The broker connection goes through SnapTrade using the broker's official authorization — you log in on your broker's own page, and we never see or store your brokerage password. Trading permissions are limited to placing orders; withdrawals to third parties aren't possible through the connection.
Who owns my strategies?
You. Strategies are private by default — your code, your ideas, your edge. Nobody sees them unless you share them.