AlphaProve tests crypto trading strategies against historical data, with trading costs and out-of-sample checks.
BTCUSDT.P · +0.29%
ETHUSDT.P · +0.85%
SOLUSDT.P · +0.22%
XRPUSDT.P · +0.53%
DOGEUSDT.P · +0.39%BTCUSDT.P · +0.29%ETHUSDT.P · +0.85%SOLUSDT.P · +0.22%XRPUSDT.P · +0.53%DOGEUSDT.P · +0.39%BTCUSDT.P · +0.29%ETHUSDT.P · +0.85%SOLUSDT.P · +0.22%XRPUSDT.P · +0.53%DOGEUSDT.P · +0.39%BTCUSDT.P · +0.29%ETHUSDT.P · +0.85%SOLUSDT.P · +0.22%XRPUSDT.P · +0.53%DOGEUSDT.P · +0.39%BTCUSDT.P · +0.29%ETHUSDT.P · +0.85%SOLUSDT.P · +0.22%XRPUSDT.P · +0.53%DOGEUSDT.P · +0.39%BTCUSDT.P · +0.29%ETHUSDT.P · +0.85%SOLUSDT.P · +0.22%XRPUSDT.P · +0.53%DOGEUSDT.P · +0.39%Your AI can read the complete AlphaProve SDK reference before it creates or checks a strategy.
Connect your AI- 01
Share your idea
Test a BTC trend-following strategy.
- 02
Create the strategy
Your AI turns it into a strategy and checks it for errors.
- 03
Run a backtest
Test it on historical data, then open the strategy and results in AlphaProve.
Describe your strategy and test it against historical data.
Trades indexed
That is six years of Bybit perpetuals: every trade and every one-minute candle. Every fill you see was priced against it.
Try your strategy on live prices with simulated money.
Paper trading keeps running when you close the tab. Review trades and positions, pause the session or close every position.
Explore crypto paper trading →- 09:35:50Filled · long 0.5 @ 64,357.6
- 09:26:40Exit @ 64,251.1−0.15%
- 09:18:45Filled · short 0.5 @ 64,154.4
Give the Lab a saved strategy to refine and test, then compare the results.
A strategy needs more than a good-looking backtest. Account for trading costs, test on unseen data, and understand where it fails.
Check how your strategy holds up on unseen data and under different trading costs.
Use forward-walk tests, Monte Carlo analysis and parameter sweeps to see how much the result depends on your assumptions.
How walk-forward testing works →
Here is what people usually try instead, and what each one leaves out.
| AlphaProve | Chart-tool tester | DIY notebook | Institutional platform | |
|---|---|---|---|---|
| L2 order-book fills | — | — | ||
| Forward-walk validation | — | By hand | ||
| Monte-Carlo risk simulation | — | By hand | ||
| Describe a strategy in plain English | — | — | — | |
| Paper trading on hosted servers | — | |||
| Priced for one person, not a desk | — |
- L2 order-book fills
- Forward-walk validation
- Monte-Carlo risk simulation
- Describe a strategy in plain English
- Paper trading on hosted servers
- Priced for one person, not a desk
- L2 order-book fills—
- Forward-walk validation—
- Monte-Carlo risk simulation—
- Describe a strategy in plain English—
- Paper trading on hosted servers
- Priced for one person, not a desk
- L2 order-book fills—
- Forward-walk validationBy hand
- Monte-Carlo risk simulationBy hand
- Describe a strategy in plain English—
- Paper trading on hosted servers—
- Priced for one person, not a desk
- L2 order-book fills
- Forward-walk validation
- Monte-Carlo risk simulation
- Describe a strategy in plain English—
- Paper trading on hosted servers
- Priced for one person, not a desk—
Start free. Pay only when you have an edge worth proving.
- ·20 AI credits / month
- ·90-day data lookback
- ·L2 order-book backtests
- ·Backtests + tearsheets
- ·300 AI credits / month
- ·2-year data lookback
- ·Code editor
- ·Forward-walk optimization
- ·Monte-Carlo risk sim
- ·Up to 50× leverage
- ·Priority backtest queue
- ·Everything in Trader
- ·1,500 AI credits / month
- ·Up to 100× leverage
- ·Highest-priority queue
- ·Unlimited lookback
- AI strategy assist: describe an edge, get a strategy
- L2 orderbook fills with depth-based slippage
- Bybit margin, funding & liquidation modelling
- Library, version history & full tearsheets
- Forward-walk validation + Meta-ML
- Sandboxed strategy execution running fully on our end
- Hosted paper trading, nothing to install
- Connect your AI with MCP
- ML strategy generator
- Strategy marketplace
- Stock-market integration
- Forex pairs
- Fundamental-analysis signals
- AI that takes the trades for you
- Engineering & research blog
Start with the evidence.
Are backtests predictive?
No. A backtest tells you how a strategy would have behaved in the past. Testing with trading costs and on data you haven’t tuned it to helps you judge the result, but cannot predict future returns.
Where does the data come from?
From Bybit. We store historical candles and order-book snapshots to simulate trades with spreads, fees and slippage.
What markets are supported?
Five Bybit USDT perpetuals: BTC, ETH, SOL, XRP and DOGE, on timeframes from one minute to one day. Paper trading covers the same five.
Do I need to know how to code?
No. Describe your idea to the strategist or build the rules in the no-code editor. There’s also a code editor if you prefer to write your own.
Can I use my own AI assistant?
Yes. Connect Claude, Claude Code, Codex, ChatGPT or another compatible MCP client to create strategies and run backtests. Your AI can read the full strategy SDK, and your results are saved in AlphaProve.
What access does my AI get?
You approve permissions once per connection. Runs use the same plan allowances as the website. You can review permissions or disconnect an app in Settings → Agent access.
Is this beginner-friendly?
You can create and test a strategy without writing code. Paper trading lets you try it on live data with simulated money before deciding whether to trade it.
Which plan should I choose?
Start on Free, with 90 days of data. Trader adds two years of data, forward-walk validation, Monte Carlo analysis and the code editor. Quant includes the full history and priority in the queue.
What are AI credits?
Credits cover work by the built-in strategist, such as drafts, revisions and fixes. Optimizer campaigns and Pine exports use more. Free includes 20 a month, Trader 300, Quant 1,500. Backtests themselves do not cost AI credits.
Where does my strategy run?
On AlphaProve’s servers, in an isolated sandbox. Your strategies are private to your account.
Is my data safe?
We store account and strategy data in the European Union and never sell it. Chats with the built-in strategist go to our AI provider; connected AI apps receive the data you authorize them to access. You can request a copy or deletion of your data. See our privacy policy for details.
What does the private beta include?
Request access using the signup form. We review new accounts before enabling sign-in. Once approved, you can use the beta within your plan’s limits.
Prove it before you trade it.
Test your strategy with historical data before putting money behind it.