Backtest crypto strategies on Binance history
Build the rules in a visual editor, replay them bar by bar over spot and USDT-M futures since 2020, and see what fees, funding and drawdowns would have done to the result. No code, nothing to install.
How it works
1Describe the rules
Combine indicators and conditions in the visual editor, start from a template, or ask an AI assistant to write the config. Either way you end up with a JSON document, not code.
2Run the backtest
The server replays the strategy bar by bar on spot or futures, charging fees, slippage and funding and liquidating positions the way Binance would.
3Read the report
Equity and drawdowns, every trade on the chart, monthly returns, market regimes and Monte Carlo. Over 60 metrics in all.
4Try to break it
Compare runs with a significance test, sweep the parameters with walk-forward, and keep only what survives out of sample.
Strategies are JSON documents
You can read one, diff two versions and keep a history of changes. Nothing in a config is executed as code, which is what makes it safe to run on a shared server.
- AND / OR / NOT groups, crosses, ranges, rising / falling and look-back windows
- Higher timeframes and other symbols inside the same rule, read without look-ahead
- Several entry layers, pyramiding, DCA grids, resting stop and limit orders
- Take-profit and stop-loss in %, ATR or R: ladders, break-even, trailing and time exits
- Five sizing modes up to volatility targeting, plus daily loss and drawdown guards
- Named parameters, version history and a dry run before the full backtest
1{2 "name": "Trend pullback",3 "market": { "type": "spot", "exchange": "binance" },4 "instrument": { "symbol": "BTCUSDT", "timeframe": "4h" },5 "sides": ["LONG"],6 "risk": { "feesPct": 0.1 },7 "entries": [{8 "id": "pullback", "side": "LONG", "sizePct": 50,9 "conditions": { "operator": "AND", "list": [10 { "left": { "indicator": "EMA", "params": { "period": 50 } },11 "op": "GT",12 "right": { "indicator": "EMA", "params": { "period": 200 } } },13 { "indicator": "RSI", "params": { "period": 14 },14 "op": "CROSS_UP", "value": 30 }15 ] }16 }],17 "exit": {18 "atrRef": { "indicator": "ATR", "params": { "period": 14 } },19 "stopLossAtrMult": 2,20 "takeProfitR": 321 }22}
What is inside
The market data is already loaded: candles, funding, open interest, and the perpetuals Binance has since delisted.
- Strategy builder
- 100+ indicators in 14 groups, from RSI, MACD and Ichimoku to market structure, candle and chart patterns, open interest and funding. A preview chart shows where the rules fire while you edit.
- Backtest engine
- Spot and USDT-M perpetual futures: long and short, 1–125× leverage, isolated or cross margin, historical funding every 8 hours and liquidation by the Binance formula.
- Backtest report
- Sharpe, Sortino, Calmar, max drawdown, profit factor, expectancy, buy-and-hold alpha and more. Bar-by-bar trade replay, regime breakdowns, a printable tear-sheet and CSV export.
- Compare & batch
- Put up to four runs side by side with a significance test, shift the start date or sweep a symbol matrix in one batch, and preview how several runs combine as a portfolio.
- Optimizer
- Grid and random search over up to 12 parameters, walk-forward with rolling or anchored windows, robustness scoring and sensitivity heatmaps. Promote a winner as a new strategy version.
- Research Lab
- Event studies and feature scans against forward returns, with multiple-testing correction, a held-out half and the edge left after costs. Ideas go through it before they become strategies.
- Analog Lab
- Finds the moments in history that looked like the current one and shows what followed: a fan of paths against a baseline, and a TP/SL plan tested on every episode.
- Market data
- Binance candles since 2020 plus funding, open interest, long/short and taker ratios, delisted perpetuals included. Export candles, indicators and labels as CSV datasets.
- Strategies written by AI
- Describe an idea in plain words to ChatGPT, Claude or another assistant: with the strategy reference Janorium publishes, it writes a config you paste into the editor. Scoped API keys open the same API to your own scripts.
Where backtests usually flatter a strategy
Six common ways a backtest makes a strategy look better than it is, and what the engine and the report do about each.
Fees and slippage left out
Maker and taker fees, slippage (constant, ATR-scaled or by volume impact), a cap on the share of bar volume a fill can take, and historical funding.
Look-ahead
Higher-timeframe reads only see closed bars, and orders fill at the bar close or the next open.
Stop and target in the same bar
When one bar touches both, the stop counts first, unless the 1m or 5m bars inside it show which came first.
Only the coins that survived
Hundreds of delisted perpetuals stay in the data, so a symbol universe can include coins that later died.
Too many variants tried
A Deflated Sharpe note that counts how many variants were tried, q-values and held-out halves in Research Lab, and a randomized-entry baseline.
One lucky stretch
Monte Carlo resampling, drawdown episodes and regime breakdowns show how much a result depends on luck or on a single market phase.
A candle with a J cut out of it
The candle is the basic unit of every price chart. The J stands for Janorium, and also for Janus, the Roman god who looks into the past and the future at once. Janorium is about the first look: what a strategy would have done on real history, so the decisions about what comes next rest on evidence rather than hope.
Questions people ask first
What is backtesting?
Backtesting means running a strategy's rules over historical market data to see how it would have traded: every entry and exit, the fees it paid, the drawdowns along the way. It shows whether an idea held up in the past before you put money behind it — not that it will hold up in the future.
Read the full guideCan I look around before signing up?
Yes. Open any strategy from the leaderboard. Public ones open in full — config, version history, every run and the complete report with trades — exactly as their owner sees them. Private ones show their results without their rules: headline metrics, equity curves and the summary of every run. An account lets you run your own backtests, clone public strategies and use the labs.
Do I need to write code?
No. A strategy is a declarative config: you build it in the visual editor or paste JSON. Nothing you submit is ever executed as code.
Which markets and data are covered?
Binance spot and USDT-M perpetual futures with history from 2020: 1h and 4h bars for every tracked symbol, down to 5m for the major pairs. Futures also carry funding rates, open interest, long/short and taker ratios; BTC and ETH add mark and index prices and order-book depth.
Does a good backtest mean the strategy will make money?
No. A backtest shows how a set of rules would have behaved on past data. Janorium works hard to make that picture honest — costs, funding, no look-ahead, overfitting checks — but markets change, and nothing here is financial advice.
How is the leaderboard ranked?
By the Janorium score, not by the best return. It multiplies the quality of the typical run — median annual return, Sharpe, average drawdown, share of profitable runs — by the evidence behind it: how many distinct backtests, across how many symbols, timeframes and years, with how many trades. Only runs of the current version that paid fees count, so one lucky fee-free run cannot top the table.
Can an AI write a strategy for me?
Yes. Give ChatGPT, Claude or another assistant the strategy reference Janorium publishes and describe your idea in plain words — it writes the JSON config. Paste it into the editor and press Apply; if Janorium reports an error, send it back to the assistant to fix.
Read the full guideIs it free?
Yes. Janorium is an independent portfolio project — sign up and run backtests at no cost.
Run your first backtest
Sign up, start from a template or from your own rules, and read the report. It is free.
Create an account