Build your bot.
Prove your strategy.
Put your rules up against the market.
Real Bitcoin data. Simulated money. Every move explained.
No deposit. No exchange account. Just your edge.
REAL MARKET PAPER TRADES
YOUR STARTING LINEUP / 20 STRATEGIES
Four bots. One arena.
Five research finalists per bot. Pick a strategy for each contender.
Your next run · all four bots share the selected candle interval.
How these 20 were selected
50 variants per bot, each tested on 1m, 5m and 15m BTC/USDT candles: 600 combinations. The five highest selection-period returns per bot were locked before checking the recent period. Rank uses each variant’s best eligible interval; changing the chart interval changes the displayed results, not the shortlist order.
Selection: Jan 2024–Jun 2026. Recent check: Jul–13 Oct 2026, 10:30 UTC. Full research: Jan 2022–13 Oct 2026. Every test starts with 1,000 USDT and includes 0.10% fee + 0.02% slippage per fill. These historical costs stay fixed when you edit costs for a new run. Active strategies needed at least 20 selection-period buys; HOLDER needed one. A different interval may have fewer buys and is labeled limited evidence.
HOLDER variants change the cash allocation only. A top-five position is relative to the tested family; some finalists lost money. Recent inactivity means no new buys, not confirmed performance. Historical results do not establish future returns.
DUEL LAB / YOUR PRIVATE WORKSPACE
An idea. A test. Your next move.
Shape your own rules, test real history, then take them live with paper funds.
01 / WRITE IT YOUR WAY
Start with a thought.
Describe the behavior you want. We turn broad ideas into proposed, editable dip-buying rules.
Text is sent to OpenAI. Proposed values are assumptions, not investment advice. You approve the rules before they are used.
02 / REAL HISTORICAL CANDLES
Make the rules prove it.
Test the exact rules currently in your editor. All four bots face the same data and modeled costs.
1,000 virtual USDT per bot. No future data in decisions. Testing the same period repeatedly can overfit the past. Beta: 6 tests per workspace daily, with shared capacity.
My recent backtests
Loading your test history…
03 / YOUR PERSONAL PLAYBOOK
Keep what you learn.
Save rules to your private workspace. Download a strategy sheet to inspect, print or adapt yourself.
Loading your strategies…
PROPOSED RULES / NOT ACTIVE
You decide what makes the cut.
Assumptions to review
Outside this bot’s capabilities
Clarify your idea
Your idea, written as exact instructions
Loads a draft only. Backtest or confirm a paper run when you are ready.
HISTORICAL SIMULATION / AFTER MODELED COSTS
Here is what the rules did.
| Contender | Final value | Return | Max drawdown | Fills | Fees |
|---|
Loading rules does not start or change a run. Review the editor, choose a duration and confirm. Downloads contain instructions and results; exchange execution must be implemented separately.
Real data appears here when the source is ready.
Confirm your rules to start a background run.
The standings
AFTER MODELED COSTSEach agent starts with 1,000 USDT. Open BTC positions are marked to the displayed price.
Every move. Explained.
RULE RECEIPTSWaiting for your first run. No invented trades.
KEEPS RUNNING WHEN YOU LEAVE
Your background agents
No runs yet. Start one from the strategy panel.
Up to 3 active runs per private workspace, for 1 hour, 24 hours or 7 days. Runs continue on the server when you close the tab. Public beta: up to 3 new background runs per network in 24 hours, with 30 active slots overall. Try a historical replay if you reach a limit. Finished runs: up to 30 days and the latest 30 per workspace (300 overall).
Your runs belong to this browser’s private code. Save it before switching device or clearing browser data. Coming from the preview? Save your code there, then use Restore my runs on stratduel.com. Keep the code private: anyone with it can control your runs.
Restore my runs
Current contenders & transparent assumptions
Spot BTC/USDT from Binance public candle data, refreshed about every 5 seconds. USDT is the quote asset; it is not labeled as USD. Decisions use only closed candles and a 60-candle warmup. Replay is a historical simulation on known data, with 240 playable candles where available. Live orders are modeled at the next candle open after the close is observed; this is not an actual exchange fill. Fees and slippage are assumptions. Pool liquidity, latency, order book depth and execution feasibility are not reconstructed. At expiry, the final mark uses the last eligible candle close/open. Each run retains 240 frames and 500 recent events. Replay links use the latest 1,000 stored snapshots; older snapshots expire. Outcomes are not evidence of future returns.
LET THE RULES DO THE TALKING.
Less “I would’ve.”More “here’s what happened.”
Watch the rule. Inspect the fill.
Change one thing. Run it again.Back to the arena