1. Define rules you can actually test
A backtest applies trading rules to historical market data. It helps you inspect what those rules would have done under a particular set of assumptions. It does not establish what will happen in the future.
Replace “buy when the trend looks strong” with a precise condition. Specify the exchange, market type, symbol, timeframe, entry, exit, position size, and whether decisions use completed candles. Record the rules before inspecting results.
- Choose one market and timeframe for your first test.
- Include an exit rule and decide whether short positions are allowed.
- Use a separate period to evaluate the rules after developing them.
2. Prepare the data and settings
Record the date range, timezone, initial capital, position sizing, commission, and slippage. Check that the selected symbol has data covering your intended period. A missing candle or a different market type can change a result.
Indicators need historical observations before their values are meaningful. Allow enough warm-up data and inspect when the first eligible signal occurs. Never use information from a later candle to decide a trade on an earlier candle.
SeekAlgo documents these controls in Platform settings. Keep a copy of the actual settings returned by each run so that another test can reproduce the same assumptions.
3. Draft, validate, and run with your assistant
Connect a compatible AI assistant through SeekAlgo MCP and sign in. Ask it to find the exact market symbol, draft your rules, show you the script, validate it, and then run a backtest. Review the script before accepting its interpretation of your idea.
For example: “Draft a long-only moving-average crossover strategy. Enter on an upward cross and exit on a downward cross. Show the parameters, sizing, costs, candle timing, and data range before running it.” The assistant may need your choices before this becomes a complete test.
Validation checks a script; it does not certify the quality of its trading logic. Compare a few signals and trades with the price chart, especially around entry and exit boundaries.
4. Read beyond net profit
Inspect the trade list, drawdown, number of trades, exposure, and costs. A high return from very few trades can be fragile. A winning percentage alone says little without the sizes of gains and losses.
Compare against a suitable benchmark over the same dates. Repeat on an untouched evaluation period and modestly different parameters. Report weaker results as well as stronger ones; repeatedly choosing the best historical settings can overfit noise.
5. Observe the strategy in paper trading
If the strategy is eligible, create a paper bot and inspect its signals and simulated trades as new market data arrives. Record differences between historical and forward behavior before changing the rules.
SeekAlgo paper bots do not place real orders. Simulated fills cannot establish actual exchange execution quality, and neither backtesting nor paper trading guarantees returns.
Product references
Try this workflow with SeekAlgo
Build and validate a strategy in the workspace, then inspect the backtest. Historical and paper results are simulations, not a promise of future performance.
Explore the workspace →