Compare the approach
Choose a method by its requirements, costs, and evidence. This starting framework is not a ranked provider list.
Affiliate disclosure: we may earn commissions from links explicitly marked as affiliate links. The framework below contains no paid placements. Read the disclosure
| Approach | Use case | What to verify | Main limitation |
|---|---|---|---|
| Backtest | Historical strategy experiment | Dataset, timestamps, cost model, unseen sample | Optimistic fills or overfitting |
| Paper trading | Observe automated behaviour | Simulator rules, signal timing, rejected actions | Simulated fills differ from execution |
| Documentation review | Understand advertised capability | Feature documentation, API permissions, dated terms | Capabilities not verified hands-on |
| Recorded live test | Observe actual fills and failures | Explicit scope, logs, costs, reconciliation | Limited sample and actual capital exposure |
Compare like with like
A profitable backtest is a result to investigate. It is not evidence that a bot will produce the same return with real orders.
Keep assumptions visible
Use the same amount, market, network, time window, and cost basis when those factors apply. Record the original source and check date for product-specific information. A low advertised fee is incomplete without the conditions that make it available.
Check the evidence
Separate an actual test result from a calculation or a statement in product documentation. A review should describe its method, scope, and unresolved questions. We do not assign star ratings or publish benchmark winners without evidence.
Run the numbers
Estimate the monthly trading return needed to cover a subscription and execution costs. This is a cost model, not a return forecast.
Open calculatorGo to the source
Understand the mechanics
Build a useful trading bot evaluation: account for fees and slippage, separate training and test data, and examine failures before trusting a backtest.
Evaluate a trading bot