Method: Official documentation and pricing checked September 28, 2026. No benchmarks were run. Recommendations are based on documented features and billing, not hands-on results.
A strategy ranking needs reproducible evidence. This guide explains how to examine a backtest using Freqtrade’s documentation. It does not report a completed experiment, a live portfolio, or a strategy that has been shown to make money.
What a useful backtest record includes
Before comparing results, retain the strategy source, configuration, software version, exchange, trading pairs, data range, candle interval and fee assumptions. State the starting balance and position limits. Without that context, a percentage return is difficult to interpret or reproduce.
Freqtrade’s backtesting documentation explains the historical-data requirement, result reports and simulation assumptions. It supports an explicit fee ratio applied on entry and exit. The documentation also warns that some dynamic pairlists cannot produce reproducible historical results.
Fees belong in the model
Kraken’s fee schedule currently lists its entry spot tier at 0.40% maker and 0.80% taker. Check the applicable product, pair and account tier before using those rates. Maker and taker describe liquidity treatment; a limit order is not automatically a maker fill.
For an illustrative $100 trade notional, 0.40% is $0.40 on one side. Two $100 notionals at that rate would incur $0.80 in fees. Actual entry and exit notionals can differ, and spread and execution effects are separate. This arithmetic is not a simulated or realized strategy result.
Check for information from the future
Freqtrade’s lookahead analysis is designed to help find strategies that inadvertently use future information. A favorable result produced with information unavailable at the decision time does not establish a tradable strategy.
Use the documented analysis and inspect any reported issues before treating the return table as evidence. Passing one diagnostic does not validate every assumption in the experiment.
Check indicator initialization
Recursive analysis examines how indicator values change with different amounts of starting data. That matters when historical calculations and an operating bot may begin with different histories.
Record the initialization choice alongside the strategy. Review whether the indicators are stable enough for the intended use rather than assuming a large historical dataset settles the question.
Separate development from evaluation
A practical research plan should define the rule and evaluation period before reviewing the result. Keep later data separate from the period used to adjust the strategy. Report losing periods and drawdowns alongside any gains, and compare with a stated baseline using consistent dates and costs.
Also retain rejected variations. Showing only the best version conceals how many attempts produced that result. These are recommended research controls, not a claim that this site completed them.
Recommendation
Start with a documented, reproducible simulation and investigate fees and bias before considering live execution. No profitable strategy is recommended here, and no backtest guarantees a future return. This page omits the previous strategy rankings and live-account claims because supporting results are not established by the vendor documentation cited below.
Sources, checked September 28, 2026: Freqtrade backtesting, Freqtrade lookahead analysis, Freqtrade recursive analysis, Kraken fees.