How ArrayManager Warm-Up Length Can Change RSI Backtests

Summary

This forum question examines why changing the initialization length of a trading system’s ArrayManager can materially alter a backtest. The strategy uses RSI generated through TA-Lib, and the author suspects that the indicator’s path dependence makes its values sensitive to the amount of history loaded before the test begins. When trades open or close on threshold crossings, even modest value differences can change signal timing.

The author adds a floating take-profit mechanism, which can amplify those initial differences: a different early position or exit may lead to a different position later in the backtest. The example contrasts five versus twenty preload days and asks whether such sensitivity can be resolved or indicates weak strategy stability. The document presents the problem and a practitioner’s reasoning, but provides no responses, solution, comparison study, or evidence about how much warm-up is sufficient. It therefore raises useful concerns about indicator initialization and path-dependent trade sequences without establishing a diagnosis.

Key ideas

• The amount of preloaded history can affect RSI values calculated during a backtest.
• Threshold-based entries and exits can turn small indicator differences into different trades.
• A floating take-profit rule may amplify early trade differences into divergent later positions.
• The question compares five and twenty preload days but reports no confirmed solution.
• Sensitivity to initialization is a reason to investigate robustness, though the post does not prove the strategy is unstable.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.