This page explains how BitThor frames indicator-combination research inside a historical simulation workflow. It is a theory guide for comparing signal families, not a prescription for live deployment.
This indicator-selection methodology page is educational, theoretical, and informational only. It discusses how indicator families can be compared inside historical simulations, but it does not tell you which indicators to deploy, does not guarantee future performance, and does not constitute financial advice, investment advice, or a recommendation to trade. Cryptocurrency trading involves substantial risk of loss, and past performance does not guarantee future results.
BitThor treats indicator selection as a hypothesis-design exercise. The research question is not "Which indicator wins forever?" but rather "Which indicator family showed a better fit inside one symbol, one candle range, and one archived simulation window?" By holding the market window constant, users can compare signal families without mixing in future data.
Moving averages, MACD-style structures, and related trend tools can be studied as directional filters. In theory, these families are most useful when the goal is to test whether sustained price movement remained coherent across the archived window.
RSI-style and rate-of-change families can be studied as acceleration or exhaustion measures. In a theory-first workflow, they are compared for how consistently they aligned with the next observed move inside the same archived test window.
Volume-weighted and volume-derived families can be treated as context layers. The theory use case is not prediction by itself, but measuring whether price signals looked more coherent when paired with participation or flow information.
Volatility bands and related tools can be framed as regime classifiers. They help users study whether a simulated window behaved more like a compression, expansion, or reversal environment before interpreting any other archived signal family.
ML-driven indicators can be treated as experimental comparison layers rather than guaranteed upgrades. In BitThor's theory framing, they are most useful when their archived fit can be compared against simpler baseline families over the exact same symbol and candle range.
One common theory model is to compare a directional family with a confirmation family. Example research framing: a trend family provides the directional hypothesis, while momentum or volume families are inspected for whether they reduced contradiction inside the same archived window.
Another theory model treats volatility or regime tools as gatekeepers. In that framing, a volatility family is not scored as the full strategy, but as a context layer that may help explain why another indicator family fit one archived period better than another.
Users can also compare a transparent baseline family against an experimental family such as ML. The theory question is whether the experimental overlay improved historical fit in the same archived window, not whether it can be trusted outside that window.
A final theory model compares concentrated setups to diversified ones. The research frame is whether one archived session looked more stable when decision weight came from several analytical viewpoints instead of one repeated family.
BitThor's methodology surfaces focus on comparing indicators against the exact symbol and candle window that produced the archived simulation result. That constraint is intentional: it reduces look-ahead leakage, keeps the comparison tied to historical evidence, and makes "Recommended Next Step" suggestions suitable for re-simulation rather than live activation.
This indicator-selection methodology page is educational, theoretical, and informational only. It discusses how indicator families can be compared inside historical simulations, but it does not tell you which indicators to deploy, does not guarantee future performance, and does not constitute financial advice, investment advice, or a recommendation to trade. Cryptocurrency trading involves substantial risk of loss, and past performance does not guarantee future results.