#61 — Continuation vs exhaustion inside resolvable moves

Hypothesis — what it tests

Whether, inside a move that is already underway (rather than at a cold standstill bar), indicator readings predict continuation (more favourable pips remaining) or exhaustion (the move is nearly spent).

Description

The cold direction-at-a-bar framing (entry #60) was arguably the wrong question — the median optimal hold is ~85 seconds, under one M5 bar, so single-bar direction tests were doomed. This experiment reframed it per the user's intuition: given a move is already underway, do indicators tell you whether it is continuing or finished? Using the oracle's optimal trades as ground-truth move segments, every bar inside a move of ≥10 minutes was scored against each causal indicator (oriented by the move's direction), measuring correlation with the favourable pips remaining to the move's end and with the fraction of the move elapsed. Indicators do carry signal here — |r_cont| up to 0.18, move-age correlation +0.42, consistent across AUD_JPY and GBP_JPY — but uniformly in the exhaustion direction: the stronger an indicator reads in the move's favour, the LESS move is left. The conclusion redirected the whole indicator library toward the exit/fade role, feeding the later fade/retrace and exit-overlay work.

Key result

Indicators

Algorithms

Code