Markov Trend Probabilities with ATR Regimes and Brier Calibration

Summary

This indicator estimates whether recent market regimes favor an uptrend or downtrend and tracks the calibration of its own probability forecasts. It classifies each bar using price change over a lookback window scaled by ATR. Changes beyond a positive or negative threshold set the regime; values inside the dead zone retain the previous state. Counts over a rolling history window produce empirical state probabilities and first-order transition rates. A separate rolling Brier Score compares the previous bar’s uptrend probability with the state subsequently observed, with lower scores indicating better recent forecast accuracy.

The text explains how to interpret the probability lines and calibration score, and suggests uses as a trend-following filter, a mean-reversion screen, or an input to position sizing. It also documents implementation choices in a ProRealTime port, including delayed prediction pairing to avoid using future outcomes. These are indicator mechanics and application suggestions, not evidence of trading profitability: no performance study is reported, and estimates depend on chosen windows, thresholds, and regime labels.

Key ideas

• Trend states are classified from ATR-scaled price changes, with a dead zone that carries forward the previous state.
• Rolling state counts estimate uptrend and downtrend probabilities, while consecutive states estimate persistence.
• The Brier Score compares lagged probability forecasts with later observed states to track calibration.
• The indicator can filter trend-following or mean-reversion strategies and inform exposure sizing.
• The document describes implementation mechanics but provides no profitability or robustness results.

Tags

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