How ChainPulse works

Regime states are determined using a composite momentum and volatility-adjusted scoring function derived from live Binance market data. Each asset is scored across three timeframes simultaneously — macro (1D), trend (4H), and execution (1H).

Scores are classified into five states: Strong Risk-On, Risk-On, Neutral, Risk-Off, and Strong Risk-Off.

Coherence measures directional alignment across the three timeframe layers. High coherence indicates all layers agree — highest conviction signal. Fragmented coherence indicates transition noise.

ChainPulse applies survival analysis to historical regime duration data. Given that a regime has lasted X hours, what is the conditional probability it continues for another Y hours?

This transforms regime detection from a static label into a time-aware probabilistic model. A Strong Risk-On regime lasting 200 hours has materially different survival characteristics than one that just started — even if the current score is identical.

The hazard function estimates the instantaneous probability of regime transition at any given moment. A rising hazard rate indicates increasing statistical fragility — even if the regime label has not yet changed.

Hazard rate is expressed as a percentage relative to historical baseline for that regime type. A hazard rate above 70% means the current regime is experiencing more failure pressure than 70% of historical regimes at equivalent age.

The recommended exposure percentage is derived from a weighted composite of regime strength, survival probability, hazard rate, and coherence. The output is a single number: how much capital belongs in this regime.

This is not a signal engine. It is a risk allocation framework. The output tells you how much to deploy — not what to buy.

Every time a user logs their actual exposure, the model compares it to the regime recommendation. Discipline score tracks protocol adherence over time. Behavioral alpha analysis identifies specific patterns that cost the user money.

This is the feature that most signal services do not build. Knowing the regime is valuable. Knowing whether you actually followed the model — and what it cost you when you didn't — is the edge.

ChainPulse is a decision-support framework. It is not a prediction engine and does not guarantee outcomes. The model can and does produce incorrect signals.

The system is designed to counter the most common and costly behavioral error in active trading: sizing up late in mature trends and sizing down early in persistent ones. Even a modest improvement in exposure timing produces asymmetric risk-adjusted returns over time.