How the heat readings work
The model asks how unusually stretched or active a market is compared with completed historical data. It does not calculate the probability of a top.
Reading the scale
Each input is normalised to the same 0–100 heat scale before the weights below are applied. Values can move sharply as markets move; labels are descriptions, not instructions.
Bitcoin Heat
Pi proximity compares the 111-day moving average with twice the 350-day moving average. The model measures proximity before a crossover as well as the other independent signals.
Altcoin Market Heat
Breadth uses CoinMarketCap’s 90-day Altcoin Season Index. Absolute non-Bitcoin market growth is included so relative outperformance during a falling market cannot look extremely hot on its own.
Overall Market Heat
The combined score is 55% Bitcoin Heat and 45% Altcoin Market Heat. An Extreme overall label also requires genuine agreement: at least one side must be 80 or above and the other 60 or above. A high average without that agreement is capped at Hot.
Coin Heat
Coin Heat requires at least 210 completed daily observations and reliable volume. Stablecoins, wrapped/bridged/staked representations, very new assets and coins that cannot be matched safely are left unscored.
Timing, smoothing and sources
Heat uses completed UTC daily data, avoiding a score that changes continuously inside an unfinished candle. The market status uses a three-day median after three daily snapshots exist. Live portfolio prices update separately and do not change the daily Coin Heat score.
Primary references: CoinMarketCap public market data and Binance public market-data endpoints.
Historical sanity checks
Early checks produced high readings around several known 2021 speculative peaks and cool readings during tested 2022 bear-market lows. This is useful model-development evidence, but it is not an independent audit, a complete backtest or proof that future cycles will behave similarly.