How the verdict works

The rule

Right now that reads 7 of 13 bullish, so the verdict is YES, held from its last answer because neither side leads by 3.

Why a lead of 3 rather than a simple majority: rebuilt day by day from free historical data back to March 2018, a simple majority changed its answer about 20 times a year, and its median answer lasted six days, because the count sat within one vote of the line on 28% of days. Needing a lead of 3, held for about a day, cut that to about four changes a year. Its answers also matched hindsight bull and bear phases (counting any fall of 50% or more from a peak as a bear market) on 80% of days, against 72%. The cost is speed at some turns: after the October 2025 top it would have said NO after 39 days, not 25. The full rebuild, including what the original rule got right and wrong, is written up in how accurate this tracker is.

That last answer is the only thing this site stores. If the store is unreachable, a simple majority decides until it is back. If a data source is down, its signal is excluded and the lead is counted over the signals that remain. An outage never counts as a bearish vote.

The 13 counted signals

They deliberately span five independent categories (price trend, sentiment, institutional flows, retail attention, and on-chain activity) so no single kind of market action can flip the verdict alone.

A note on correlation, before anyone counts: these are not 13 fully independent measurements. Four of the votes read the Bitcoin price series over different horizons, miner revenue largely co-moves with price, the two ETF votes read the same flow series, Fear & Greed is counted at two speeds, and search trends feed into Fear & Greed itself. The realized price trend is on-chain, but coins realise more profit when prices are high, so it agreed with the 200-day momentum vote on 86% of days between April 2023 and September 2026. That structure is deliberate. Price regime is weighted heavily because price regime is most of what “bull market” means, and the remaining votes (flows, stablecoins, sentiment, on-chain activity) exist precisely so a price squiggle alone cannot flip the verdict. Group them yourself and it is roughly seven independent inputs; the verdict needs breadth across them, not unanimity within any one.

Context indicators (shown, not counted)

Eight indicators appear on the dashboard but never vote. The halving cycle is a calendar, not a measurement; it would stay “bullish” through a crash, and days since the cycle top has the same problem. BTC dominance and funding rates were counted until September 2026 and moved here because both got stuck: dominance has sat above 55% on every day since early 2025, and funding’s 7-day average cleared its +11% baseline on 3% of days in 2025 and none in 2026, so each could only vote no. A vote that can give only one answer carries no information. Pump.fun activity is the most tempting to count, because Solana memecoin speculation is a vivid read on risk appetite. It stays out for two reasons: the platform has only existed since 2024, so a threshold would rest on a single cycle, and it covers one venue on one chain while the counted signals are Bitcoin or market-wide. PumpSwap volume is the same venue after tokens graduate, and carries the same caveats. Total DeFi TVL is excluded for a different reason: it is priced in dollars, so it falls when prices fall, and counting it would quietly weight price more heavily than the thirteen-signal split claims.

Changes to the signals

September 2026: a clear lead to change the answer, and 15 signals down to 13. The verdict now changes only when one side leads by 3, for the reasons under the rule above. BTC dominance and funding rates became context indicators, because each could only vote no. And the all-time-high signal’s recovery leg now measures from the lowest close since the top rather than the 52-week low, which after the 2017 peak still held 2017 prices and read bullish on 218 of 288 days of the 2018 bear market.

September 2026: Coinbase app rank retired, realized price trend added. Apple’s public chart feed only lists the top 25 finance apps, and Coinbase was usually outside it, so the signal rarely had a rank to read and was counted as bearish by default. Realized price replaced it. It had been shown as context, because its textbook rule (price above realized price, or MVRV above 1) held on every day of the previous twelve months, including a fall of more than half from the October 2025 high. It now votes on its 60-day direction instead, which turned bearish in January 2026 while the price was still falling. The evidence for that choice is on the realized price page.

Data & freshness

Every signal is fetched server-side from public APIs and cached for roughly an hour (some slower-moving on-chain series longer). Realized price comes from a free feed that runs about a week behind. The verdict’s last answer is kept in a small store so the lead-of-three rule can hold it between hourly updates. The page you see is prerendered with the latest data and refreshes itself automatically. All sources are attributed on each card and on the signal explainer pages.

What this is not

Not financial advice, not a trading system, and not a prediction. The thresholds are reasonable but ultimately judgment calls. The lead-of-three rule and the September 2026 changes were checked against a day-by-day rebuild of the verdict from free historical data, but they were chosen after seeing that data, so the rebuild flatters them. It is an honest, transparent aggregation of the indicators people actually argue about, built so you can check the state of the market in one glance and dig into any signal you disagree with.