What State Is the Market's Hidden Liquidity In?
Before price moves, money changes state: it parks, it builds charge, and eventually it deploys or unwinds. PsiCharge models that hidden state from venue-level dollar-liquidity dynamics our recorders capture around the clock. The inputs stay proprietary; the receipts do not: every state's measured next-24h outcome is published below, split by year, wins and losses alike.
PsiCharge is at 47 and the market is in SUPERPOSITION. Charge is building and holding: liquidity is parked, direction undecided.
Today's read: what could happen from here?
Over the last 12 months, from this exact state, Bitcoin's next 24 hours averaged -0.30% (median +0.03%, closed up 51% of the time; next 72h avg -0.81%, up 48%).
Measured history, not a promise: states describe liquidity, price decides itself.
When Psi speaks, what does price do?
The Psi score (colored river) against Bitcoin (white, right axis) over the last 120 days. Diamonds mark state alarms on the price itself: red = Purge, green = Collapse. Judge the model by where its marks land, not by our words.
The receipts: every state, every year
Measured next-24h Bitcoin outcome for each alarm state, split by year, weak years included. Rules were frozen on 2022-2024 data; everything after is out-of-sample. This table updates as history accumulates.
| State | Year | n (hours) | Avg 24h BTC | Closed up |
|---|---|---|---|---|
| COLLAPSE | 2024 | 44 | +0.13% | 66% |
| PURGE | 2024 | 53 | -0.86% | 32% |
| COLLAPSE | 2025 | 416 | +0.75% | 62% |
| PURGE | 2025 | 174 | -0.11% | 53% |
| COLLAPSE | 2026 | 303 | -0.21% | 47% |
| PURGE | 2026 | 165 | -0.89% | 42% |
Model record starts 2024-08-07; states are computed hourly from our own venue recorders.
Methodology, honestly bounded
PsiCharge is a proprietary multi-input liquidity state model computed from our own venue-level recorders. The inputs and weights are not disclosed. What is disclosed, permanently: the measured outcome of every state, split by year, including the years it underperformed.
The rules were frozen on 2022-2024 data and everything since is out-of-sample. Purge has been the most consistent state across years and assets; Collapse is regime-dependent and is labeled as such. Neither is a trade signal: they describe the liquidity state, not your position.
PsiCharge is not a crash predictor, and we say so because we tested it: major single-day crashes were not systematically preceded by Purge. What Purge has measured is a persistent next-24h downward tilt with slightly elevated volatility.
What exactly does PsiCharge measure?
The state of dollar liquidity on the dominant trading venue: whether it is parking, holding charge, deploying into the market, or unwinding stored stress. The specific inputs are proprietary; the state definitions have been stable since the model was frozen.
Why should I trust a model whose inputs are secret?
Because the outputs are not. Every state's measured outcome is published year by year, including the weak years, and the rules were frozen before the out-of-sample period. Judge the receipts, not the recipe.
What is the difference between Collapse and Purge?
Both are discharge states, but they differ in how much stress the system had stored. Collapse is a low-stress discharge, historically constructive in up-trending regimes. Purge is a high-stress discharge, historically the most consistent risk-off reading across years, on Bitcoin and Ethereum alike.
How often does it alarm?
Purge has fired about twice a month and Collapse about five times a month in the recent record. Rare enough to mean something; each alarm posts to our feed automatically.