LIQUIDATION REVERSAL LAB
Where forced flow breaks
Measure trigger reclaims, pivot depth, and multi-factor sweep confluence.
LIQUIDATION REVERSAL LAB
Where forced flow breaks.
A live view of liquidation cascades in BTC perpetuals — depth, depletion, and confirmed events across five venues — and a research record that, so far, argues against its own edge. Trading is disabled.
LIVE CONFIRMED LIQUIDATIONS
WHAT WE FOUND
The cheapest round trip we can price exceeds every effect measured across the completed studies. The current preregistered experiment is classified FAILED under its frozen rule. No strategy has been promoted.
Bybit public L50 depth, five-second depletion and recovery, exhaustion. No account.
Liquidation mapOPENProjected forced-flow exposure with confirmed cross-venue events on the same axis.
Confirmed tape & archiveOPENEvery liquidation from Kraken, Bybit, Binance, Bitget and Hyperliquid, with venue coverage and price basis attached.
HOW TO READ THIS
Nothing here is a calibrated probability. A composite that has not been fitted out of sample stays a −1…+1 score.
Resting liquidity, projected exposure and confirmed executions can share a chart. Their notionals are never added.
Missing data breaks the line rather than dropping to zero, and a venue-sampled feed is labelled sampled.
Secondary market detail Kraken taker flow
Trade tape
Taker side from Kraken Futures, newest execution first.
| Time | Taker | Type | Price | Size | Notional |
|---|
Cross-venue liquidations
Connecting to four public exchange feeds.
| Time | Venue / coverage | Position | Price | Price basis | Size | Approx. notional |
|---|
CONFIRMED EVENT ARCHIVE
Price-level liquidations across venues
Loading recent BTC events from the local archive and Hyperliquid history.
The local companion records Binance and Bitget continuously while it is running; Kraken and Bybit add BTC, ETH, and SOL. 0xArchive supplies bounded Hyperliquid completed-event history even when the companion was offline. Kraken and Hyperliquid prices are executions. Bybit reports bankruptcy prices. Binance and Bitget public streams are sampled by each venue. Venue, coverage, and price basis remain attached to every record.
EXCHANGE HISTORY
Liquidations beyond this session
Loading Kraken's five-minute liquidation history.
Kraken reports aggregate liquidation volume without position side. USD values are estimates: each five-minute liquidation volume bucket is multiplied by that bucket's Kraken close. Bybit's live price is its reported bankruptcy price.
The heatmap combines Kraken's visible resting order book with confirmed events from Kraken, Bybit, Binance, Bitget, and Hyperliquid. 0xArchive's Hyperliquid forced-liquidation map is computed from clearinghouse positions and margin state, then bucketed around a roughly 45-minute snapshot. Those hexagons are projected estimates—not completed liquidations, exact wallet liquidation prices, or trade instructions.
The horizontal axis is historical relative volume, the vertical axis is the selected performance return, bubble size is market capitalization, and color indicates gain or loss. The ranked candidate table provides the same assets as text.
Ranked candidates
Deterministic rotation score; candidates are research prompts, not orders.
| Rank | Asset | Bias | Score | Return | RVOL | 24h volume |
|---|
CROSS-SOURCE CONFIRMATION
Perpetual positioning behind the shortlist
Loading public Bybit positioning and Kraken liquidation context.
| Asset / contract | Bias | Funding | Basis | Open interest | OI / turnover | Spread | Kraken liq. 24h |
|---|
Funding, open interest, mark/index basis, turnover, and top-of-book spread come from Bybit's public linear-contract ticker snapshot. Kraken contributes exchange-reported five-minute liquidation volume for BTC, ETH, and SOL; USD values use Kraken candle closes and remain estimates. Missing contracts stay missing. No synthetic values are invented. These fields are context only and do not alter the rotation score.
FORWARD EVIDENCE
Test the shortlist on unseen prices
Local forward observation, experimental. Freeze a candidate cohort, then evaluate only later snapshots that arrive near the selected horizon.
Load a snapshot, then capture a fixed cohort to save market-only observations in this browser.
Bars show directional hit rate and a line shows average direction-adjusted return for each score band. Long returns retain their sign and short returns are inverted.
Evidence by score band
Current cohort only
| Score | N | Hit rate | Avg signed | Long / short |
|---|
This user-triggered convenience sample is prospective observation, not a historical backtest. Results are hypothetical directional observations, not probabilities, expected returns, trades, or evidence of Bybit profitability. Assets captured from one snapshot are correlated and are not independent trials. Aggregate prices exclude fees, spread, slippage, latency, funding, and execution failure. The page must be refreshed near each target; late windows are counted as missed and excluded. Records contain public market data only, remain in this browser profile, and can be erased or altered with browser storage. Samples below 30 hide percentages. Trading remains disabled.
Cross-asset snapshot: Sharpe heatmap API. Derivatives confirmation: Bybit public tickers and Kraken market analytics. These sources provide current and historical market observations, not future returns. This interface never places a trade.
HISTORICAL STUDY
Liquidation reversal study
Loading dataset…
Extension versus event size
Reversal rate by side
Event inspector
Review the trigger, extreme, and reclaim window candle by candle.
SIGNAL RESEARCH / FORWARD TEST
Live directional factor lab
Research objective: predict the sign of BTC's next close from market information available now.
Waiting for live inputs
Connecting to the market feeds.
Measured factors
DOWN -1 / UP +1| Factor | Reading | Score | Source |
|---|
Automatic scheduled evidence
Loading unbiased five-minute collection snapshots.
| Captured (UTC) | Direction | Consensus | Coverage | Outcome | Signed move |
|---|
Prospective evidence
No forward observations captured in this browser.
| Captured (UTC) | Horizon | Direction | Consensus | Status | Signed move |
|---|
This is an untrained, equally weighted research consensus—not a probability, expected return, trade, or order. Factors are hypotheses. Captures are stored only in this browser and outcomes exclude fees, spread, slippage, latency, funding, and execution failure. Percentages remain hidden until 30 resolved observations. Trading stays disabled.
Filtered event ledger
Audit every event used in the visible reversal rate.
| Time (UTC) | Side | Liquidations | Z-score | Outcome | Extension | Reclaim bars |
|---|
Study definition
Predictive research score
The live factor lab averages six signed measurements from -1 (downward pressure) to +1 (upward pressure): nearby liquidation-map pressure, L2 depth imbalance, aggressive taker flow, confirmed liquidation impulse, two-minute momentum, and contrarian derivatives crowding. Missing feeds reduce coverage instead of being replaced with invented values.
The consensus is intentionally uncalibrated. It must accumulate timestamped forward observations and survive out-of-sample evaluation before it can support any probability claim or trading rule.