Structure study

Bitcoin after a confirmed Weekly Bearish Breakdown

Performance after a completed weekly close loses the latest confirmed pivot low by 1%.

VERY LOW · N=7 MONITORING Last event 2026-06-08 Model 1.1.0
HOW TO READ THIS

Context before conviction.

After removing overlapping forward windows, N=7 independent observations remain from 13 raw events. Their 26W median return was -21.49% with a 42.9% win rate and -42.69% median adverse excursion.

A 7-event independent sample can produce extreme win rates by chance. Treat this result as exploratory.
OVERLAP CONTROL7 of 13 completed 26W observations retained · 46.2% overlapping events excluded from the primary estimate.
FORWARD RETURNS

Outcome by horizon

Primary non-overlapping results are shown beside the complete raw event population.

HorizonPopulationSampleMedianMeanWin rate25th percentile75th percentileMedian MAE
4WIndependentN=14+1.2%0.0%64.3%-11.0%+7.8%-9.4%
4WRaw eventsN=14+1.2%0.0%64.3%-11.0%+7.8%-9.4%
8WIndependentN=13+0.3%0.0%53.8%-14.2%+5.5%-15.9%
8WRaw eventsN=14+1.4%+1.3%57.1%-13.8%+14.0%-14.0%
13WIndependentN=10+4.8%+6.3%60%-18.8%+26.4%-18.6%
13WRaw eventsN=13+2.2%+8.0%61.5%-8.3%+31.1%-17.1%
26WIndependentN=7-21.5%+6.3%42.9%-31.4%+51.5%-42.7%
26WRaw eventsN=13-17.4%+9.8%46.2%-27.0%+47.5%-42.7%
52WIndependentN=4+55.0%+36.6%75%+10.6%+81.0%-50.6%
52WRaw eventsN=11+52.9%+135.7%72.7%+7.1%+111.5%-43.7%
OUTCOME GEOMETRY

Distribution and drawdown

Inspect the stored event-level outcomes behind the aggregate table.

FORWARD RETURN · 26WDistribution
N=7
Most populated range-50.2% to -18.6%4 observations
MAXIMUM ADVERSE EXCURSION · 26WDrawdown by event
NON-OVERLAPPING
0.0%-14.5%-29.0%-43.5%-58.0%
10 Nov 2025-42.7% MAEReturn -21.5%MFE +2.7%
OUT-OF-SAMPLE CHECK

Temporal validation

Chronological 70/30 holdouts and expanding-window folds use only independent observations. Later outcomes never influence event selection or thresholds.

DETAIL · 4WUnstable

1 of 3 expanding folds retained the training median direction.

TRAIN · EARLIER 70%+4.8%N=9 · 2018-06-25 2022-11-14
TEST · LATER 30%-2.9%N=5 · 2024-07-08 2026-06-08
HorizonIndependent NTrain medianTest medianDirectionStable foldsAssessment
4WN=14+4.8%-2.9%Changed1/3 · 33.3%Unstable
8WN=13+2.6%-6.2%Changed1/3 · 33.3%Unstable
13WN=10+0.7%+12.5%Stable1/2 · 50%Mixed
26WN=7Insufficient Data
52WN=4Insufficient Data
EXPANDING-WINDOW FOLDS · 4W
FoldTrain throughFuture test windowTrain / test NTrain medianTest medianDirection
12022-05-162022-06-202024-07-087 / 3+8.8%+4.1%Stable
22024-07-082025-03-102025-11-1010 / 2+4.4%-8.3%Changed
32025-11-102026-02-022026-06-0812 / 2+2.7%-7.0%Changed
PARAMETER ROBUSTNESS

Sensitivity analysis

One detector setting changes at a time around the published configuration. Returns never determine which configuration is canonical.

ASSESSMENT · 26WInsufficient Data

At least 8 independent observations are required for the canonical and compared variants.

INDEPENDENT MEDIAN RANGEEligible configurations only
SAMPLE RANGENon-overlapping observations
ConfigurationParametersRaw NIndependent NMedianWin rateDirection
Breakout buffer 0.5%Pivotstrength 3 · Retestendweek 8 · Reteststartweek 2 · Failurebufferpct 1 · Breakoutbufferpct 0.5 · Retesttolerancepct 1.5N=13N=7-21.5%42.9%Too small
Canonical detectorPublishedPivotstrength 3 · Retestendweek 8 · Reteststartweek 2 · Failurebufferpct 1 · Breakoutbufferpct 1 · Retesttolerancepct 1.5N=13N=7-21.5%42.9%Baseline
Breakout buffer 2%Pivotstrength 3 · Retestendweek 8 · Reteststartweek 2 · Failurebufferpct 1 · Breakoutbufferpct 2 · Retesttolerancepct 1.5N=12N=6-24.3%33.3%Too small
Pivot strength 2Pivotstrength 2 · Retestendweek 8 · Reteststartweek 2 · Failurebufferpct 1 · Breakoutbufferpct 1 · Retesttolerancepct 1.5N=17N=10+39.4%60%Changed
Pivot strength 4Pivotstrength 4 · Retestendweek 8 · Reteststartweek 2 · Failurebufferpct 1 · Breakoutbufferpct 1 · Retesttolerancepct 1.5N=11N=7+37.7%57.1%Too small

This is a robustness check, not model selection. Each row changes one assumption while every other detector rule remains fixed.

VENUE ROBUSTNESS

Cross-exchange benchmark

The published detector is replayed independently on each venue over the shared completed-candle window. Exchange populations are never mixed.

ASSESSMENT · 26WInsufficient Data

Both venues need at least 8 independent outcomes before direction is compared.

SHARED COMPLETED HISTORY2015-07-20 → 2026-09-01Identical calendar boundary for both detector replays
CANONICAL SOURCECOINBASEBTC-USD
VERY LOW · N=7
Completed candles4062
Raw / independent N13 / 7
Independent median-21.5%
Win rate42.9%
COMPARISON SOURCEBITSTAMPBTC/USD
VERY LOW · N=7
Completed candles4062
Raw / independent N13 / 7
Independent median-21.5%
Win rate42.9%

Coinbase remains the canonical research source. Bitstamp is an external robustness check; its candles, events, and forward-return population remain separate.

EVENT DEFINITION

Weekly close ≤ pivot low − 1% · one event per pivot

The signal is evaluated only with information available at the event close. Future observations are used solely to calculate outcomes.

STUDY RULES

Support is the latest strict weekly pivot low already confirmed by three right-side bars.

A completed weekly close must finish at least 1% below that pivot level.

Only the first qualifying breakdown is counted for each confirmed pivot.

Forward returns begin at the breakdown close; intraweek wicks do not qualify.

Primary confidence and headline statistics use the earliest chronological non-overlapping sample for each horizon; raw events remain reported for audit.

Temporal validation uses the independent sample only: a chronological 70/30 holdout plus expanding-window folds with at least two later observations per fold and no threshold tuning.

Sensitivity analysis changes one detector parameter at a time around the published canonical configuration, compares independent 26W outcomes, requires at least 8 observations per comparable variant, and never selects a parameter from performance.

Cross-exchange robustness replays the canonical detector independently on Coinbase BTC-USD and Bitstamp BTC/USD over their shared completed-daily history. Each source keeps its own signal and non-overlapping outcome population, with at least 8 observations required per benchmark.

Research, not a recommendation. Live market data can be delayed or temporarily unavailable. Historical results are displayed only when they have been computed and stored by the database research pipeline.