IMMUTABLE METHODOLOGY RELEASEModel 1.6.0
Model 1.6.0
Buffered weekly local structure reclaim
Requires a small confirmation margin above the previous completed week high before weekly local repair can move Broken structure to Improving.
0a8f79a31d90bbd806fa5b9a5e24ba0697622fd7bbef876d488658034f98fc6cThe pipeline recomputes this hash before every research rebuild. If code attempts to change this release under the same version number, publication stops and requires a new version.
RELEASE NOTES
What this version established
- Added a 0.25% confirmation buffer to the weekly local-structure reclaim branch after comparing 0%, 0.25%, and 0.50% across full history.
- Separated the immutable cycle origin from the active recovery level used to confirm current structural repair.
- Allowed the active recovery level to ratchet downward only after a continuation breakdown satisfies the production confirmation rule.
- Required two consecutive completed weekly closes above the active recovery level before Improving can return to Intact.
- Changed failed recovery to require a newly confirmed structural breakdown while keeping the existing cycle origin.
- Added earlier local recovery after two completed weeks without a new 20-day low, confirmed by a local 20-day high reclaim or a weekly close above the previous completed week high.
- Added an explicit Intact, Neutral, Broken, and Improving state machine with weekly breakdown confirmation and reclaim hysteresis.
- Replaced score-only regime assignment with a score-and-structure gating matrix plus a seven-day score-momentum tie-breaker.
- Added Compressed, Normal, and Expanding volatility context based on 20-day ATR relative to its trailing 50-session median.
- Historical cards now display regime, opportunity conditions, price structure, and volatility separately.
- Published a four-horizon comparison of low Opportunity Score entries with intact versus broken structure.
- Kept Opportunity Score inputs and weights unchanged.
FIXED SPECIFICATION
Reproduction contract
Schema 1 · implementation btc-regime-model-v1.6
data contract
Data contract
- Coinbase Exchange BTC-USD is the canonical technical source from 2015-07-20 onward.
- Bitstamp BTC/USD is used only as a separately replayed robustness benchmark.
- Signals, indicators, regimes, and research outcomes use completed UTC candles only.
- Daily candles are aggregated into UTC weekly and calendar-month candles without generated gaps.
regime score
Opportunity Score and final regime
- Opportunity Score weights are Trend 40, Momentum 25, Mean reversion 25, and Confirmation 10.
- Opportunity conditions are Low 0–40, Mixed 41–55, Supportive 56–70, and Strong 71–100; these bands describe model alignment, not market direction.
- The final regime is selected from opportunity conditions and point-in-time price structure before volatility confirmation is considered.
- Seven-day score momentum is Rising at +10 or more, Falling at -10 or less, and Stable otherwise; it is used only as a tie-breaker.
- Historical regimes are recomputed point in time and begin only after the 50-month EMA and its prior value are available.
canonical studies
Canonical detector studies
- Daily RSI crosses 75 with a seven-day cooldown; weekly RSI crosses 50 with an eight-week cooldown.
- Weekly bullish and bearish divergence use confirmed three-left/three-right pivots, 3–26 week separation, 1% price separation, and two RSI points of opposing momentum.
- Weekly higher-low and lower-high studies admit confirmed lifecycle events only.
- Weekly breakout and breakdown require a 1% completed-close pivot break; successful retests use weeks 2–8 and 1.5% level tolerance.
- Macro studies use completed monthly crossings of the contemporaneous 50-month EMA.
- The drawdown study records the first crossing of 30% below the running all-time high.
- The bottom-zone study records the first completed stabilization after a 30% drawdown cycle; the 20-day EMA must be reclaimed, its five-day slope must be positive, and daily RSI must be at least 45.
bottom zone
Point-in-time bottom-zone classification
- A bottom-zone cycle begins at a completed daily close at least 30% below the running all-time closing high and resets only after recovery to within 20% of that high.
- Capitulation marks a completed close at least 50% below the running high; Deep Drawdown means the cycle is active without completed stabilization.
- Stabilizing is assigned only on the first completed close above the 20-day EMA when that EMA is above its level five sessions earlier and RSI(14) is at least 45.
- Early Recovery and Retest describe the post-stabilization close relative to the completed 20-day EMA; no eventual low or forward outcome enters the classification.
price structure
Point-in-time price structure
- Prior 20-day and 50-day high and low levels exclude the current daily candle; no future-confirmed pivot is used.
- Intact moves to Neutral after a daily close below the prior 20-day low.
- Broken requires a completed weekly close below the support known at the start of that week and at least two daily closes below the same level during that week.
- The support captured on the first transition to Broken becomes the immutable cycle origin and is retained only for history and audit context.
- The same initial support becomes the active recovery level; later continuation breakdowns can ratchet it downward only when the production breakdown confirmation rule is satisfied.
- Broken moves to Improving only after two consecutive completed weeks without a new 20-day low and either a daily close above the prior 20-day high or a completed weekly close at least 0.25% above the previous completed week high.
- Improving moves to Intact only after two consecutive completed weekly closes above the active recovery level.
- A newly confirmed structural breakdown returns Improving to Broken and may ratchet the active recovery level lower without starting a new cycle.
- Broken never expires with time; both cycle levels reset only after structure returns to Intact.
regime gating
Regime gating and volatility
- Low maps to Caution with Intact or Neutral structure, Bearish with Broken structure, and Repair with Improving structure.
- Mixed maps to Bearish with Broken structure and Repair with Improving structure; Intact upgrades from Caution to Repair only when score momentum is Rising.
- Supportive maps Broken or Neutral structure to Repair, Improving to Repair or Early Bullish based on score momentum, and Intact to Early Bullish or Bullish based on trend confirmation.
- Strong maps Broken to Repair, Improving or Neutral to Early Bullish, and Intact to Bullish; Strong Trend additionally requires score at least 86, confirmed trend, and Expanding volatility.
- Volatility is Compressed at an ATR ratio of 0.8 or less, Expanding at 1.2 or more, and Normal otherwise, using 20-day ATR percent versus its trailing 50-session median.
statistics
Statistical integrity
- Every study keeps raw events and selects the earliest chronological non-overlapping population independently for each horizon.
- Confidence is based on independent N: Very Low below 10, Low 10–29, Medium 30–99, and High at 100 or more.
- Temporal validation requires N=8 and uses a chronological 70/30 holdout plus expanding-window folds.
- Sensitivity changes one declared detector parameter at a time and never selects the canonical rule from returns.
- Cross-exchange validation replays each detector separately and never pools venue populations.
- The low-score structure study selects global chronological non-overlap per horizon before partitioning intact and broken observations.
low score structure study
Low Opportunity Score by Market Structure
- An event occurs on entry to Opportunity Score 40 or below, or when the structure group changes during that low-score spell.
- Intact-group events include Intact, Neutral, and Improving states; Broken-group events require the contemporaneous Broken state.
- The study reports 7D, 30D, 90D, and 180D return, win rate, MAE, MFE, path drawdown, follow-through, and sample count.
- Follow-through requires a positive endpoint return and at least 5% maximum favorable excursion inside the completed forward window.
rsi matrix
Predeclared RSI research
- Daily RSI thresholds are fixed at 65, 70, 75, and 80; weekly thresholds are fixed at 45, 50, 55, and 60.
- Global non-overlap is selected per threshold and horizon before point-in-time regime segmentation.
- Missing point-in-time regime history remains Unclassified and no cell is interpreted below independent N=8.
- Threshold variants may share source events and are descriptive comparisons rather than independent experiments.
RELEASE EVIDENCE
Immutable study snapshots
These are the results first published under this model, not today's rolling study values. The JSON release record includes their complete filters, results, and methodology.