# Pre-breakout study v3 — fixed before results Purpose: identify an unbroken 20-bar range boundary, not claim to know the future. Requested Nov 16 2026. Compare full-position 5%, 10%, 20% favorable price moves independently, with identical entries. Never add these overlapping target cohorts together. Universe: existing 60-stock universe and 12 supported USD crypto pairs. Price-source coverage is reported; no inference of whole-market volume. Stock splits are normalized exactly once. Crypto uses completed UTC calendar days only; weekends count. Missing days are not forward-filled. Skip crypto signals without 210 consecutive preceding days and skip confirmation/trade windows crossing a missing daily bar. Prices and volume represent the provider's venue feed, not every crypto exchange. Fixed setup: daily trend has >=4 of 5 clues in one direction, close remains on the inside of the previous 20-bar high/low and <=1 ATR from that trigger. Recent 5-day high-low range is <=60% of previous 20-day range. RSI 45–70 for long, 30–55 for bearish. CMF supports side (>0 long/<0 short), 20-bar return beats/trails benchmark in that direction. Benchmark SPY for stocks, BTC for crypto (BTC compared with ETH to avoid self-comparison). Price/volume quality must have positive mean 20-bar volume and >=15 positive-volume bars out of 20. Wait up to five subsequent completed bars for a close beyond the frozen trigger with >=1.2x the signal-date prior-20 mean volume. Then enter next bar open only on correct side of trigger and within 0.5R, where R=1.5 signal-date ATR. Model stop from actual entry at 1.5 signal-date ATR; allow adverse gap fill at opening price. Independent targets 5%, 10%, 20% from actual entry, whole position each; stop first on ambiguous daily bars. Exit at 20-bar close if no target/stop. Stock friction 0.20% round trip / 0.50% sensitivity; crypto 0.60% / 1.00%, modeled assumptions, not a venue quote. Bearish crypto results are hypothetical directional short scenarios; source spot assets are not shortable and fees do not include funding/borrow. No within-symbol overlapping candidate waits or positions: advance beyond expiry or final exit of longest target cohort. A signal must have the full five-bar confirmation opportunity inside a period or is boundary-pending; target outcomes must finish within the same period. Awaiting triggers, expired signals, gap skips and unfinished trades are counted separately. Periods 2024, 2025 and 2026 through latest completed bar. This is a newly specified exploratory study on already-used history, not independent holdout evidence. No parameter tuning or live promotion based on results. Report every period/side/target denominator, target-hit fraction, mean return, cost sensitivity, worst loss and sample uncertainty. At least 30 completed observations are needed even to label a target hit-rate sample substantial; no calibrated future prediction is claimed. Cross references: stock-sector analyst-selected ETF proxies, stock vs sector and sector vs SPY returns, SPY/QQQ/DIA regimes, 60-stock sample breadth (not exchange-wide breadth), RSP/SPY participation proxy, IWM/SPY size proxy, HYG/LQD mixed credit/duration proxy, TLT bond-price proxy, GLD gold and UUP dollar proxy. Calculate dated prices; don't equate these relationships with a guarantee. These contextual features are reported separately and are not added as retroactively optimized filters. VIX, intraday VWAP, options skew/open interest, dealer gamma, crypto funding/open interest/order-book imbalance are not connected and must not influence scores. References explain concepts; they do not endorse this algorithm. AQR diversified long-horizon momentum evidence does not validate 5–20% short-horizon single-asset predictions.