BtcDailyThreeWhiteSoldiersMomentumContinuationLong

BTC Daily Three White Soldiers Momentum Continuation Long with 200-SMA Bull Regime Filter

A long-only single-instrument 3-BAR CANDLESTICK CONTINUATION-PATTERN strategy on BTCUSDT perpetual futures using daily bars and OHLCV-only data, with a 200-SMA bull regime filter. The trigger is the classic 'Three White Soldiers' pattern documented by Steve Nison (Japanese Candlestick Charting Techniques, 1991). Three White Soldiers is a SPECIFIC 3-BAR TEMPORAL SEQUENCE indicating SUSTAINED buying pressure across three consecutive trading sessions: 3 consecutive bullish candles where each closes near its high, each opens within the prior bar's body (no gaps signaling exhaustion), and each closes higher than the prior bar's close. This is mechanistically NOVEL within the portfolio AND COMPLEMENTARY to BnbDailyMorningStarReversalLong: Morning Star is a 3-bar REVERSAL pattern (bear → indecision → bull), while Three White Soldiers is a 3-bar CONTINUATION pattern (sustained bull). Together they cover the two primary 3-bar pattern archetypes. Three White Soldiers captures a fundamentally different microstructure than any existing portfolio strategy: it's NOT a regime gate (Golden Cross), NOT volume-confirmation (Vol Breakout), NOT volatility-compression (Inside Day), NOT a 1-bar rejection (Hammer), NOT a 2-bar reversal (Engulfing), NOT a count-based event (consecutive-down-day), NOT a continuous-statistic extreme (RSI/BB/VWAP), NOT a drawdown depth. It's a temporal momentum-CONFIRMATION pattern requiring 3 specific bars in sequence to show sustained directional commitment with no exhaustion gaps. The economic mechanism: each successive bullish close above the prior close represents fresh buyers entering the market AND prior shorts being forced to cover; the requirement that each bar opens WITHIN the prior body (not gapping up) eliminates parabolic-blow-off setups that frequently fail. BTC daily is chosen because (a) BTC has zero daily-timeframe strategy currently — this fills the asset+timeframe gap (BTC's existing 2 strategies are 4H Vol Breakout and Spot Drawdown), (b) BTC daily futures data path is empirically used by multiple in-pipeline strategies' regime filters and has been promoted for ETH Golden Cross etc., (c) BTC has the deepest institutional flow making 3-bar momentum patterns statistically more meaningful than on smaller-cap altcoins, (d) candlestick patterns have their canonical Nison-tested foundation at the daily timeframe. Only 3 explicit parameters (regime SMA period, body-to-range ratio threshold, stop-loss pct) — minimum-parameter-count for a 3-bar pattern.

Iteration 2 fix (Layer 3 — no trades over 2317 bars): the pattern was over-constrained so it never triggered. Made the smallest loosening changes that keep the Three White Soldiers economic meaning intact: (1) decoupled the 'close near high' upper-wick test from body_ratio — it previously required upper wick <= (1-body_ratio)*range = 0.4*range at body_ratio 0.6, which combined with body>=0.6*range left almost no feasible candles; now a fixed 40%-of-range tolerance; (2) relaxed the anti-exhaustion open test from the two-sided 'open within prior body' to the single essential condition 'no gap-up above prior close', dropping the restrictive lower bound; (3) removed the redundant extra 4th-bar (c1>c0) rising requirement, keeping the 3-bar rising-close chain; (4) lowered default body_ratio from 0.6 to 0.5. Entry gate structure, regime filter, sizing, and all exit logic are unchanged, so earlier-passing layers do not regress.

Signal too rare for single-asset daily optimization (not a data wall — full BTC daily history of 2,369 bars loaded; not a code bug — the pattern is implemented correctly and the 2 signals genuinely match). The strict Three White Soldiers definition (3 consecutive strong-bodied bullish candles closing near their highs, rising closes, no gap-up on bars 2 & 3) plus a 200-SMA bull filter fired only 2 times in ~6.4 years, and both trades lost (win_rate 0.0, PF 0.0, metrics_reliable=false). 2 trades cannot support a 3-window walk-forward + 15-day holdout — the optimizer would fit pure noise. This is the factory's recurring rare-single-asset-daily-chart-pattern outcome (cf. double-top/bottom and BB lower-band-bounce abandons): the sample is structurally too small, and it is NOT fixable by parameter tuning within this hypothesis — loosening body_ratio or dropping the anti-exhaustion no-gap/close-near-high constraints enough to reach ~60+ trades would destroy the very pattern the hypothesis is testing (it becomes a generic 'three up days in an uptrend' filter with no distinct edge). RESEARCH-LEAD REFRAME (not a developer parameter change): the only way to make a 3-bar continuation candlestick pattern statistically testable is to run it as a MULTI-ASSET BASKET across many liquid symbols simultaneously, pooling entries to reach the trade floor while keeping the pattern strict — a single-asset daily instance of any strict 3-bar pattern will always be sample-starved. Recommend abandoning this single-asset BTC-daily instance and, if the mechanism is still desired, re-proposing it as a cross-sectional basket.

Long-only BTCUSDT daily Three White Soldiers (Nison) momentum-continuation strategy with a 200-SMA bull regime filter. Enters long when three consecutive bullish daily candles each have a strong body (>= body_ratio of range), close near their high (upper wick <= 40% of range), form a rising close sequence, and open with no gap-up above the prior close (anti-exhaustion) — and price is above the 200-SMA. Exits via Donchian trailing-low stop, regime break (close < SMA), or a hard stop_loss_pct; no fixed take-profit so winners run. OHLCV-only, leverage 1.0, three tunable params.

Verification failed (Layer 3 — sandbox backtest): No trades produced Bar type used: BTCUSDT.BINANCE-1-DAY-LAST-EXTERNAL, Bars processed: 9398 Diagnostics: should_enter() returned a side 0 times over 2317 evaluated bars -> your ENTRY CONDITION never triggered. Loosen the entry logic / thresholds. Ensure your strategy produces trades with the given data and parameters.

Code is correct and the mechanism is faithfully implemented: strict Three White Soldiers detection (3 strong bullish bodies, close-near-high, rising closes, anti-exhaustion no-gap on bars 2 & 3) gated by a 200-SMA bull regime; the 2 entries that fired genuinely match the pattern.

Clean single-instrument BTC daily design on a fully-available data path (2,369 bars loaded, no data wall), sensible risk-based sizing and multi-exit (hard stop, regime break, Donchian trail).

SIGNAL TOO RARE: only 2 entries in 2,167 evaluated bars (~6.4 years). The strict pattern + 200-SMA filter fires ~0.3 times/year — nowhere near the walk-forward sample floor (the hypothesis itself cited ~62 trades as the level that killed ADA daily; this is 2).

Both trades lost (win_rate 0.0, avg_win $0.0, profit_factor 0.0, expectancy -$1,800) and metrics_reliable=false — there is no edge to measure and nothing for an optimizer to fit but noise.

The rarity is intrinsic to the hypothesis's defining strictness (three consecutive gap-free strong bullish closes in an uptrend); relaxing filters enough to reach a statistical sample would dissolve the pattern's anti-exhaustion thesis into a generic '3 up days' filter.

This strategy faithfully implemented the Three White Soldiers continuation pattern on BTC daily futures with a 200-SMA bull filter, and the code was verified as correct — the 2 signals it produced genuinely matched the pattern. However, across ~6.4 years of full BTC daily history the strict definition plus regime filter fired only twice, and both trades lost (win rate 0.0, profit factor 0.0, expectancy -$1,800). At the backtest-review gate the analyst abandoned it because 2 trades cannot support a 3-window walk-forward and 15-day holdout, and loosening the filters enough to reach a viable sample would dissolve the pattern's defining anti-exhaustion thesis. The recommended path forward was to re-propose the mechanism as a cross-sectional multi-asset basket rather than a single-asset daily instance.

A strict multi-bar candlestick pattern on a single asset's daily chart is inherently too rare to reach a statistically testable trade count — such patterns need a multi-asset basket to pool enough entries without loosening the definition.

The analyst issued an 'abandon' verdict at the pre-optimization backtest-review gate: with only 2 trades (vs. a ~60-trade floor) the sample was structurally too small to support walk-forward optimization, so no optimization or later analyst stage was ever run.

A long-only BTCUSDT daily-futures strategy that entered on the classic 'Three White Soldiers' 3-bar bullish continuation pattern (three strong-bodied rising closes with no exhaustion gaps) gated by a 200-SMA bull-regime filter.

Over ~6.4 years (2,167 evaluated bars) the pattern fired only 2 entries, both of which lost, giving a 0.0 win rate, 0.0 profit factor, expectancy of -$1,800/trade, total return -3.41, and a Sharpe of -4.20 with metrics_reliable=false.