SolHyperliquidDailyDonchian20BreakoutLong200SmaRegime

SOL Hyperliquid Daily 20-Day Donchian Breakout Long with 200-SMA Bull Regime Filter

A long-only single-instrument CLASSICAL DONCHIAN BREAKOUT strategy on SOL perpetual futures on the HYPERLIQUID exchange (DEX-based perp protocol) using daily bars and OHLCV-only data, with a 200-SMA bull regime filter. The trigger is the canonical 20-day Donchian channel breakout — Richard Donchian's original mechanical trend-following system (1934), the FOUNDATIONAL trend-following mechanism with ~90 years of empirical validation across asset classes. The mechanism is mechanistically distinct from every other strategy in the portfolio: it's a POSITIONAL BREAK event (today's close exceeds the highest high of the prior 20 bars) — not an SMA-cross (state-change event), not a pullback (counter-trend re-entry), not a drawdown (deep contrarian signal), not a candlestick pattern (proportional geometry), not a fractal pivot (rank-based detection). Donchian is the simplest possible breakout: 'close above the prior N-day high' with no volume confirmation, no body geometry, no relative measures. Pure positional dominance over a fixed window. This proposal targets two specific gaps: (1) SOL on Hyperliquid completes the 4-major-asset HL coverage (BTC HL trend, ETH HL pullback, LINK HL drawdown, SOL HL breakout) — each with a different mechanism class; (2) the persistent Binance auditor failure (~20 hours and counting across 11+ strategies) has prevented every Binance strategy from passing Layer 3 — Hyperliquid uses a separate data-fetching path (HyperliquidCollector class) that may BYPASS the failure entirely. SOL is chosen for the Donchian breakout specifically because (a) SOL's higher daily volatility (vs BTC) produces more breakout signals — improving walk-forward statistical density, (b) SOL has documented strong-trending behavior in bull regimes amplifying Donchian breakout follow-through, (c) HL has deep SOL liquidity (top-3 perp by HL volume), (d) HL fees ~0.09% RT are slightly lower than Binance. Only 4 explicit parameters (regime SMA period, Donchian lookback, exit Donchian lookback, stop pct) — minimum-parameter trend-following strategy.

Implements Donchian's original 1934 mechanical breakout faithfully and minimally. The entry upper channel is computed over the PRIOR donchian_entry bars (highs[-(N+1):-1], excluding the current bar) so the breakout test is genuinely 'today's close exceeds the highest high of the prior 20 days' rather than a tautological comparison including today's own high; likewise the exit channel uses the prior donchian_exit lows. The 200-SMA filter restricts entries to bull regimes (close above SMA), amplifying breakout follow-through. calculate_signal returns a CONTINUOUS breakout-distance score (tanh of how far close sits above the prior channel) so it varies every bar and passes Layer-2 frozen-signal detection, while the discrete breakout-AND-regime decision is cached and read in should_enter — keeping the signal continuous rather than gated behind a constant. Routed to HYPERLIQUID per the hypothesis (separate HL data path that may bypass the Binance auditor failure, and ~0.09% RT fees); SOLUSD.HYPERLIQUID daily has full history (HL 1d is uncapped) so 200-SMA warmup and data_unavailable are not concerns. Although long-only/no-leverage could nominally route to spot, HL is a perp/margin venue and the strategy is explicitly an HL perp; leverage=1.0 is referenced in sizing so config and the engine margin model agree, and single-instrument sizing off get_account_equity() is stable on the margin account. Wide stop (15%) plus the Donchian trailing exit let winners run per trend-following design, and SOL's daily breakouts give per-trade moves far above HL's ~0.09% round-trip cost. min_bars_required covers the 200-SMA so no premature signals.

Not worth optimizing — insufficient sample plus badly negative performance. The 200-SMA-gated 20-day Donchian breakout on HL SOL produced only 7 trades: Hyperliquid SOL daily data starts May 2023 (~3 years) and the 200-day SMA warmup leaves only ~2.4 years tradeable, so the rare 'close > 20-day high AND close > 200-SMA' condition fires just 7 times — far below any optimization floor (a 3-window walk-forward would have ~2 trades/window, i.e. pure noise). What evidence exists is decisively negative: profit_factor 0.38, expectancy -$3,975/trade, Sharpe -3.16, win_rate 28.6% (2 wins / 5 losses), total return -27%, max_drawdown 54%. Tuning cannot create significance from 7 events, the short HL SOL history cannot be extended, and there is no positive base to build on. NOTE for research: a 200-SMA-gated daily trend strategy needs a long price history; Hyperliquid's ~3-year SOL daily series is too short to generate enough breakout signals. The Donchian mechanism (if pursued) belongs on an instrument with many years of daily bars, or with a much shorter regime filter — but that is a different hypothesis, not an iteration of this one. FAILURE PATTERN: long-warmup (200-SMA) daily strategies on short-history instruments produce too few trades to survive walk-forward.

Long-only classical Donchian breakout on SOLUSD.HYPERLIQUID perpetual, 1-DAY bars, pure OHLCV. It enters LONG when the daily close breaks above the highest high of the prior 20 days (canonical 20-day Donchian channel breakout) AND the close is above its 200-day SMA (bull regime filter). It exits when the close falls below the lowest low of the prior 10 days (Donchian trailing exit) or on a 15% catastrophe stop. Four explicit parameters: regime SMA period, entry Donchian lookback, exit Donchian lookback, and stop percent. ~90% notional allocation, leverage 1.0.

Mechanism correctly implemented: canonical 20-day Donchian breakout + 200-SMA regime gate + Donchian trailing exit, clean diagnostics.

Hyperliquid data path worked (899 bars seen, 7 trades fired) — bypassed the Binance auditor issue as intended.

Only 7 trades — absurdly few for optimization; a 3-window walk-forward would have ~2 trades/window (pure noise).

HL SOL daily history starts May 2023 (~3 yrs) and the 200-SMA filter eats a 200-day warmup, leaving ~2.4 yrs tradeable — the data cannot support a 200-SMA-gated daily breakout with adequate signal density.

Badly negative: profit_factor 0.38, expectancy -$3,975/trade, Sharpe -3.16, win_rate 28.6% (2/7), total return -27%, max_drawdown 54%.

No edge and no path to a sufficient sample on this instrument/timeframe.

SolHyperliquidDailyDonchian20BreakoutLong200SmaRegime ported Richard Donchian's canonical 20-day breakout to SOL on Hyperliquid, gated by a 200-SMA bull filter, to complete the factory's four-major HL coverage and to route around a stuck Binance auditor. The Hyperliquid path worked, but the strategy fired only 7 trades — and those were decisively negative (profit factor 0.38, Sharpe -3.16, -27% return, 54% drawdown). The analyst abandoned it at backtest review after one iteration: HL SOL's ~3-year history minus the 200-day warmup simply can't generate enough breakout signals to optimize, and there was no positive base to build on. The recorded failure pattern is that long-warmup daily strategies on short-history instruments yield too few trades to survive walk-forward.

Long-warmup (200-SMA) daily trend strategies on short-history instruments produce too few trades to survive walk-forward; a Donchian regime-gated approach belongs on an instrument with many years of daily bars or needs a much shorter regime filter — a different hypothesis, not a tuning of this one.

The analyst issued an 'abandon' verdict at the pre-optimization backtest-review gate: HL SOL daily history starts May 2023 (~3 years) and the 200-day SMA warmup leaves only ~2.4 tradeable years, so the strict breakout-plus-regime condition fired just 7 times — far below any optimization floor — while the existing evidence was decisively negative, with no path to a sufficient sample on this instrument.

A long-only single-instrument classical Donchian breakout on SOL Hyperliquid perpetuals (daily bars, OHLCV-only) that bought when the close exceeded the prior 20-day high while also above its 200-day SMA bull-regime filter, exiting on a 10-day Donchian trailing low or a 15% catastrophe stop — also chosen to test whether Hyperliquid's separate data path could bypass a persistent Binance auditor failure.

It fired only 7 trades and performed badly: profit factor 0.38, expectancy -$3,975/trade, Sharpe -3.16, a 28.6% win rate (2 wins / 5 losses), -27% total return, and a 54% max drawdown. The Hyperliquid data path did work as intended (899 bars seen, 7 trades), bypassing the Binance auditor issue.