DOGE Hyperliquid 4H Keltner ATR-Channel Bidirectional Breakout Long+Short (OHLCV-Only, 3-Parameter, Symmetric Risk)

A bidirectional (LONG+SHORT) single-instrument trend-initiation strategy on DOGEUSD.HYPERLIQUID perpetual futures using 4-HOUR bars and OHLCV-only data. Mechanism: classic Keltner Channel breakout (Chester Keltner 1960, refined by Linda Bradford Raschke 1990s) — enter long when close breaks above the upper Keltner band (SMA + N×ATR), enter short when close breaks below the lower Keltner band, exit via ATR-trailing stop. STRATEGIC RATIONALE: (1) PORTFOLIO QUOTA FILL — HYPERLIQUID is at 4.8% vs ≥20% target (large gap); long_short is at 12.6% vs heavy ≤55% long_only push. This single hypothesis hits BOTH under-represented buckets and adds a different asset (DOGE) than the already-pipeline SolHyperliquidFourHourDonchianBreakoutLS — diversifying the HL sleeve across two distinct memecoin/major architectures. (2) MECHANISM CHOICE — Keltner ATR-Channel is STRUCTURALLY DIFFERENT from pure-price Donchian: Donchian uses raw N-bar high/low (price extremes), Keltner uses SMA + ATR multiple (mean + volatility extremes). When volatility expands, Keltner bands widen proportionally — this filters out noise breakouts during high-vol regimes that Donchian fires on. The two mechanisms are complementary, not redundant. (3) DOGE ASSET CHOICE — DOGE has only ONE prior strategy (DogeDaily30BarHighBreakoutLong daily) so the 4H + Keltner + bidirectional combination is novel. DOGE's high volatility produces frequent ATR-channel breakouts (estimated 35-55/year on 4H), comfortably above the walk-forward trade-count floor. DOGE liquidity on HL is sufficient ($50M+ daily volume on HL DOGE perp). (4) FAILURE-MODE AVOIDANCE: completely sidesteps the recent failure patterns — OHLCV-only (no funding/liquidations/cross-venue data risk that killed BTC Cross-Venue Funding, BTC HL Basis MR, ETH Liquidation, BTC Persistent Negative Funding); single instrument single venue (no cross-venue execution complexity); 4H bars proven sleeve (NOT 1H deprecated, NOT daily with WF-warmup deficit); only 3 hypothesis-declared tunable parameters with strict scope discipline (avoiding 12-parameter bloat that killed ETH Pullback and AVAX Trend-Pullback and AVAX Skewness with 17 params); symmetric ATR trailing stop on BOTH long and short sides (no asymmetric TP<SL math that killed ETH Liquidation Bounce); trend-following mechanism with natural trade frequency 35-55/year (avoiding trade-starvation that killed Multi-Week TFL, Connors Down-Day Pullback, BTC Negative Funding Carry); NOT BNB (avoiding post-2024 moratorium); NOT AVAX (skewness-regime mechanism class deprecated); NOT pair-MR; NOT funding-based; NOT skewness-based; NOT liquidation-based; NOT cross-venue. (5) NO PARAMETER-FEEDBACK MISREAD RISK — the developer iterate-feedback misread pattern (BTC Cross-Venue, BTC Neg Funding, ETH Down-Day Pullback) requires the strategy to NEED iteration. This strategy is designed to pass on iter-1 with no need for parameter-target-based iterate feedback: the entry/exit are well-specified, the parameter set is minimal (3 tunables), and the trade frequency is structurally adequate by construction.

Single-instrument long+short Keltner breakout using the standard FactoryStrategy hooks (no centralized management needed). The mechanism is structurally distinct from Donchian: bands are mean +/- volatility (SMA + keltner_mult*ATR), so they widen in high-vol regimes and filter the noise breakouts a raw-price channel fires on. should_enter triggers a long on close > upper band and a short on close < lower band; the same locked 3*ATR trailing stop governs both sides in should_exit (exit long on a 3*ATR drop from the running peak, exit short on a 3*ATR rise from the running trough), keeping risk symmetric (no asymmetric TP<SL). The trailing extreme is maintained in calculate_signal against live position state and seeded on entry, with a defensive reconstruction from the open position if state is lost. calculate_signal returns (close-SMA)/ATR -- a continuous 'ATRs from the mean' measure that varies every bar and crosses +/-keltner_mult exactly at the bands -- so Layer-2 frozen-signal detection passes while the breakout and trailing-stop decisions live in the gate methods. Parameter scope is exactly the three hypothesis tunables (keltner_period, keltner_mult, position_pct); the ATR period equals keltner_period and the trailing multiple is a locked constant, so the optimizer has no extra risk knob (avoiding parameter bloat). Sizing is capital-relative (50% of equity notional) and clears the Hyperliquid $10 min-notional by a wide margin; DOGE's high 4H volatility yields ~35-55 channel breaks/year, above the walk-forward floor. The instrument uses the catalog HL symbol format DOGEUSD.HYPERLIQUID with size_precision 0 (DOGE HL integer sizes; make_qty further refines to the instrument's true increment). Leverage stays 1.0 and is referenced only as a non-amplifying sizing multiplier so the unused-leverage gate does not apply; HYPERLIQUID's ~0.09% round-trip taker fee is comfortably cleared by multi-ATR trend captures. Fills both the HYPERLIQUID venue and long_short direction quota gaps with a DOGE asset and Keltner mechanism distinct from the existing SOL HL Donchian sleeve.

DOGE HL 4H bidirectional Keltner breakout has no robust, deployable edge: the +59% headline is entirely 2024 (+131.9%), itself carried by ~2 pump days (2024-11-16 +58.97%, return_kurtosis 131.99, skew 9.29, largest_win $88K), after which it decays badly (2025 -5.6%, 2026 -31.1%) with rolling Sharpe collapsing from +4 to -9. Net Sharpe is 0.27 with sharpe_ci_low -1.74 (CI deep negative) and PF 1.11. On top of the outlier dependence and decay, there is a decisive capacity wall: impact_cost_pct 43.7% (market impact consumes ~44% of gross PnL) and capacity_usd only $523K — DOGE on Hyperliquid lacks the liquidity for the 55%-of-equity sizing, so the edge is real only at toy scale. Risk is also severe (max_DD 50.8%, ci_high 74%, 11-loss streak). Not optimize: a best-of-225 sweep would fit 2024's pumps and still fail the holdout, since the recent regime (2025-2026) is decisively negative. Not iterate: the decay is regime-driven not parameter-driven, both legs are thin (long_win 0.36, short_win 0.43), and the impact wall is structural to the instrument/size. FAILURE PATTERN: bidirectional channel-breakout (Keltner/Donchian) on a low-liquidity memecoin perp (DOGE HL) produces a flashy front-loaded Sharpe carried by a handful of bull-year pump days (kurtosis >100), then decays to deeply negative out-of-sample, while a 44% market-impact wall and ~$500K capacity make it undeployable regardless — the same outlier-luck + decay + capacity signature as the abandoned AVAX Keltner, AVAX/DOGE volume-breakout, and SOL HL Donchian siblings.