BnbSpotDailyDrawdownAccumulationLong

BNB BINANCE_SPOT Daily Drawdown-Accumulation Long-Only (Spot Venue Quota Fill, Single-Instrument, Defined Recovery Exit, 3-Parameter, Anti-Buy-Hold Guards)

A LONG-ONLY, single-instrument, DAILY strategy on BNBUSDT.BINANCE_SPOT that accumulates deep drawdowns from a rolling high and exits on recovery. This extends the ONLY two classes still proven alive this session — drawdown-accumulation (BTC-daily sibling reached PAPER at Sharpe 3.33; ETH-spot variant submitted this session) and cointegrated pairs (ETH-BTC submitted) — onto a fresh spot major, directly per the analyst's standing recommendation: 'a spot long-only edge needs genuine UPSIDE-ASYMMETRY (e.g., the proven drawdown-accumulation BTC-daily class on a spot major), not dip-fading.' It targets the still-unfilled BINANCE_SPOT venue bucket (11.2% actual vs 15% target) — the ONLY under-represented venue that is reliably testable, since HYPERLIQUID is data-capped, COIN-M is engine-blocked, options/carry/cross-venue are tooling-blocked, and funding data terminates ~Jan-2023. It explicitly avoids every dead lane confirmed this session: NOT VWAP-excursion MR (class halted 0-for-7 with the inverted-payoff signature — this is the OPPOSITE polarity: it buys deep multi-month discounts and holds for the recovery, with avg_win STRUCTURALLY > avg_loss because winners ride tens-of-percent recoveries while the catastrophic stop caps losers); NOT funding/premium/L-S-supp; NOT a sector-analogy pair. BNB is chosen deliberately over the failed XRP target: XRP spent long stretches in secular DECLINE (why dip-fading bought falling knives there), whereas BNB has one of the strongest recovery profiles in crypto — persistent structural demand (exchange token / BNB Chain) has caused every major BNB drawdown (2018, 2020-covid, 2021-May, 2022-FTX, 2024) to recover toward new highs, which is exactly the upside-asymmetry this mechanism monetizes. It is fully testable: single-instrument long-only spot directional PnL on ample multi-year data. It engineers around the documented drawdown-accumulation failure modes: DAILY bars (not 4H — the SOL-4H/basket variants starved on rare events at sub-daily resolution); SINGLE asset (not a basket — avoids the correlated-drawdown 'diversification' fallacy); a DEFINED RECOVERY EXIT plus moderate threshold and bounded scale-ins to force MULTIPLE completed round-trips (prevents the buy-and-hold degeneration / single-position end-of-data MTM artifact that the analyst flagged, where the optimizer chose 3-trade/693-day-hold params with metrics_reliable=false). Exactly 3 tunables (dd_threshold, high_lookback_days, recovery_exit_pct); all infra params (scale-in count, catastrophic stop, time stop, min_notional) FIXED and bounded — not in the Optuna search space.

Extends the proven drawdown-accumulation class (BTC-daily sibling reached PAPER at Sharpe 3.33) onto a fresh spot major, per the analyst's recommendation that a spot long-only edge needs genuine upside-asymmetry rather than dip-fading. It fills the reliably-testable BINANCE_SPOT venue bucket and avoids the halted VWAP-excursion family -- this is the opposite polarity (buy deep multi-month discounts, hold for recovery), so avg_win is structurally larger than avg_loss because winners ride tens-of-percent recoveries while the fixed wide catastrophic stop caps the rare loser; BNB's persistent structural-demand recovery profile gives the high win-rate the edge relies on (and is why BNB is chosen over secular-decline names like XRP). Venue routing is correct: long-only, no leverage, no funding edge -> BINANCE_SPOT (CASH, leverage clamped to 1), so only OrderSide.BUY entries are issued and exits sell the position. Bounded scale-ins require adding to an open position, which the base flat-only should_enter path cannot do, so should_enter/should_exit/position_size are made inert and the lifecycle runs in _manage via _submit_entry / _submit_exit_position. Tranches are sized off a capital snapshot taken when flat because get_account_equity() returns the USDT balance, which falls after each spot buy -- snapshotting keeps the scale-ins equal-dollar. The documented failure modes are engineered around: DAILY bars (not 4H, which starved on rare events), SINGLE asset (not a correlated basket), and a defined recovery exit + fixed catastrophic stop + ~1y time stop that force MULTIPLE completed round-trips, preventing the buy-and-hold degeneration / single-position end-of-data MTM artifact (metrics_reliable=false) the analyst flagged. The returned signal is the continuous drawdown (varies every bar -> non-frozen on real data; computed over min(available, lookback) so it is valid even in Layer-2's short synthetic scenarios). Exactly three tunable parameters (dd_threshold, high_lookback_days, recovery_exit_pct); SCALE_IN_COUNT, DD_STEP, TRANCHE_PCT, CATASTROPHIC_STOP, MAX_HOLD_BARS and min_notional are fixed constants kept out of the Optuna search space, closing the optimizer-escape hole. min_bars_required is kept low (20) with the rolling high computed over the available window so the strategy is robust in both multi-year backtests and short synthetic scenarios.

BNB spot drawdown-accumulation already failed the full optimization this session, the headline metrics are artifacts, and the core premise is contradicted by the data. DIRECT EVIDENCE: the earlier BNB spot drawdown-accumulation (a trade-richer 51-trade version) went through 3-phase optimization and was abandoned with DSR 0.406, all three walk-forward OOS windows exactly [0,0,0] (the optimizer raised the drawdown threshold and produced ZERO out-of-sample trades), and a degenerate holdout; this daily-with-scale-ins version is a near-duplicate on the same asset and mechanism class, so a re-run would hit the identical wall. METRICS ARE ARTIFACTS: Sharpe 5.88 is a per-period-attribution distortion (annualized_volatility 268% from booking 68-day-hold PnL onto single exit days); the honest figures are modest (PF 1.85, max DD 51%, CI high 64%). PREMISE CONTRADICTED: the hypothesis claims winners structurally exceed losers, but avg_loss ($74.9k) is 2.2x avg_win ($33.5k) with return_skew -1.15 — the 40% catastrophic stop and time-stop produce large losers (2022 -25.8%, 2026 -19.6%), so the upside-asymmetry mechanism does not hold. The sample is also thin (36 trades / 8.5 years) and bull-recovery-concentrated (2020 +74%, 2021 +75%). The drawdown-DCA-on-alt family is 0-for-alts post-optimization this session (SOL DSR 0.0008, BNB zero-OOS, LINK degenerate artifact); only the BTC sibling reached paper. Not iterate — no 3-parameter config turns BNB's handful of bull-recovery drawdowns into a multiple-testing-significant edge when the prior, trade-richer version already produced zero OOS trades. FAILURE PATTERN: single-asset daily drawdown-accumulation re-attempted on an alt that already failed the gates (BNB) yields the same thin, bull-recovery-concentrated, artifact-Sharpe profile with a contradicted upside-asymmetry (avg_loss > avg_win from the catastrophic stop); the mechanism only clears optimization on BTC.

Long-only daily drawdown-accumulation on BNBUSDT.BINANCE_SPOT. Each day it computes the drawdown of the close from the rolling high over high_lookback_days bars. It buys the first equal-dollar tranche when drawdown >= dd_threshold and adds deeper tranches (bounded to 3) as drawdown deepens by fixed 8% steps, sizing each tranche at 20% of the capital snapshot taken when flat (so tranches stay equal-dollar on a CASH account). It sells the entire position on the earliest of: a recovery target (close >= avg_entry x (1 + recovery_exit_pct)), a fixed wide catastrophic stop (close <= avg_entry x 0.60), or a ~1-year time stop. Long-only spot, CASH account, 1x. The base entry/exit hooks are inert; the accumulate/exit lifecycle is fully managed internally.

Clean execution (36 round-trips, 0 dropped); positive in 8 of 10 years including recent (2024 +22%, 2025 +11%); scale-in design less degenerate than the ETH 10-trade variant

Sibling mechanism (BTC spot drawdown-DCA) reached paper at Sharpe 3.33; healthy capacity

BNB spot drawdown-accumulation ALREADY failed optimization this session (prior 51-trade version: DSR 0.406, all-zero walk-forward OOS, degenerate holdout) — same asset+mechanism, re-run hits the same wall

Artifact metrics: Sharpe 5.88 with annualized_vol 268% (per-period attribution on 68-day holds); real max DD 51% (CI high 64%)

Upside-asymmetry premise contradicted: avg_loss $74.9k is 2.2x avg_win $33.5k (skew -1.15) — the 40% catastrophic stop produces large losers (2022 -25.8%, 2026 -19.6%)

Thin (36 trades / 8.5 years), bull-recovery-concentrated (2020 +74%, 2021 +75%); family 0-for-alts post-opt (SOL/BNB/LINK), only BTC worked

BnbSpotDailyDrawdownAccumulationLong extended the only two still-alive mechanism classes to a fresh spot major, porting the paper-stage BTC drawdown-accumulation onto BNB with scale-ins and a defined recovery exit, deliberately chosen over the declining XRP for BNB's strong recovery profile. Its headline numbers looked strong (Sharpe 5.88, +450%), but they were per-period-attribution artifacts (268% annualized vol) on a thin, bull-recovery-concentrated 36-trade sample, and the founding upside-asymmetry premise was contradicted by avg losses 2.2x avg wins from the catastrophic stop. Decisively, a trade-richer BNB version had already failed 3-phase optimization this session with all-zero OOS windows and a degenerate holdout, so the analyst abandoned this near-duplicate before optimization. It ended after one iteration as abandoned, never advancing to optimization or risk review.

Re-attempting single-asset daily drawdown-accumulation on an alt that already failed the gates (BNB) yields the same thin, bull-recovery-concentrated, artifact-Sharpe profile with a contradicted upside-asymmetry — a wide catastrophic stop makes avg_loss exceed avg_win — and the drawdown-DCA-on-alt family is 0-for-alts post-optimization (SOL, BNB, LINK) with only the BTC sibling clearing, so no 3-parameter config turns a handful of bull-recovery drawdowns into a multiple-testing-significant edge.

It was abandoned at the pre-optimization backtest-review gate (verdict: abandon) because a trade-richer BNB spot drawdown-accumulation had already failed full optimization this session (DSR 0.406, all three walk-forward OOS windows [0,0,0], degenerate holdout), the headline Sharpe was a per-period-attribution artifact, and the core upside-asymmetry premise was refuted by avg_loss > avg_win — so re-running this near-duplicate would hit the same wall and optimization/later stages were never reached.

A long-only single-instrument daily drawdown-accumulation strategy on BNBUSDT.BINANCE_SPOT (OHLCV-only) that bought deep discounts from a rolling 180-day high (first tranche at 20% drawdown, up to 3 scale-in tranches at deeper levels) and sold all on a 25% recovery above average entry, a 40% catastrophic stop, or a 1-year time stop — a port of the paper-stage BTC-daily drawdown-accumulation sibling to a fresh spot major chosen for its strong recovery profile.

Across 36 round-trips over ~8.5 years it posted a headline Sharpe 5.88, 80.6% win rate, +450.9% return and profit factor 1.85, but the metrics were artifacts (annualized vol 268% from per-period attribution on 68-day holds; honest max drawdown 51%, CI high 64%), the result was bull-recovery-concentrated (2020 +74%, 2021 +75%), and the upside-asymmetry premise was contradicted — avg loss $74.9k was 2.2x avg win $33.5k (skew -1.15) as the 40% stop produced large losers (2022 -25.8%, 2026 -19.6%).