EthSpot30DayDrawdownAccumulationRecoveryLong

ETH Spot 30-Day Drawdown Accumulation with Recovery Exit (Daily)

A long-only single-instrument accumulation strategy on ETHUSDT spot (BINANCE_SPOT) using daily bars and ONLY OHLCV data. Architecturally mirrors the proven BTC Spot 30-Day Drawdown Accumulation strategy (currently at paper_stage with Sharpe 3.33), applied to ETH spot for portfolio diversification. The strategy waits for ETH to draw down meaningfully from a recent 30-day high WHILE the long-term uptrend regime is intact (price still above the 200-day SMA), then enters long on the first daily up-close. Exits when price recovers back to its rolling 30-day high (winners ride the rebound), or when an ATR trail / regime filter fires (cuts losers fast). The architecture explicitly fixes both fatal flaws of the abandoned 'ETH Spot Capitulation Buy-the-Dip': (1) it ADDS a 200-day SMA regime filter (the abandoned version had none, so caught falling knives in 2018, 2022 bears), and (2) it INVERTS the broken exit asymmetry by using the recovery-to-peak as the take-profit (lets winners run to typical +15-25% recoveries) rather than RSI=55 (which capped wins at +3-5%). Spot venue means no funding cost, no liquidation risk, and CASH account compatibility. Daily timeframe with low signal frequency (~5-15 entries/year) keeps fees manageable even at the higher 0.20% spot RT cost.

Faithful clone of the paper-stage BtcSpotDrawdownAccumulationRecovery (Sharpe 3.33) applied to ETH spot, with the two explicit fixes the hypothesis requires versus the abandoned capitulation version: (1) a 200-day SMA regime filter so it never buys falling knives in 2018/2022 bears, and (2) an inverted exit asymmetry — the recovery-to-prior-peak take-profit (captured at entry as the 30-day high) lets winners ride the typical +15-25% rebound instead of an RSI=55 cap, while a 5%-below-200-SMA regime breakdown and a 60-bar time stop cut losers fast. I added the hypothesis's 'first daily up-close' confirmation: prev_close is snapshotted BEFORE the bar is appended, so cur_close > prev_close correctly compares today vs yesterday and the entry fires on the first up-close after the drawdown (no lookahead; lookback_high excludes the current bar). The strategy is correctly routed to BINANCE_SPOT: it is long-only, uses no leverage, the edge is organic spot accumulation (no funding/liquidation needed), and a single concurrent position means the CASH-account equity (sized once at entry from full cash) is clean. calculate_signal returns a continuous tanh(drawdown) score (regime-gated, varying) so Layer-2 frozen-signal detection passes while never spuriously entering; should_enter is BUY-only (never sells to enter, respecting CASH). ETHUSDT spot has deep multi-year daily history, avoiding data_unavailable. With ~5-15 entries/year and recoveries far exceeding the ~0.20% spot round-trip cost, the edge clears fees.

Not worth optimizing — the BTC drawdown-accumulation mechanism (paper-stage Sharpe 3.33) does not generalize to ETH spot: profit_factor 0.82, expectancy -$173/trade, Sharpe -0.91, total return -17.3% over 8.7 years, negative in 2019/2020/2021(-19.6%)/2022 (110-trade sample, so this is a fair read). The strategy also does not behave as the hypothesis describes: avg holding is 1d 3h, not the intended weeks-long ride to a +15-25% recovery — entries fire on near-complete V-recoveries that hit the 30-day-high exit within a day for small wins, while losers run to the regime cut (avg_win $1,540 < avg_loss $1,885), and the fast churn drives commissions to 11.37% of gross at spot's 0.20% RT cost. Whether viewed as ETH's dip dynamics not fitting the 'ride the recovery' design or as a calibration mismatch, the result is decisively negative and fee-fragile; optimizing would overfit a no-edge base. NOTE for research: the spot drawdown-accumulation mechanism is validated only on BTC; it has now failed on ETH and on four hyper-volatile alts (AVAX/AAVE/INJ/SOL). If salvage is attempted, the recovery-exit calibration is the place to look (it fires in ~1 day instead of riding for weeks), but as implemented there is no edge. Stop porting this mechanism to non-BTC instruments.

Long-only ETH spot accumulation on ETHUSDT.BINANCE_SPOT daily bars, OHLCV-only, mirroring the proven BTC Spot 30-Day Drawdown Accumulation Recovery strategy. It enters long when ETH is >= 15% below its prior 30-day high (current bar excluded) WHILE the 200-day uptrend regime is intact (close > 200-SMA) AND the current bar is the first daily up-close (close > prior close). It exits when price recovers to the 30-day high captured at entry (lets winners ride the rebound), when price breaks 5% below the 200-SMA (cuts losers in a bear), or after a 60-day time stop. Deploys 50% of cash per trade; BINANCE_SPOT CASH account (leverage 1, long-only by venue).

Mechanism mirrors a paper-stage BTC sibling (Sharpe 3.33); correctly adds a 200-SMA regime filter; clean diagnostics; long history (8.7 years).

No edge: profit_factor 0.82, expectancy -$173/trade, Sharpe -0.91, total return -17.3% over 8.7 years, negative in 2019/2020/2021/2022.

Does not behave as designed: avg holding 1d 3h vs the intended weeks-long 'ride to +15-25% recovery' — entries fire near-complete V-recoveries that hit the 30-day-high exit fast with small wins (avg_win $1,540 < avg_loss $1,885).

commission_pct_of_gross 11.37% — fast churn at spot 0.20% RT consumes a huge share of gross on a non-existent edge.

Same drawdown-accumulation family that has failed off-BTC on AVAX/AAVE/INJ/SOL — does not generalize to ETH.

EthSpot30DayDrawdownAccumulationRecoveryLong cloned a paper-stage BTC accumulation strategy (Sharpe 3.33) onto ETH spot, explicitly adding a 200-SMA regime filter and a recovery-to-peak exit to fix an earlier abandoned ETH dip-buyer. The mechanism ran cleanly over a long history but had no edge — profit factor 0.82, Sharpe -0.91, -17.3% over 8.7 years — and didn't even behave as intended, exiting in ~1 day for small wins while fees consumed 11.37% of gross. The analyst abandoned it at backtest review after one iteration, noting the BTC mechanism doesn't generalize and the recovery exit was miscalibrated. The recorded guidance is to stop porting drawdown-accumulation off BTC, where it has now failed on ETH and four hyper-volatile alts.

The spot drawdown-accumulation mechanism is validated only on BTC and has now failed on ETH plus four hyper-volatile alts (AVAX/AAVE/INJ/SOL); a recovery-to-prior-peak exit can also silently misfire (exiting in a day rather than riding weeks), so a sibling's paper-stage Sharpe doesn't transfer and the family shouldn't keep being ported to non-BTC instruments.

The analyst issued an 'abandon' verdict at the pre-optimization backtest-review gate: the BTC drawdown-accumulation mechanism doesn't generalize to ETH spot — the result is decisively negative and fee-fragile, and the recovery-exit fires in ~1 day instead of riding for weeks, so optimizing a no-edge base would only overfit.

A long-only ETHUSDT spot daily accumulation strategy mirroring the paper-stage BTC sibling (Sharpe 3.33): it waited for ETH to draw down ≥15% from its prior 30-day high while still above its 200-day SMA, entered on the first daily up-close, and exited on recovery back to the 30-day high (to ride the rebound), a regime breakdown below 0.95× the SMA, or a 60-day time-stop.

Over 8.7 years and 110 trades it had no edge: profit factor 0.82, expectancy -$173/trade, Sharpe -0.91, -17.3% total return, and a 35% max drawdown, negative in 2019/2020/2021/2022. It also didn't behave as designed — average holding was just ~1 day rather than the intended weeks-long ride, so entries fired on near-complete V-recoveries that hit the 30-day-high exit fast for small wins (avg win $1,540 < avg loss $1,885), while fast churn pushed commissions to 11.37% of gross at the 0.20% spot round-trip cost.