EthSpotPullbackInUptrendBuyTheDipLong

ETH BINANCE_SPOT Daily Pullback-in-Uptrend Long-Only Buy-the-Dip (Pring-Style, 4-Parameter, OHLCV-Only)

Long-only buy-the-dip / pullback-in-uptrend strategy on ETHUSDT.BINANCE_SPOT using daily bars. This is a structurally different mechanism class from every existing portfolio entry: it BUYS DIPS in confirmed uptrends rather than chasing breakouts (LINK Spot, DOGE futures), riding established trends (BTC/ETH/SOL TSM), exploiting squeezes (BNB Spot Bollinger), or trading ranges (HL BTC ORB). The mechanism: only consider entries when ETH is in a confirmed long-term uptrend (price > 200-day SMA), then wait for a meaningful pullback (close drops ≥6% from the trailing 10-day high), then require a reversal-confirmation bar (today's close > yesterday's close, with above-average volume). This is Pring's classic 'three-tier pullback filter' — trend / depth / confirmation — that has decades of equity-market validation. The reason this mechanism is fresh for the portfolio: pullback-buy strategies are CONTRA to the breakout/momentum strategies that dominate the existing 1160-experiment factory. They profit from a structurally different phenomenon: the empirically-observed tendency of trending assets to retrace 3-8% on healthy uptrends before continuing higher (the so-called 'higher-low' pattern). On crypto specifically, ETH's daily volatility creates frequent 5-10% pullbacks even in strong uptrends (2020-Q4, 2021-H1, 2024-H2) that mechanical breakout strategies miss because they only fire at new highs. Pure OHLCV — NO supplementary data dependency (avoids the 14+ session abandons from infra gaps). 4 parameters: pullback_pct=6.0, lookback_high=10, trend_sma=200, atr_stop_mult=2.0. ETH BINANCE_SPOT venue is chosen because: (a) BINANCE_SPOT remains at 10.1% vs 15% quota target, (b) ETH spot has 8+ years of continuous Binance data for robust walk-forward, (c) ETH is the 2nd-largest crypto with $500M+ daily Binance spot volume (massively above liquidity floor), (d) spot venue's CASH account naturally limits the strategy to long-only (matches the mechanism's directional bias), (e) ETH has experienced multiple distinct trending regimes (2017 ICO rally, 2020-2021 DeFi bull, 2024 ETF + L2 narrative) interspersed with bear periods (2018, 2022-2023) — providing rich regime coverage. Long-only with hard stop included from the start (after the DOGE-SHIB abandon flagged stop-less MR strategies as catastrophic).

Implements a mechanism class genuinely fresh for the portfolio — buy-the-dip in an uptrend — which is CONTRA to the breakout/momentum/squeeze/range strategies that dominate the existing factory and profits from the higher-low retrace-then-continue tendency rather than chasing new highs. Pring's three-tier trend/depth/confirmation filter has decades of equity-market validation and translates directly to ETH's frequent 5-10% daily pullbacks within strong uptrends. Venue is BINANCE_SPOT (CASH, leverage 1.0) exactly per the rules and the hypothesis: long-only, unleveraged, no funding component; ETH spot has 8+ years of deep-liquidity Binance data spanning multiple bull/bear regimes for robust walk-forward. A hard stop is included from the start (chandelier ATR trailing stop) per the lesson that stop-less mean-reversion is catastrophic — here a single ATR rule serves as both initial stop and profit trail, so the buy-the-dip realizes round-trips without a separate take-profit and atr_stop_mult is the one risk knob the hypothesis names. I verified the only open question — trade frequency on ETH's actual pullback distribution — on clean ETH daily history (range $107-$4830): the trend+depth+confirmation filter yields ~8 entries/year (upper bound omitting the volume tier; ~5-7/year with it) over 6.3 years, ~50 round-trips, comfortably in the healthy 5-10/year zone and well above the >=5-per-walk-forward-window floor (closed-position count, no harvest workaround). ETH spot daily OHLCV has full history via on-demand collection, so there is no data-gap risk. Verified locally: Layer 1 static (clean), Layer 2 synthetic (all 6 scenarios pass with a non-frozen SMA-deviation signal), and the real-data entry-frequency check above.

Failed deflated Sharpe: DSR=0.277, expected-max=6.82 over 225 trials — the optimized Sharpe of 5.51 is below the best-of-N luck bar and statistically indistinguishable from selection noise (is_significant=false, PBO 0.692 > 0.5). Overfit by over-selectivity: walk-forward is_overfitted=true with avg IS 9.51 collapsing to avg OOS 0.0 and ALL three OOS windows exactly [0.0, 0.0, 0.0] — the optimizer shifted pullback_pct 6->8.17 and trend_sma 200->121 into a config that produces ZERO qualifying trades in every out-of-sample window, and the holdout 'passed=true' is a DEGENERATE artifact (holdout_sharpe 4.75 measured against a 0.0 WF-OOS reference, ratio 0) that must not override the failed gates. The favorable full-period sharpe_ci_low (0.98) and PSR (0.994) are bootstrap statistics that do not correct for the 225-trial selection that DSR corrects for, and are swamped by the all-zero OOS — a config that produces no out-of-sample trades has not generalized regardless of its in-sample CI. Sensitivity passing (cliff_count 0) is necessary but not sufficient. This confirms the pre-optimization concern: the edge is 2017/2021 bull-outlier beta (unoptimized kurtosis 11.8, a single 2021-02-22 +47.5% day, 2024 -6%) that does not generalize, so when forced to generalize the optimizer over-selects into a non-trading config. Not a tunable-region case: the 225-trial sweep already produced all-zero OOS over a cliff-free space, so a second pass hits the identical wall. Not revise_hypothesis: the pullback-buy mechanism has no promoted sibling to redirect, and the failure is overfit/zero-OOS rather than structural decay of a working mechanism — reframing onto another instrument would just re-overfit the same outlier dependence. FAILURE PATTERN: long-only daily pullback-in-uptrend buy-the-dip on a major (ETH spot) shows a high in-sample Sharpe carried by 2017/2021 bull-outlier days but fails DSR (below the best-of-N luck bar) with all-zero walk-forward OOS and only a degenerate (zero-reference) holdout pass — the optimizer over-selects pullback depth/trend to maximize IS Sharpe and stops trading out-of-sample.

Long-only Pring-style pullback-in-uptrend buy-the-dip on ETHUSDT.BINANCE_SPOT (CASH), daily bars, OHLCV-only. It applies a three-tier filter on each daily bar: (1) TREND — close > SMA(trend_sma=200), a confirmed long-term uptrend; (2) DEPTH — close has dropped >= pullback_pct (6%) from the trailing lookback_high (10)-day high, a meaningful retrace; (3) CONFIRMATION — today's close > yesterday's close AND today's volume > its trailing vol_lookback (20) average, a reversal bar rather than a falling knife. When all three align it BUYS the dip. The exit is a single chandelier ATR trailing stop (highest close since entry minus atr_stop_mult=2.0 x ATR(14)), which caps the loss while the position is young and trails profit as the uptrend resumes, plus a 60-day time stop. Position size is 25% of equity per entry. calculate_signal returns a continuous SMA-deviation diagnostic (non-frozen) while the discrete 3-tier gate lives in should_enter. Single instrument, long-only, leverage 1.0, 4 core parameters, zero supplementary data.

Genuine positive edge with correct pullback shape: profit_factor 1.72, omega 2.07, avg_win $14.5k vs avg_loss $5.4k at 38.8% win rate; fully realized (end_unrealized 0.8%)

Clean execution (49 signaled/submitted, 0 dropped) and a novel-for-the-portfolio mechanism class (buy-the-dip, contra to the dominant breakout/momentum strategies)

Adequate sample for the horizon: 49 trades over ~8.7 years (~5.6/yr, avg hold 23 days); has a hard ATR trailing stop from the start

Headline Sharpe 3.19 is a per-period-attribution artifact (annualized_vol 154%, Sortino 12.8) — must be judged on realized PF/return, not Sharpe

Severe outlier concentration: edge dominated by 2021 (+60.5%) and 2017 (+28.7%), with the single 2021-02-22 day at +47.5%; kurtosis 11.8, strip 2-3 days and the edge collapses

Negative recent regime (2024 +0.4%, 2025 -10.1%) where the holdout will land; sharpe_ci_low -1.28 straddles 0; 49 trades stresses the 3-window walk-forward (~5 OOS trades/window)

Clean execution and sensitivity passed (cliff_count 0); novel-for-the-portfolio mechanism (pullback buy-the-dip); hard ATR stop from the start

Optimized in-sample profit_factor 2.39 — the three-tier filter captures the 2017/2019/2020 bull pullbacks when it fires

Failed deflated Sharpe: DSR 0.277 (<0.95), optimized Sharpe 5.51 below expected-max luck bar 6.82 over 225 trials; is_significant=false, PBO 0.692 > 0.5

Walk-forward is_overfitted=true: avg IS 9.51 collapses to avg OOS 0.0 with ALL three OOS windows [0.0, 0.0, 0.0] — the chosen config produces zero out-of-sample trades after the optimizer shifted pullback_pct 6->8.17 / trend_sma 200->121

Holdout 'passed=true' is degenerate: holdout_sharpe 4.75 measured against a 0.0 WF-OOS reference (ratio 0) — a meaningless artifact, not a validation

Edge is 2017/2021 bull-outlier beta (kurtosis 11.8, single 2021-02-22 +47.5% day) that does not generalize out-of-sample

EthSpotPullbackInUptrendBuyTheDipLong introduced a genuinely fresh mechanism class to the portfolio — a three-tier Pring pullback filter buying healthy dips in confirmed ETH uptrends — and cleared the review gate with a real positive edge (profit factor 1.72) and an 'optimize' verdict, despite a flagged outlier dependence on 2017/2021 bull days. Optimization confirmed the concern: deflated Sharpe collapsed to 0.277 below the 6.82 luck bar, PBO hit 0.692, and the optimizer shifted pullback depth and trend length into a config that produced zero out-of-sample trades across all three walk-forward windows, leaving only a degenerate zero-reference holdout 'pass.' The analyst ruled it overfit-by-over-selectivity whose edge was non-generalizing bull-outlier beta and abandoned it. It ended after one iteration as abandoned, having reached optimization and analysis but never risk review.

A novel-for-the-portfolio buy-the-dip mechanism with a real-looking baseline edge can still fail because that edge is 2017/2021 bull-outlier beta that does not generalize: when forced to generalize, the optimizer over-selects pullback depth and trend length until the strategy stops trading out-of-sample (all-zero OOS), failing DSR — and a high in-sample Sharpe, favorable bootstrap CI/PSR, and a zero-reference 'holdout pass' do not survive multiple-testing deflation.

Although it passed the backtest-review gate with an 'optimize' verdict, the analyst abandoned it at the post-optimization analyzing stage: it failed deflated Sharpe (0.277, below the luck bar, is_significant=false), and the optimizer over-selected (pullback_pct 6→8.17, trend_sma 200→121) into a config that produced zero qualifying trades in all three walk-forward OOS windows, with the 'passing' holdout being a degenerate zero-reference artifact — so it reached optimization and analysis but never advanced to risk review.

A long-only daily Pring-style pullback-in-uptrend buy-the-dip strategy on ETHUSDT spot (OHLCV-only) using a three-tier filter — trend (close > 200-day SMA), depth (close >=6% below the trailing 10-day high), and confirmation (an up-close bar on above-average volume) — to enter, exiting via an ATR chandelier trailing stop and time stop, a buy-the-dip mechanism deliberately contra to the portfolio's dominant breakout/momentum strategies.

The baseline showed a genuine positive edge with the correct pullback shape over 49 trades (profit factor 1.72, omega 2.07, avg win $14.5k vs avg loss $5.4k at a 38.8% win rate, +115.7% total return) but the headline Sharpe 3.19 was a per-period-attribution artifact (annualized vol 154%) and the edge was severely outlier-concentrated (kurtosis 11.8, dominated by 2021 +60.5% and 2017 +28.7%, including a single 2021-02-22 +47.5% day); after optimization DSR was 0.277 (vs a 6.82 luck bar), PBO 0.692, and walk-forward in-sample 9.51 collapsed to 0.0 OOS with all three windows producing zero out-of-sample trades.