SolSpotSma200RegimeHoldLong

SOL Spot Long-Hold with 200-SMA Regime Filter (Daily, Pure Regime-Flat Exit)

A long-only single-instrument PURE regime-conditional buy-and-hold strategy on SOLUSDT spot (BINANCE_SPOT) using daily bars and OHLCV-only data. This is the simplest possible strategy in the pipeline: just one tunable parameter (the SMA period) and a binary regime decision: hold spot SOL when above 200-day SMA, hold cash when below. No ATR trail. No drawdown threshold. No EMA crossover events. No vol regime filter. No multi-condition AND-gates. The single dominant filter is the 200-day SMA regime check. This proposal explicitly addresses the recently-documented failure patterns (8+ this session): rather than auto-replicate complex architectures across assets (which has failed repeatedly), this uses the SIMPLEST POSSIBLE mechanism that has empirical validity — buy and hold during confirmed bull regimes. Crucially, this proposal explicitly fixes the failure mode of SOL Spot Drawdown DCA (-11.4% over 5.6 years): that strategy failed because the 5% trailing stop was too tight for SOL's ~90% volatility, capturing only 30% of upside before triggering on normal pullbacks. THIS proposal HAS NO TRAIL — just a binary 200-SMA regime check. The volatility-calibration trap that doomed SOL Spot Drawdown CANNOT apply because there is no trail to mis-calibrate. Asset selection is justified by SOL-SPECIFIC empirical evidence: SOL has TWO highly successful strategies (SOL Daily 21/55 EMA Sharpe 3.81 paper_stage, SOL 4H Volume Breakout Sharpe 4.42 promoting) confirming SOL responds CLEANLY to daily regime filtering. The mechanism selection is justified by the explicit failure mode of SOL Spot Drawdown — by REMOVING the ATR trail and drawdown threshold (which were the failure points), and KEEPING only the 200-SMA regime filter, this proposal isolates the proven element of regime classification. NO bait-and-switch — strictly SOLUSDT.BINANCE_SPOT throughout. ONE tunable parameter only (SMA period) means cliff_count cannot exceed 1 — structural defense against the BNB Golden Cross cliff_count=4 fragility failure. Calibrated for ~3-8 entries per market cycle (very low frequency, very high holding period).

Implements the hypothesis's deliberately minimal design: isolate the single empirically-valid element (200-day SMA bull-regime classification) and remove every component that caused the SOL Spot Drawdown DCA failure — there is NO trailing stop and NO drawdown threshold, so the volatility-calibration trap that tight-stopped that strategy structurally cannot apply. calculate_signal() returns the continuous SMA distance (close-sma)/sma so the decision variable varies every bar and is in natural units; should_enter() and should_exit() use the same zero boundary so the position is held continuously through the bull regime and flattened exactly when price loses the SMA (no whipsaw band beyond the SMA itself). Venue routing follows the rule for long-only / no-leverage / non-funding edges: BINANCE_SPOT (CASH, leverage forced to 1) rather than futures, so it pays the spot fee schedule without wasting margin/funding it doesn't use. invest_fraction=0.95 deploys near-full equity (appropriate for a buy-and-hold regime strategy) while leaving a cash buffer for fees/precision. min_bars_required = sma_period + 1 guarantees a full window before the first decision. With only the SMA period tunable, the optimizer's cliff_count cannot exceed 1 — the structural defense the hypothesis calls for. Holding periods of months dwarf the ~0.20% spot round-trip cost, so fee drag is negligible.

Failed deflated Sharpe: DSR=0.0, expected-max=16.15 over 225 trials — the optimized Sharpe of 2.84 is far below the best-of-N luck bar and statistically indistinguishable from selection noise (is_significant=false, sharpe_ci_low -13.9 straddles 0, PBO 0.7725 >> 0.5). Textbook overfit/outlier-driven failure: walk-forward is_overfitted=true with avg IS 4.80 collapsing to avg OOS -8.42 (windows [-29.73, 4.47, 0.0]), and a failed holdout (holdout Sharpe 0.0, ratio 0 vs WF-OOS -8.42). The optimizer pushed sma_period 200->101, thinning to 54 whipsaw trades (win rate 13%, 11 consecutive losses, 82% max DD); the 1957% headline is a mirage — end_unrealized_pct 120% means it is entirely open-position mark-to-market, and the realized edge lives in the 2021 (+2518%) and 2024 (+274%) bull outliers while every other year is deeply negative (2022 -46%, 2023 -51%, 2025 -21%, 2026 -4.5%). This is overfit on a 1-parameter space, not a tunable-region case: the sweep already established its best config loses across every out-of-sample window, so a second pass hits the identical wall — there is no robust region to tune toward (sensitivity passing is necessary but not sufficient). Not revise_hypothesis either: the pure-200-SMA-regime-hold mechanism has no proven sibling to redirect (the cited SOL EMA/volume-breakout strategies are different mechanisms), and the failure is overfit/single-trade concentration, not structural decay of a working mechanism — reframing onto another instrument would just re-overfit. FAILURE PATTERN: single-instrument 200-SMA regime-hold on a high-vol alt produces a headline carried by 1-2 bull-year outliers plus end-of-backtest open-position unrealized, passes sensitivity (1 trivial param) but fails DSR (0.0)/PBO/holdout with deeply negative walk-forward OOS as the optimizer thins sma_period and amplifies whipsaw.

Pure regime-conditional buy-and-hold on SOLUSDT.BINANCE_SPOT, daily bars, OHLCV only. The strategy holds spot SOL (95% of equity) whenever the close is above the 200-day SMA and exits fully to cash whenever the close drops to or below the SMA. There is exactly one tunable parameter (the SMA period). No trailing stop, no drawdown threshold, no crossover-event logic, no volatility filter — just the binary 200-SMA regime check. Long-only on a CASH spot account: the only SELL is the regime exit of a held long. Very low frequency, very high holding period (~3–8 regime switches per market cycle).

Strategy functions correctly and trades exactly match the hypothesis: long-only spot, enters above 200-SMA, holds long through bull regimes (326d/267d/122d), exits to cash on regime breaks. No bait-and-switch (strictly SOLUSDT.BINANCE_SPOT), no real leverage (0.95-0.97).

Trade frequency (21 trades / 5.7yr) is reasonable for the low-frequency regime-hold design and sufficient to support a 1-parameter sweep without overfitting.

Structurally minimal: one tunable parameter (sma_period) means cliff_count cannot exceed ~1 — genuine defense against the multi-parameter fragility that failed siblings.

Regime filter demonstrably works as intended — it sat out the 2023 chop with small losses and captured the 2021 and 2024 bull legs.

Edge is concentrated in 2 of 21 trades (pnl_pct 925% and 502%); the remaining 19 are predominantly small losers, win_rate 19%. Outlier-dependence is the exact pattern that has repeatedly failed DSR/walk-forward/holdout in optimization.

Headline metrics are distorted by a reporting artifact (whole-trade PnL attributed to the exit day), inflating annualized_volatility to 3331% and pushing sharpe_ci_low to -19 — the point-estimate Sharpe 4.4 is unreliable until the robustness block is computed.

max_drawdown 65% and end_unrealized_pct 52.9% — half the headline return is an open mark-to-market position at backtest end; realized track record is more modest.

Most of the cumulative return tracks SOL's own appreciation during bull regimes; benchmark-relative alpha is unknown (alpha/IR null here).

enough for 1-param sweep

2 of 21 trades carry P&L

watch

yes

must match

Implementation is clean and matches the hypothesis (binary 200-SMA regime hold, one tunable parameter); the dominant trades are real SOL bull-regime holds, not data artifacts.

Sensitivity is effectively clean — sma_period is not a cliff (only the trivial invest_fraction>1.0 spot-cap cliff), confirming the structural-simplicity claim.

Deflated Sharpe = 0.0 vs the 16.15 best-of-225 luck bar — the optimized Sharpe 2.84 is below the noise floor; is_significant=false, sharpe_ci_low -13.9 straddles 0.

PBO 0.7725 (>>0.5): the parameter selection is more likely overfit than not.

Walk-forward is_overfitted=true: avg IS 4.80 collapses to avg OOS -8.42, OOS windows [-29.73, 4.47, 0.0] — no robust region.

Holdout failed (Sharpe 0.0, ratio 0 vs WF-OOS -8.42).

Entire 1957% headline is open-position unrealized (end_unrealized_pct 120%); realized edge is carried by 2021/2024 bull outliers with every other year deeply negative and recent regime (2025 -21%, 2026 -4.5%) negative. Optimized config whipsaws (13% win rate, 11 consecutive losses, 82% max DD).

expected-max 16.15

SolSpotSma200RegimeHoldLong distilled SOL down to its simplest viable mechanism — a single 200-day SMA regime filter — deliberately removing the trailing-stop machinery that had sunk the SOL Spot Drawdown DCA. It cleared the pre-optimization gate (with reviewers explicitly flagging that 2 of 21 trades and a reporting artifact carried the headline), but optimization confirmed the concern: deflated Sharpe of zero, PBO 0.77, an in-sample 4.80 collapsing to a -8.42 OOS average, a failed holdout, and an optimized config that whipsawed to an 82% drawdown. The analyst judged it overfit, single-trade-concentrated SOL bull-beta with no robust tuning region and no proven sibling to redirect — neither iterable nor reframable — and abandoned it. It reached optimization and analysis but was abandoned after a single iteration, never advancing to risk review.

Single-instrument 200-SMA regime-hold on a high-vol alt produces a headline carried by one or two bull-year outliers plus end-of-backtest open-position unrealized PnL; it passes sensitivity (one trivial parameter) but fails DSR/PBO/holdout with deeply negative walk-forward OOS — clean sensitivity is necessary but not sufficient, and the result is regime-filtered SOL beta, not differentiated alpha.

After passing the backtest-review gate (verdict: optimize), the analyst abandoned it at the post-optimization stage: it failed the deflated Sharpe test (DSR 0.0 vs expected-max luck bar 16.15 over 225 trials, is_significant=false, PBO 0.77, sharpe_ci_low -13.9), the walk-forward was flagged overfit (avg IS 4.80 collapsing to avg OOS -8.42), and the holdout failed (ratio 0) — despite clean sensitivity from its single parameter.

The simplest possible strategy in the pipeline: a long-only single-parameter regime-conditional buy-and-hold on SOLUSDT.BINANCE_SPOT daily bars (OHLCV only) that held spot SOL while price was above its 200-day SMA and moved fully to cash when it fell below — a pure binary regime-flat exit deliberately stripped of the ATR trail and drawdown threshold that doomed the earlier SOL Spot Drawdown DCA.

The base backtest over 2020-08-11 to 2026-05-18 showed a huge but distorted headline across 21 trades — total return +2497%, CAGR 134.9%, Sharpe 4.41 (CI [-19.0, 7.95]), profit factor 2.28, a 65% max drawdown, 19% win rate — with the edge concentrated in just 2 of 21 trades and ~53% of the return being open-position unrealized; optimization (Sharpe 2.84, sma_period thinned to 101) whipsawed to a 13% win rate, 11 consecutive losses, and 82% drawdown.