Feature Iteration Run #1876: zz_sylvain

# AUTOLOOP: zz_sylvain (3 iterations)

> **Phase**: Y4.7 Option B (DSL → Python code → FeatureSet candidate)
> **Target indicator**: zz_sylvain
> **Base feature_set_id**: 1748
> **Claim**: (default) Propose a new k_* feature that improves risk-adjusted returns
> **Objective mode**: scalarize
> **Metrics**: ['profit_factor', 'sharpe_ratio', 'win_rate', 'max_drawdown', 'test_accuracy']
> **Total duration**: 7027.84s
> **Total accepted**: 0
> **Total rejected**: 3

## Iterations 詳情

### Iteration 1
- DSL: `rsi_val = ta.rsi(close, 14)
macd_line = ta.ema(close, 12) - ta.ema(close, 26)
macd_norm = macd_line / (ta.atr(high, low, close, 14) + 1e-10)
result = (rsi_val - 50) / (1 + ta.atr(high, low, close, 5) / close) + macd_norm` (feature=k_rsi_atr_macd_momentum)
- DSL valid: True
- FeatureSet candidate id: 1843
- baseline PF: 2.049 → modified PF: 1.769
- t_stat: 0.000, p_value: 0.476, cohens_d: 0.120
- scalar_score: -0.1039
- accepted: False, reason: normalized_score -0.0659 <= 0.0, profit_factor_delta -0.2800 < 0.25, sharpe_delta -3.3876 (-16.31%) < -10%, champion trades 43 < 100 — PF statistically meaningless, hypothesis test not confirmed, DSR gate failed: champion SR=7.0143 ≤ expected_max=4.9782 (N=29 effective trials, PSR=0.9202)

**Metrics summary:**
  - profit_factor: baseline=2.049, modified=1.769, diff=-0.280
  - sharpe_ratio: baseline=20.771, modified=17.383, diff=-3.388
  - win_rate: baseline=0.545, modified=0.558, diff=0.013
  - max_drawdown: baseline=0.045, modified=0.040, diff=-0.004
  - test_accuracy: baseline=0.988, modified=0.949, diff=-0.038

### Iteration 2
- DSL: `fast_ema = ta.ema(close, 8)
slow_ema = ta.ema(close, 34)
ema_diff = fast_ema - slow_ema
atr_v = ta.atr(high, low, close, 14)
ema_diff_norm = ema_diff / (close * (1 + atr_v / close))
range_n = ta.highest(high, 14) - ta.lowest(low, 14) + 1e-10
result = ema_diff_norm * (1 - 2 * atr_v / range_n)` (feature=k_ema_norm_range_pct)
- DSL valid: True
- FeatureSet candidate id: 1844
- baseline PF: 2.049 → modified PF: 1.340
- t_stat: 0.000, p_value: 0.003, cohens_d: 0.299
- scalar_score: -3.6867
- accepted: False, reason: normalized_score -0.1885 <= 0.0, profit_factor_delta -0.7088 < 0.25, sharpe_delta -13.0903 (-63.02%) < -10%, hypothesis test not confirmed, DSR gate failed: champion SR=4.1806 ≤ expected_max=4.5678 (N=29 effective trials, PSR=0.4622)

**Metrics summary:**
  - profit_factor: baseline=2.049, modified=1.340, diff=-0.709
  - sharpe_ratio: baseline=20.771, modified=7.681, diff=-13.090
  - win_rate: baseline=0.545, modified=0.500, diff=-0.045
  - max_drawdown: baseline=0.045, modified=0.125, diff=0.080
  - test_accuracy: baseline=0.988, modified=0.987, diff=-0.001

### Iteration 3
- DSL: `denom = high - low
price_pos = denom != 0 ? (close - low) / denom : 0.5
vol_ratio = volume / (ta.sma(volume, 20) + 1e-10)
result = price_pos * vol_ratio` (feature=k_vol_price_pos)
- DSL valid: True
- FeatureSet candidate id: 1845
- baseline PF: 2.049 → modified PF: 1.649
- t_stat: 0.000, p_value: 1.000, cohens_d: 0.000
- scalar_score: -0.1920
- accepted: False, reason: normalized_score -0.1064 <= 0.0, profit_factor_delta -0.4003 < 0.25, sharpe_delta -6.6869 (-32.19%) < -10%, champion trades 61 < 100 — PF statistically meaningless, hypothesis test not confirmed

**Metrics summary:**
  - profit_factor: baseline=2.049, modified=1.649, diff=-0.400
  - sharpe_ratio: baseline=20.771, modified=14.084, diff=-6.687
  - win_rate: baseline=0.545, modified=0.508, diff=-0.037
  - max_drawdown: baseline=0.045, modified=0.083, diff=0.038
  - test_accuracy: baseline=0.988, modified=0.972, diff=-0.015