跳到主要內容

Feature Iteration Run #518: zz_sylvain

Status: completed | Started: 2026-07-17 18:42:45 | Ended: 2026-07-17 18:44:56

📊 Iterations

Iter Y Label Model Warm Proxy Parent FS Feature DSL DSL Valid Baseline PF Modified PF Scalar Score t-stat p-value Cohen's d Result Reason
1 — gradient_boosting — — — k_loop_iter_1_95ceaa k_close - i_ema(k_close, 20) ✅ 1.286 1.031 -1.8597 0.000 1.000 0.000 ❌ REJECT p=1.000 ≥ 0.05, cohens_d=0.000 ≤ 0.3, constraints: ['sharpe_ratio >= 1.0: got 0.8314']
2 — gradient_boosting — — — k_loop_iter_2_8ba4bb atr_norm * (1 - price_position) ✅ 1.286 1.187 -0.5523 0.000 1.000 0.000 ❌ REJECT p=1.000 ≥ 0.05, cohens_d=0.000 ≤ 0.3
3 — gradient_boosting — — — k_loop_iter_3_f8c027 dist_to_high * vol_ratio / (i_atr(k_hig… ✅ 1.286 1.252 -0.3372 0.000 1.000 0.000 ❌ REJECT p=1.000 ≥ 0.05, cohens_d=0.000 ≤ 0.3

📄 Report Markdown

# AUTOLOOP: zz_sylvain (3 iterations)

> **Phase**: Y4.7 Option B (DSL → Python code → FeatureSet candidate)
> **Target indicator**: zz_sylvain
> **Base feature_set_id**: 7
> **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**: 131.64s
> **Total accepted**: 0
> **Total rejected**: 3

## Iterations 詳情

### Iteration 1
- DSL: `k_close - i_ema(k_close, 20)` (feature=k_loop_iter_1_95ceaa)
- DSL valid: True
- FeatureSet candidate id: 578
- baseline PF: 1.286 → modified PF: 1.031
- t_stat: 0.000, p_value: 1.000, cohens_d: 0.000
- scalar_score: -1.8597
- accepted: False, reason: p=1.000 ≥ 0.05, cohens_d=0.000 ≤ 0.3, constraints: ['sharpe_ratio >= 1.0: got 0.8314']

**Metrics summary:**
  - profit_factor: baseline=1.286, modified=1.031, diff=-0.255
  - sharpe_ratio: baseline=7.879, modified=0.831, diff=-7.048
  - win_rate: baseline=0.553, modified=0.430, diff=-0.123
  - max_drawdown: baseline=0.075, modified=0.108, diff=0.033
  - test_accuracy: baseline=0.986, modified=0.970, diff=-0.016

### Iteration 2
- DSL: `atr_norm * (1 - price_position)` (feature=k_loop_iter_2_8ba4bb)
- DSL valid: True
- FeatureSet candidate id: 579
- baseline PF: 1.286 → modified PF: 1.187
- t_stat: 0.000, p_value: 1.000, cohens_d: 0.000
- scalar_score: -0.5523
- accepted: False, reason: p=1.000 ≥ 0.05, cohens_d=0.000 ≤ 0.3

**Metrics summary:**
  - profit_factor: baseline=1.286, modified=1.187, diff=-0.099
  - sharpe_ratio: baseline=7.879, modified=5.779, diff=-2.100
  - win_rate: baseline=0.553, modified=0.540, diff=-0.013
  - max_drawdown: baseline=0.075, modified=0.133, diff=0.058
  - test_accuracy: baseline=0.986, modified=0.949, diff=-0.037

### Iteration 3
- DSL: `dist_to_high * vol_ratio / (i_atr(k_high, k_low, k_close, 14) + 1e-9)` (feature=k_loop_iter_3_f8c027)
- DSL valid: True
- FeatureSet candidate id: 580
- baseline PF: 1.286 → modified PF: 1.252
- t_stat: 0.000, p_value: 1.000, cohens_d: 0.000
- scalar_score: -0.3372
- accepted: False, reason: p=1.000 ≥ 0.05, cohens_d=0.000 ≤ 0.3

**Metrics summary:**
  - profit_factor: baseline=1.286, modified=1.252, diff=-0.033
  - sharpe_ratio: baseline=7.879, modified=6.583, diff=-1.296
  - win_rate: baseline=0.553, modified=0.470, diff=-0.083
  - max_drawdown: baseline=0.075, modified=0.177, diff=0.102
  - test_accuracy: baseline=0.986, modified=0.969, diff=-0.017

📊 Multi-Objective Metrics

Iteration 1: k_loop_iter_1_95ceaa

Y Label: — | Parent FS: #—

Objective mode: scalarize

Scalar score: -1.8597

Failed constraints: sharpe_ratio >= 1.0: got 0.8314

MetricBaselineModifiedDiff
win_rate 0.553 0.430 -0.123
max_drawdown 0.075 0.108 0.033
sharpe_ratio 7.879 0.831 -7.048
profit_factor 1.286 1.031 -0.255
test_accuracy 0.986 0.970 -0.016

Iteration 2: k_loop_iter_2_8ba4bb

Y Label: — | Parent FS: #—

Objective mode: scalarize

Scalar score: -0.5523

MetricBaselineModifiedDiff
win_rate 0.553 0.540 -0.013
max_drawdown 0.075 0.133 0.058
sharpe_ratio 7.879 5.779 -2.100
profit_factor 1.286 1.187 -0.099
test_accuracy 0.986 0.949 -0.037

Iteration 3: k_loop_iter_3_f8c027

Y Label: — | Parent FS: #—

Objective mode: scalarize

Scalar score: -0.3372

MetricBaselineModifiedDiff
win_rate 0.553 0.470 -0.083
max_drawdown 0.075 0.177 0.102
sharpe_ratio 7.879 6.583 -1.296
profit_factor 1.286 1.252 -0.033
test_accuracy 0.986 0.969 -0.017

📁 Offline Report

Markdown backup: reports/autoloop/zz_sylvain/20260718_024456.md