gihyeon/cointrader
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: remove ADX hard filter from dataset builder, add ADX as ML feature

Browse Source 
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
21in7
2026-03-03 21:17:49 +09:00
parent 0aeb15ecfb
commit 9c6f5dbd76
1 changed files with 5 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
src/dataset_builder.py
View File

@@ -116,12 +116,6 @@ def _calc_signals(d: pd.DataFrame) -> np.ndarray:
# 둘 다 해당하면 HOLD (충돌 방지) # 둘 다 해당하면 HOLD (충돌 방지)
signal_arr[long_enter & short_enter] = "HOLD" signal_arr[long_enter & short_enter] = "HOLD"
# ADX 횡보장 필터: ADX < 25이면 추세 부재로 판단하여 진입 차단
if "adx" in d.columns:
adx = d["adx"].values
low_adx = (~np.isnan(adx)) & (adx < 25)
signal_arr[low_adx] = "HOLD"
return signal_arr return signal_arr
@@ -212,6 +206,10 @@ def _calc_features_vectorized(
side = np.where(signal_arr == "LONG", 1.0, 0.0).astype(np.float32) side = np.where(signal_arr == "LONG", 1.0, 0.0).astype(np.float32)
# ADX (ML 피처로 제공 — rolling z-score 정규화)
adx_raw = d["adx"].values.astype(np.float64) if "adx" in d.columns else np.zeros(len(d), dtype=np.float64)
adx_z = _rolling_zscore(adx_raw)
result = pd.DataFrame({ result = pd.DataFrame({
"rsi": rsi.values.astype(np.float32), "rsi": rsi.values.astype(np.float32),
"macd_hist": macd_hist.values.astype(np.float32), "macd_hist": macd_hist.values.astype(np.float32),
@@ -226,6 +224,7 @@ def _calc_features_vectorized(
"ret_5": ret_5_z, "ret_5": ret_5_z,
"signal_strength": strength, "signal_strength": strength,
"side": side, "side": side,
"adx": adx_z,
"_signal": signal_arr, # 레이블 계산용 임시 컬럼 "_signal": signal_arr, # 레이블 계산용 임시 컬럼
}, index=d.index) }, index=d.index)