admin/quant-trader-service
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Use actual plan edge for Entry PM training

Browse Source
This commit is contained in:
Codex
2026-06-28 07:26:59 +08:00
parent 3c0f2d0d91
commit 2a86a6e2fa
6 changed files with 129 additions and 63 deletions
Download Patch File Download Diff File Expand all files Collapse all files
training/trader_training/training.py
+7 -4
View File
@@ -39,8 +39,8 @@ TARGETS = {
"heads": [
("long_entry_prob", "binary", "long_entry_target", ["long_entry_prob"], ["longEntryProb"]),
("short_entry_prob", "binary", "short_entry_target", ["short_entry_prob"], ["shortEntryProb"]),
("long_expected_net_edge_bps", "regression", "long_expected_net_edge_bps", ["long_expected_net_edge_bps"], [None]),
("short_expected_net_edge_bps", "regression", "short_expected_net_edge_bps", ["short_expected_net_edge_bps"], [None]),
("long_expected_net_edge_bps", "regression", "long_actual_plan_net_edge_bps", ["long_expected_net_edge_bps"], [None]),
("short_expected_net_edge_bps", "regression", "short_actual_plan_net_edge_bps", ["short_expected_net_edge_bps"], [None]),
],
},
"CONTINUE": {
@@ -119,11 +119,12 @@ def train_small_models(args: Any) -> None:
head_results.extend(_fit_head(item, x_train_scaled, x_tune_scaled, train, tune, scaler))
for result in head_results:
logging.info(
"trader.training.model_head_trained runId=%s model=%s head=%s kind=%s metrics=%s",
"trader.training.model_head_trained runId=%s model=%s head=%s kind=%s targetSource=%s metrics=%s",
args.run_id,
model_name,
result.field,
result.kind,
result.metrics.get("target_source"),
result.metrics,
)
for result in head_results:
@@ -221,7 +222,9 @@ def _fit_head(item, x_train, x_tune, train: pd.DataFrame, tune: pd.DataFrame, sc
model.fit(x_train, y_train)
pred = model.predict(x_tune)
weight, bias = _fold_scaler(model.coef_.reshape(1, -1).T, np.array([model.intercept_]), scaler)
return [HeadResult(fields[0], None, "identity", weight, bias, _regression_metrics(y_train, y_val, pred), pred.reshape(-1, 1), y_val)]
metrics = _regression_metrics(y_train, y_val, pred)
metrics["target_source"] = target
return [HeadResult(fields[0], None, "identity", weight, bias, metrics, pred.reshape(-1, 1), y_val)]
raise ValueError(f"unsupported head kind: {kind}")