Skip to content

"Vibe-Trading: Your Personal Trading Agent"

fix(quantlib): support multi-segment test sets and guard empty training partitions in CV

santhreal71a483d0b720
diff --git a/agent/src/quantlib/crossvalidation.py b/agent/src/quantlib/crossvalidation.py
index be8d41217..686538067 100644
--- a/agent/src/quantlib/crossvalidation.py
+++ b/agent/src/quantlib/crossvalidation.py
@@ -178,27 +178,32 @@ def _apply_purge_and_embargo(
n = label_ends.size
starts = np.arange(n)
test_positions = np.flatnonzero(test_mask)
- first_test, last_test = int(test_positions[0]), int(test_positions[-1])
-
- # The test set's own footprint runs from its first observation to the last
- # index any of its labels reaches -- not merely to its last row.
- test_span_end = int(max(last_test, label_ends[test_positions].max()))
+ if test_positions.size == 0:
+ return (starts, 0, 0)
candidate = ~test_mask
- # Closed intervals on both ends: [start_i, end_i] and [first_test, test_span_end]
- # overlap when start_i <= test_span_end and end_i >= first_test.
- overlaps = (starts <= test_span_end) & (label_ends >= first_test)
- purged_mask = candidate & overlaps
- candidate_after_purge = candidate & ~overlaps
+ breaks = np.where(np.diff(test_positions) > 1)[0]
+ segment_starts = np.insert(test_positions[breaks + 1], 0, test_positions[0])
+ segment_ends = np.append(test_positions[breaks], test_positions[-1])
- if embargo_size > 0:
- embargo_end = min(test_span_end + embargo_size, n - 1)
- in_embargo = (starts > test_span_end) & (starts <= embargo_end)
- else:
- in_embargo = np.zeros(n, dtype=bool)
+ purged_mask = np.zeros(n, dtype=bool)
+ embargo_mask = np.zeros(n, dtype=bool)
+
+ for seg_start, seg_end in zip(segment_starts, segment_ends):
+ seg_positions = np.arange(seg_start, seg_end + 1)
+ seg_span_end = int(max(seg_end, label_ends[seg_positions].max()))
- embargo_mask = candidate_after_purge & in_embargo
- train_mask = candidate_after_purge & ~in_embargo
+ overlaps = (starts <= seg_span_end) & (label_ends >= seg_start)
+ purged_mask |= (candidate & overlaps)
+
+ if embargo_size > 0:
+ emb_end = min(seg_span_end + embargo_size, n - 1)
+ in_emb = (starts > seg_span_end) & (starts <= emb_end)
+ embargo_mask |= (candidate & in_emb)
+
+ candidate_after_purge = candidate & ~purged_mask
+ embargo_mask = candidate_after_purge & embargo_mask
+ train_mask = candidate_after_purge & ~embargo_mask
return (
np.flatnonzero(train_mask),
@@ -261,6 +266,8 @@ def purged_kfold_splits(
train, purged, embargoed = _apply_purge_and_embargo(
label_ends, test_mask, embargo_size
)
+ if train.size == 0:
+ raise ValueError(f"Purge and embargo removed all training samples for fold {fold}")
yield Split(
train=train,
test=np.arange(start, stop),
@@ -342,6 +349,8 @@ def group_purged_kfold_splits(
train_rows_list.append(group_to_rows[g])
train_rows = np.concatenate(train_rows_list) if train_rows_list else np.array([], dtype=int)
+ if train_rows.size == 0:
+ raise ValueError(f"Purge and embargo removed all training samples for fold {fold}")
train_rows.sort()
test_rows.sort()
@@ -489,6 +498,8 @@ def combinatorial_purged_splits(
train, purged, embargoed = _apply_purge_and_embargo(
label_ends, test_mask, embargo_size
)
+ if train.size == 0:
+ raise ValueError("Purge and embargo removed all training samples for fold")
yield Split(
train=train,
test=np.sort(test),
@@ -533,22 +544,36 @@ def detect_boundary_leakage(
overlapping = np.array([], dtype=int)
else:
label_ends = _as_label_spans(label_end_times, n_samples)
- first_test = int(split.test.min())
- test_span_end = int(max(split.test.max(), label_ends[split.test].max()))
+ test_positions = np.sort(split.test)
+ breaks = np.where(np.diff(test_positions) > 1)[0]
+ segment_starts = np.insert(test_positions[breaks + 1], 0, test_positions[0])
+ segment_ends = np.append(test_positions[breaks], test_positions[-1])
+
train_starts = split.train
train_ends = label_ends[split.train]
- mask = (train_starts <= test_span_end) & (train_ends >= first_test)
- overlapping = train_starts[mask]
+ overlapping_mask = np.zeros(len(train_starts), dtype=bool)
+ for seg_start, seg_end in zip(segment_starts, segment_ends):
+ seg_positions = np.arange(seg_start, seg_end + 1)
+ seg_span_end = int(max(seg_end, label_ends[seg_positions].max()))
+ overlapping_mask |= (train_starts <= seg_span_end) & (train_ends >= seg_start)
+ overlapping = train_starts[overlapping_mask]
if embargo_size > 0:
if label_end_times is None:
- span_end = int(split.test.max())
+ label_ends = np.arange(n_samples or (int(split.test.max()) + 1))
else:
label_ends = _as_label_spans(label_end_times, n_samples)
- span_end = int(max(split.test.max(), label_ends[split.test].max()))
- violations = split.train[
- (split.train > span_end) & (split.train <= span_end + embargo_size)
- ]
+ test_positions = np.sort(split.test)
+ breaks = np.where(np.diff(test_positions) > 1)[0]
+ segment_starts = np.insert(test_positions[breaks + 1], 0, test_positions[0])
+ segment_ends = np.append(test_positions[breaks], test_positions[-1])
+
+ violations_mask = np.zeros(len(split.train), dtype=bool)
+ for seg_start, seg_end in zip(segment_starts, segment_ends):
+ seg_positions = np.arange(seg_start, seg_end + 1)
+ seg_span_end = int(max(seg_end, label_ends[seg_positions].max()))
+ violations_mask |= (split.train > seg_span_end) & (split.train <= seg_span_end + embargo_size)
+ violations = split.train[violations_mask]
else:
violations = np.array([], dtype=int)
diff --git a/agent/tests/quantlib/test_crossvalidation.py b/agent/tests/quantlib/test_crossvalidation.py
index 96893c024..e9ef260c5 100644
--- a/agent/tests/quantlib/test_crossvalidation.py
+++ b/agent/tests/quantlib/test_crossvalidation.py
@@ -300,6 +300,16 @@ def test_mismatched_label_length_rejected():
list(purged_kfold_splits(100, np.arange(50), n_folds=4))
+def test_excessive_embargo_removing_all_training_samples_rejected():
+ # A 2-fold split with 80% embargo removes all training samples
+ with pytest.raises(ValueError, match="removed all training samples"):
+ list(purged_kfold_splits(10, np.arange(10), n_folds=2, embargo_fraction=0.8))
+
+ groups = [1, 1, 2, 2]
+ with pytest.raises(ValueError, match="removed all training samples"):
+ list(group_purged_kfold_splits(groups, n_folds=2, embargo_fraction=0.8))
+
+
@pytest.mark.parametrize("n_test_groups", [0, 6, 7])
def test_bad_combinatorial_group_count_rejected(n_test_groups):
with pytest.raises(ValueError, match="n_test_groups"):