Datasets:
Tasks:
Tabular Classification
Formats:
parquet
Languages:
English
Size:
< 1K
Tags:
economics
quantitative-finance
causal-inference
macroeconomics
housing-economics
market-microstructure
License:
File size: 4,575 Bytes
ebcde1f | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 | from __future__ import annotations
from typing import Any, cast
import polars as pl
import pytest
from microstructure.data.schemas import SCHEMA_VERSION, table_from_records
from microstructure.research.features import (
ResearchDataError,
build_cancellation_intensity_features,
model_feature_columns,
)
def _delta(
update_id: int,
*,
bids: list[tuple[int, int]],
asks: list[tuple[int, int]],
continuity_id: str = "epoch-a",
) -> dict[str, Any]:
event_ts_ns = 1_700_000_000_000_000_000 + update_id * 1_000
return {
"schema_version": SCHEMA_VERSION,
"venue": "binance_spot",
"symbol": "BTCUSDT",
"event_ts_ns": event_ts_ns,
"received_ts_ns": event_ts_ns + 100,
"available_ts_ns": event_ts_ns + 100,
"availability_basis": "local_receive_time",
"capture_seq": update_id,
"continuity_id": continuity_id,
"first_update_id": update_id,
"last_update_id": update_id,
"previous_update_id": update_id - 1 if update_id > 1 else None,
"bids": [
{"price_ticks": price_ticks, "quantity_lots": quantity_lots}
for price_ticks, quantity_lots in bids
],
"asks": [
{"price_ticks": price_ticks, "quantity_lots": quantity_lots}
for price_ticks, quantity_lots in asks
],
"tick_size": 0.01,
"lot_size": 0.001,
"source_artifact_id": f"{update_id:064x}",
}
def _frame(records: list[dict[str, Any]]) -> pl.DataFrame:
table = table_from_records("depth_deltas", records)
return cast(pl.DataFrame, pl.from_arrow(table))
def test_cancellation_intensity_counts_only_observable_zero_quantity_deletes() -> None:
frame = _frame(
[
_delta(1, bids=[(10_000, 0)], asks=[(10_002, 5)]),
_delta(2, bids=[(9_999, 0)], asks=[(10_003, 0)]),
_delta(3, bids=[(10_000, 4)], asks=[]),
_delta(
10,
bids=[(10_000, 7)],
asks=[],
continuity_id="epoch-b",
),
]
)
result = build_cancellation_intensity_features(frame, windows=(2,))
rows = result.sort(["continuity_id", "decision_sequence"]).to_dicts()
assert [row["cancellation_deletes_current"] for row in rows] == [1, 2, 0, 0]
assert [row["depth_updates_current"] for row in rows] == [2, 2, 1, 1]
assert [row["cancellation_deletes_w2"] for row in rows] == [1, 3, 2, 0]
assert [row["depth_updates_w2"] for row in rows] == [2, 4, 3, 1]
assert [row["cancellation_intensity_w2"] for row in rows] == pytest.approx(
[0.5, 0.75, 2.0 / 3.0, 0.0]
)
assert set(result.get_column("cancellation_observation_policy")) == {
"zero_quantity_level_deletes_only"
}
assert not result.get_column("nonzero_reduction_classified_as_cancellation").any()
assert result.get_column("max_feature_source_ts_ns").equals(
result.get_column("feature_cutoff_ts_ns")
)
assert result.get_column("max_feature_source_sequence").equals(
result.get_column("decision_sequence")
)
assert model_feature_columns(result) == ("cancellation_intensity_w2",)
def test_future_depth_mutation_cannot_change_past_cancellation_features() -> None:
records = [
_delta(1, bids=[(10_000, 0)], asks=[]),
_delta(2, bids=[(10_000, 5)], asks=[]),
_delta(3, bids=[], asks=[(10_002, 0)]),
_delta(4, bids=[(9_999, 8)], asks=[]),
]
before = build_cancellation_intensity_features(_frame(records), windows=(3,))
mutated = [dict(record) for record in records]
mutated[3] = _delta(4, bids=[(9_999, 0)], asks=[(10_004, 0)])
after = build_cancellation_intensity_features(_frame(mutated), windows=(3,))
columns = [
"decision_sequence",
"cancellation_deletes_w3",
"depth_updates_w3",
"cancellation_intensity_w3",
"max_feature_source_ts_ns",
"max_feature_source_sequence",
]
assert (
before.filter(pl.col("decision_sequence") < 4)
.select(columns)
.equals(after.filter(pl.col("decision_sequence") < 4).select(columns))
)
def test_cancellation_features_fail_closed_on_unsegmented_sequence_gap() -> None:
frame = _frame(
[
_delta(1, bids=[(10_000, 0)], asks=[]),
_delta(3, bids=[(10_000, 0)], asks=[]),
]
)
with pytest.raises(ResearchDataError, match="stale/gapped sequence"):
build_cancellation_intensity_features(frame, windows=(2,))
|