first commit

tests/test_opportunity_engine.py

@@ -0,0 +1,115 @@
+"""Opportunity engine math tests."""
+from __future__ import annotations
+
+from decimal import Decimal as D
+
+from arbitrage.book.l2 import BookRegistry
+from arbitrage.engine.opportunity import EngineConfig, EventIndex, OpportunityEngine
+from arbitrage.models import Event, Outcome
+
+
+def _two_outcome_event() -> Event:
+    return Event(
+        id="e1",
+        slug="e1",
+        title="two-outcome",
+        is_neg_risk=True,
+        end_date=None,
+        outcomes=(
+            Outcome(token_id="A", name="A", outcome_index=0),
+            Outcome(token_id="B", name="B", outcome_index=1),
+        ),
+    )
+
+
+def _engine(
+    reg: BookRegistry,
+    index: EventIndex,
+    *,
+    min_bps: int = 50,
+    max_basket_usd: D = D("500"),
+) -> OpportunityEngine:
+    return OpportunityEngine(
+        books=reg,
+        index=index,
+        config=EngineConfig(
+            min_net_edge_bps=min_bps,
+            fees_per_share_usd=D("0"),
+            gas_per_basket_usd=D("0.10"),
+            max_basket_usd=max_basket_usd,
+        ),
+    )
+
+
+class TestEvaluate:
+    def test_detects_arb_when_asks_sum_below_one(self) -> None:
+        ev = _two_outcome_event()
+        idx = EventIndex()
+        idx.upsert(ev)
+        reg = BookRegistry()
+        reg.apply_snapshot("A", bids=[], asks=[(D("0.40"), D("200"))])
+        reg.apply_snapshot("B", bids=[], asks=[(D("0.50"), D("200"))])
+        opp = _engine(reg, idx).evaluate(ev)
+        assert opp is not None
+        assert opp.event_id == "e1"
+        assert opp.net_edge_bps >= 50
+        assert opp.max_baskets > 0
+
+    def test_rejects_when_asks_sum_above_one(self) -> None:
+        ev = _two_outcome_event()
+        idx = EventIndex()
+        idx.upsert(ev)
+        reg = BookRegistry()
+        reg.apply_snapshot("A", bids=[], asks=[(D("0.70"), D("200"))])
+        reg.apply_snapshot("B", bids=[], asks=[(D("0.50"), D("200"))])
+        assert _engine(reg, idx).evaluate(ev) is None
+
+    def test_rejects_when_any_leg_has_no_asks(self) -> None:
+        ev = _two_outcome_event()
+        idx = EventIndex()
+        idx.upsert(ev)
+        reg = BookRegistry()
+        reg.apply_snapshot("A", bids=[], asks=[])
+        reg.apply_snapshot("B", bids=[], asks=[(D("0.50"), D("200"))])
+        assert _engine(reg, idx).evaluate(ev) is None
+
+    def test_rejects_when_edge_below_threshold(self) -> None:
+        ev = _two_outcome_event()
+        idx = EventIndex()
+        idx.upsert(ev)
+        reg = BookRegistry()
+        # Sum = 0.995 -> 50bps gross, net will be below 50 after gas
+        reg.apply_snapshot("A", bids=[], asks=[(D("0.495"), D("200"))])
+        reg.apply_snapshot("B", bids=[], asks=[(D("0.500"), D("200"))])
+        assert _engine(reg, idx, min_bps=50).evaluate(ev) is None
+
+    def test_depth_clips_basket_count(self) -> None:
+        ev = _two_outcome_event()
+        idx = EventIndex()
+        idx.upsert(ev)
+        reg = BookRegistry()
+        # Leg A is very thin
+        reg.apply_snapshot("A", bids=[], asks=[(D("0.40"), D("12"))])
+        reg.apply_snapshot("B", bids=[], asks=[(D("0.50"), D("10000"))])
+        opp = _engine(reg, idx).evaluate(ev)
+        assert opp is not None
+        assert opp.max_baskets <= D("12")
+
+    def test_vwap_degrades_with_size(self) -> None:
+        """At large sizes we consume worse levels; edge per basket must shrink."""
+        ev = _two_outcome_event()
+        idx = EventIndex()
+        idx.upsert(ev)
+        reg = BookRegistry()
+        # Both legs: cheap top, expensive deep levels
+        reg.apply_snapshot(
+            "A", bids=[], asks=[(D("0.40"), D("10")), (D("0.48"), D("10000"))]
+        )
+        reg.apply_snapshot(
+            "B", bids=[], asks=[(D("0.50"), D("10")), (D("0.52"), D("10000"))]
+        )
+        opp = _engine(reg, idx).evaluate(ev)
+        assert opp is not None
+        # At size 10, sum_vwap = 0.90. At larger sizes it'll rise toward 1.00.
+        # The engine chose the size maximizing expected profit, so sum is <= 1.0.
+        assert opp.sum_vwap_asks <= D("1.0")