feat: implement read-only Polymarket data collection layer with market discovery and price fetching

src/data_collection/polymarket_readonly.py

@@ -690,9 +690,15 @@ class PolymarketReadOnlyLayer:
                 bucket["no_buy"] = no_buy
             if no_sell is not None:
                 bucket["no_sell"] = no_sell
-            if yes_midpoint is not None:
-                bucket["market_price"] = yes_midpoint
-                bucket["probability"] = yes_midpoint
+            reference_price = yes_midpoint
+            if reference_price is None and yes_buy is not None and yes_sell is not None:
+                reference_price = (yes_buy + yes_sell) / 2.0
+            if reference_price is None:
+                reference_price = yes_buy if yes_buy is not None else yes_sell
+            if reference_price is not None:
+                reference_price = max(0.0, min(1.0, float(reference_price)))
+                bucket["market_price"] = reference_price
+                bucket["probability"] = reference_price
             if yes_prices.get("quote_source"):
                 bucket["quote_source"] = yes_prices.get("quote_source")
             if yes_prices.get("quote_age_ms") is not None:
@@ -819,7 +825,6 @@ class PolymarketReadOnlyLayer:
                 ended_at = parsed
                 break
 
-        now_utc = datetime.now(timezone.utc)
         tradable = True
         reason = None
         if closed:
@@ -831,9 +836,6 @@ class PolymarketReadOnlyLayer:
         elif accepting_orders is False:
             tradable = False
             reason = "not_accepting_orders"
-        elif ended_at is not None and ended_at <= now_utc:
-            tradable = False
-            reason = "past_end_time"
 
         return {
             "active": active,
@@ -1397,6 +1399,8 @@ class PolymarketReadOnlyLayer:
                     "sell": sell,
                     "midpoint": midpoint,
                     "last_trade_price": last_trade,
+                    "quote_source": "polymarket_clob_client",
+                    "quote_age_ms": 0,
                     "book": book,
                     "book_liquidity": book_liquidity,
                 }
@@ -1420,6 +1424,8 @@ class PolymarketReadOnlyLayer:
             "sell": sell,
             "midpoint": midpoint,
             "last_trade_price": last_trade,
+            "quote_source": "polymarket_clob_rest",
+            "quote_age_ms": 0,
             "book": book,
             "book_liquidity": book_liquidity,
         }
@@ -1635,6 +1641,8 @@ class PolymarketReadOnlyLayer:
                     no_sell = max(0.0, min(1.0, 1.0 - yes_sell))
 
                 market_slug = str(market.get("slug") or "").strip()
+                row_yes_token_id = str(yes_token.get("token_id") or "").strip()
+                row_no_token_id = str(no_token.get("token_id") or "").strip()
                 top_rows.append(
                     {
                         "label": self._extract_market_bucket_label(market, bucket_temp),
@@ -1646,8 +1654,8 @@ class PolymarketReadOnlyLayer:
                         "yes_sell": yes_sell,
                         "no_buy": no_buy,
                         "no_sell": no_sell,
-                        "yes_token_id": yes_token_id or None,
-                        "no_token_id": no_token_id or None,
+                        "yes_token_id": row_yes_token_id or None,
+                        "no_token_id": row_no_token_id or None,
                         "quote_source": yes_prices.get("quote_source"),
                         "quote_age_ms": _safe_int(yes_prices.get("quote_age_ms"), 0),
                         "slug": market_slug or None,

tests/test_polymarket_readonly.py

@@ -59,6 +59,7 @@ def test_fetch_token_market_data_prefers_orderbook_executable_prices():
     assert data["sell"] == 0.24
     assert data["midpoint"] == 0.5
     assert data["last_trade_price"] == 0.49
+    assert data["quote_source"] == "polymarket_clob_client"
 
 
 def test_get_token_market_data_prefers_fresh_ws_cache():
@@ -117,6 +118,56 @@ def test_price_analysis_computes_edge_kelly_and_lock():
     assert analysis["best_side"] == "yes"
 
 
+def test_trade_state_keeps_open_markets_tradable_after_gamma_end_date():
+    layer = PolymarketReadOnlyLayer()
+
+    state = layer._market_trade_state(
+        {
+            "active": True,
+            "closed": False,
+            "acceptingOrders": True,
+            "endDate": "2020-01-01T00:00:00Z",
+        }
+    )
+
+    assert state["tradable"] is True
+    assert state["reason"] is None
+    assert state["ended_at_utc"] == "2020-01-01T00:00:00+00:00"
+
+
+def test_hydrate_bucket_prices_uses_executable_quotes_without_midpoint():
+    layer = PolymarketReadOnlyLayer()
+    buckets = [
+        {
+            "temp": 14.0,
+            "yes_token_id": "yes-token",
+            "no_token_id": "no-token",
+        }
+    ]
+
+    def _fake_get_token_market_data(token_id):
+        if token_id == "yes-token":
+            return {
+                "buy": 0.66,
+                "sell": 0.70,
+                "quote_source": "polymarket_clob_rest",
+                "quote_age_ms": 0,
+            }
+        return {"buy": 0.30, "sell": 0.36}
+
+    layer._get_token_market_data = _fake_get_token_market_data
+
+    layer._hydrate_bucket_prices(buckets)
+
+    assert buckets[0]["yes_buy"] == 0.66
+    assert buckets[0]["yes_sell"] == 0.70
+    assert buckets[0]["no_buy"] == 0.30
+    assert buckets[0]["no_sell"] == 0.36
+    assert round(buckets[0]["market_price"], 6) == 0.68
+    assert round(buckets[0]["probability"], 6) == 0.68
+    assert buckets[0]["quote_source"] == "polymarket_clob_rest"
+
+
 def test_build_top_temperature_buckets_dedupes_same_temperature():
     layer = PolymarketReadOnlyLayer()
 
@@ -195,6 +246,11 @@ def test_build_top_temperature_buckets_dedupes_same_temperature():
     )
 
     values = [row.get("value") for row in rows]
+    token_ids = [row.get("yes_token_id") for row in rows]
     assert len(values) == len(set(values))
+    assert len(token_ids) == len(set(token_ids))
     assert rows[0]["value"] == 14.0
+    assert rows[0]["yes_token_id"] == (
+        "highest-temperature-in-ankara-on-march-12-2026-14c-or-higher|yes"
+    )
     assert all(not str(row.get("label") or "").startswith("<=") for row in rows)