phase 17: pass all tracked tickers to extractor, soften prompt for macro-to-company relevance

2026-04-12 09:18:08 -07:00
parent 59f89d03d2
commit 57d0fc7d33
6 changed files with 142 additions and 4 deletions
@@ -0,0 +1,44 @@
+from minio import Minio
+import os, json
+
+mc = Minio(os.environ["MINIO_ENDPOINT"], access_key=os.environ["MINIO_ACCESS_KEY"], secret_key=os.environ["MINIO_SECRET_KEY"], secure=False)
+raw_objs = list(mc.list_objects("stonks-llm-results", recursive=True))
+
+# Count companies found vs not found
+total = 0
+with_companies = 0
+empty_companies = 0
+wrong_tickers = 0
+our_tickers = {"AAPL","MSFT","NVDA","AMZN","GOOGL","JPM","JNJ","XOM","TSLA","META"}
+ticker_hits = {}
+
+for o in raw_objs:
+    data = json.loads(mc.get_object("stonks-llm-results", o.object_name).read())
+    if not data.get("success"):
+        continue
+    attempts = data.get("attempts", [])
+    if not attempts:
+        continue
+    raw_out = attempts[-1].get("raw_output", "")
+    try:
+        parsed = json.loads(raw_out)
+        companies = parsed.get("companies", [])
+        total += 1
+        if companies:
+            with_companies += 1
+            for c in companies:
+                t = c.get("ticker", "")
+                if t in our_tickers:
+                    ticker_hits[t] = ticker_hits.get(t, 0) + 1
+                else:
+                    wrong_tickers += 1
+        else:
+            empty_companies += 1
+    except:
+        pass
+
+print(f"Total successful extractions: {total}")
+print(f"  With companies: {with_companies}")
+print(f"  Empty companies: {empty_companies}")
+print(f"  Wrong/unknown tickers: {wrong_tickers}")
+print(f"  Our ticker hits: {ticker_hits}")