Add intraday percent, the move since today's open

chg measures against yesterday's close, so a stock that gapped up 5%
and then sold off all day still reads as unchanged if it lands back at
yesterday's price. gap_pct only sees the overnight jump. Neither one
tells you what happened *during* the session. intraday_pct fills that
gap: (price - open) / open, screenable as `intraday` or `sinceopen`.

Live-priced like gap_pct, since it depends on today's open and the
live price ticking against it; null rather than zero when open is
missing.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_017b8HGHfwwreddySUhn7Reh

Author: Ben05-sys

app/screen.py

@@ -36,7 +36,7 @@
            "earnings_when", "peg", "mktcap", "chg", "is_adr",
            "rs_sector", "sector_change_pct",
            "dollar_volume", "avg_dollar_volume", "day_range_pct", "gap_pct",
-           "true_range", "atr_pct", "sma_spread"]
+           "true_range", "atr_pct", "sma_spread", "intraday_pct"]
 COLUMNS = [n for n, _ in store.UNIVERSE_COLUMNS] + DERIVED
 
 TEXT_COLUMNS = {"symbol", "name", "sector", "industry", "country",
@@ -58,7 +58,7 @@
     # moves with all three of its.
     "dollar_volume", "avg_dollar_volume", "day_range_pct",
     # Built from today's open, which a stale snapshot may not have yet.
-    "gap_pct",
+    "gap_pct", "intraday_pct",
     # Built from today's volume, which only rises as the session runs.
     "rel_volume_10d",
     # Built from today's high/low, which only widen as the session runs.
@@ -128,6 +128,7 @@
     "liquidity": "avg_dollar_volume",
     "dayrange": "day_range_pct", "rangepos": "day_range_pct",
     "inrange": "day_range_pct", "gap": "gap_pct",
+    "intraday": "intraday_pct", "sinceopen": "intraday_pct",
     "tr": "true_range", "truerange": "true_range",
     "atr": "atr_pct", "atrpct": "atr_pct",
     "when": "earnings_when", "earningswhen": "earnings_when",

app/universe.py

@@ -281,6 +281,15 @@ def safe(a, b):
         df["gap_pct"] = safe(df["open"] - df["prev_close"], df["prev_close"]) * 100
     else:
         df["gap_pct"] = np.nan
+    # The move since *today's open*, not yesterday's close. `chg` conflates
+    # the overnight gap with what has happened during the session; a name
+    # can gap up 5% and then sell off all day, which is a very different
+    # stock from one that opened flat and has run up 5% since. Splitting
+    # the two is what `gap_pct` (overnight) and this (intraday) are for.
+    if "open" in df.columns:
+        df["intraday_pct"] = safe(df["price"] - df["open"], df["open"]) * 100
+    else:
+        df["intraday_pct"] = np.nan
     # The real daily range, in dollars: `day_high - day_low` alone misses
     # the overnight gap, so a stock that gapped down hard and then sat
     # still reads as barely having moved. skipna=False so a missing input

tests/test_screen.py

@@ -455,6 +455,31 @@ def when_for(hour, minute=0):
               "sma50": [1.0], "sma200": [1.0], "earnings_ts": [np.nan],
           }))["gap_pct"].isna().all())
 
+    print("\nintraday percent")
+    check("gapped up but sold off since the open reads negative",
+          abs(row["CHEAP"]["intraday_pct"] - (-100 / 21)) < 1e-9,
+          row["CHEAP"]["intraday_pct"])
+    check("distinct from chg: CHEAP is flat on the day yet moved intraday",
+          row["CHEAP"]["chg"] == 0.0 and row["CHEAP"]["intraday_pct"] != 0.0,
+          (row["CHEAP"]["chg"], row["CHEAP"]["intraday_pct"]))
+    check("gapped down but ran since the open reads positive",
+          abs(row["DEAR"]["intraday_pct"] - 300 / 97) < 1e-9,
+          row["DEAR"]["intraday_pct"])
+    check("opened at price is a real zero, not null",
+          row["FLAT"]["intraday_pct"] == 0.0, row["FLAT"]["intraday_pct"])
+    check("intraday resolves", screen.resolve("intraday") == "intraday_pct")
+    check("sinceopen resolves too", screen.resolve("sinceopen") == "intraday_pct")
+    check("intraday needs a live price and today's open, so it isn't prefiltered stale",
+          "intraday_pct" in screen.LIVE_COLUMNS
+          and "intraday_pct" not in screen.STATIC_SAFE)
+    check("a frame with no open derives rather than raising",
+          derive(pd.DataFrame({
+              "symbol": ["X"], "name": ["X"], "sector": ["T"], "price": [1.0],
+              "change_pct": [0.0], "volume": [1.0], "avg_volume_3m": [1.0],
+              "market_cap": [1.0], "week52_high": [1.0], "week52_low": [1.0],
+              "sma50": [1.0], "sma200": [1.0], "earnings_ts": [np.nan],
+          }))["intraday_pct"].isna().all())
+
     print("\ntrue range")
     tr = derive(pd.DataFrame({
         "symbol": ["NORMAL", "GAPPED"], "name": ["N", "G"], "sector": ["T"] * 2,