Add sma_spread, the golden-cross distance in percent

sma50 > sma200 says which side of the cross a name is on but not how
far — a name 0.1% above its 200-day and one 15% above it both just
read "true". sma_spread turns that into a sortable percent, so a
golden-cross screen can be ranked by how developed the cross is
instead of only filtered on the boolean.

It depends on the two moving averages alone, not on today's price, so
unlike pct_from_sma50/200 it classifies as STATIC_SAFE and narrows
before a screen re-prices.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AmxGXcSkCKBA9i2sQDoQBk

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"]
+           "true_range", "atr_pct", "sma_spread"]
 COLUMNS = [n for n, _ in store.UNIVERSE_COLUMNS] + DERIVED
 
 TEXT_COLUMNS = {"symbol", "name", "sector", "industry", "country",
@@ -80,6 +80,10 @@
     "week52_low", "week52_change_pct", "sma50", "sma200", "avg_volume_3m",
     "avg_volume_10d", "analyst_rating", "earnings_ts", "earnings_days",
     "earnings_when", "peg", "prev_close", "is_adr",
+    # Both inputs are themselves slow-moving averages a snapshot old by a
+    # few hours does not change; unlike `pct_from_sma50` it does not also
+    # depend on today's price.
+    "sma_spread",
     # Average dollar volume is a liquidity floor, not a reading of the
     # tape: `avgdvol > 20m` means "this name normally trades enough to get
     # out of", and half a day of price drift does not change that answer.
@@ -105,6 +109,7 @@
     "from52high": "pct_from_52w_high", "from52low": "pct_from_52w_low",
     "fromsma50": "pct_from_sma50", "fromsma200": "pct_from_sma200",
     "sma50pct": "pct_from_sma50", "sma200pct": "pct_from_sma200",
+    "smaspread": "sma_spread", "crossdist": "sma_spread",
     "rating": "analyst_rating", "earnings": "earnings_days",
     "ipo": "ipo_year", "type": "quote_type", "state": "market_state",
     "lag": "lag_seconds",

app/universe.py

@@ -306,6 +306,11 @@ def safe(a, b):
         df["price"] - df["week52_low"], df["week52_low"]) * 100
     df["pct_from_sma50"] = safe(df["price"] - df["sma50"], df["sma50"]) * 100
     df["pct_from_sma200"] = safe(df["price"] - df["sma200"], df["sma200"]) * 100
+    # SMA50 vs SMA200 as a percent apart — the golden-cross distance.
+    # `sma50 > sma200` alone only says which side of the cross a name is
+    # on; this is how far, so a screen can sort names by how developed the
+    # cross is rather than just filtering on the boolean.
+    df["sma_spread"] = safe(df["sma50"] - df["sma200"], df["sma200"]) * 100
     # Relative strength against a stock's own sector. "Up 3%" on a day the
     # whole sector rose 3% is not strength; this separates the move from
     # the tide. Cap-weighted, because an unweighted sector average is

tests/test_screen.py

@@ -498,6 +498,26 @@ def when_for(hour, minute=0):
     check("atr_pct needs today's range and price, so it isn't prefiltered on a stale one",
           "atr_pct" in screen.LIVE_COLUMNS and "atr_pct" not in screen.STATIC_SAFE)
 
+    print("\nsma spread")
+    ss = derive(pd.DataFrame({
+        "symbol": ["CROSSED", "BELOW"], "name": ["C", "B"], "sector": ["T"] * 2,
+        "price": [100.0, 100.0], "change_pct": [0.0] * 2,
+        "volume": [1e6] * 2, "avg_volume_3m": [1e6] * 2, "market_cap": [1e9] * 2,
+        "week52_high": [110.0] * 2, "week52_low": [90.0] * 2,
+        "sma50": [110.0, 95.0], "sma200": [100.0, 100.0],
+        "earnings_ts": [np.nan] * 2,
+    }))
+    ssrow = dict(zip(ss["symbol"], ss.to_dict("records"), strict=True))
+    check("above the 200-day: spread is positive and sized",
+          abs(ssrow["CROSSED"]["sma_spread"] - 10.0) < 1e-9,
+          ssrow["CROSSED"]["sma_spread"])
+    check("below the 200-day: spread is negative",
+          abs(ssrow["BELOW"]["sma_spread"] - (-5.0)) < 1e-9,
+          ssrow["BELOW"]["sma_spread"])
+    check("smaspread resolves", screen.resolve("smaspread") == "sma_spread")
+    check("sma_spread depends only on two slow-moving averages, so it narrows safely stale",
+          "sma_spread" in screen.STATIC_SAFE and "sma_spread" not in screen.LIVE_COLUMNS)
+
     # Today's dollar volume only rises, so prefiltering on a stale one would
     # drop the names that have since crossed the line. The average is a
     # standing fact about the name and narrows safely.