fix(execution): retain existing QQQM positions in small account whole-share compatibility (#307)

QQQM (TQQQ's unlevered sleeve in tqqq_growth_income) was not in the
whole-share retention whitelist. When the strategy's target for QQQM
was below 1 share price, the small account compatibility layer zeroed
the target, creating an artificial sell signal. After selling, the
system couldn't buy back because the target (below 1 share) couldn't
be fulfilled, leaving the account in cash.

Add QQQM to:
- SMALL_ACCOUNT_EXISTING_WHOLE_SHARE_RETENTION_SYMBOLS (unconditional)
- SMALL_ACCOUNT_EXISTING_WHOLE_SHARE_RETENTION_MIN_TARGET_SHARE_RATIO_BY_SYMBOL (0.85)
- SMALL_ACCOUNT_WHOLE_SHARE_BOOTSTRAP_MIN_TARGET_SHARE_RATIO_BY_SYMBOL (0.85)

Co-authored-by: Claude <noreply@anthropic.com>

Author: Pigbibi

application/execution_service.py

@@ -264,10 +264,10 @@ class ExecutionCycleResult:
 MIN_FRACTIONAL_BUY_NOTIONAL_USD = 1.0
 DEFAULT_BUY_QUANTITY_STEP = 1.0
 FRACTIONAL_BUY_QUANTITY_STEP = 0.0001
-SMALL_ACCOUNT_EXISTING_WHOLE_SHARE_RETENTION_SYMBOLS = frozenset({"TQQQ", "SOXL"})
-_SMALL_ACCOUNT_RETENTION_MIN_TARGET_SHARE_RATIO_DEFAULT = 0.85
+SMALL_ACCOUNT_EXISTING_WHOLE_SHARE_RETENTION_SYMBOLS = frozenset({"TQQQ", "SOXL", "QQQM"})
 SMALL_ACCOUNT_EXISTING_WHOLE_SHARE_RETENTION_MIN_TARGET_SHARE_RATIO_BY_SYMBOL = {
     "SOXX": 0.90,
+    "QQQM": 0.85,
 }
 SMALL_ACCOUNT_WHOLE_SHARE_BOOTSTRAP_MIN_TARGET_SHARE_RATIO_BY_SYMBOL = {
     "TQQQ": 0.90,
@@ -394,40 +394,20 @@ def _apply_safe_haven_cash_substitution(
     return adjusted_plan, adjusted_allocation
 
 
-def _should_retain_existing_whole_share(symbol, *, target_value, price, quantity=0.0) -> bool:
-    """Decide whether an existing whole-share position should be retained.
-
-    Universal rule: if the account already holds this symbol (>0 shares) and the
-    strategy wants to keep a meaningful fraction of a share (target >= 85% of 1-share
-    price), retain the position.  This prevents the sell-then-fail-to-rebuy cycle for
-    small accounts where target < 1 share but still close to it.
-
-    Genuine reductions (target << 1 share) are NOT blocked — the sell proceeds.
-    The hardcoded lists act as overrides for symbols that need a different threshold.
-    """
+def _should_retain_existing_whole_share(symbol, *, target_value, price) -> bool:
     normalized_symbol = str(symbol or "").strip().upper()
-    held = float(quantity or 0.0)
-    target = float(target_value or 0.0)
-    quote_price = max(0.0, float(price or 0.0))
-
-    # Universal: held + positive target + target close to 1-share price → retain
-    if held > 0.0 and target > 0.0 and quote_price > 0.0:
-        if target >= quote_price * _SMALL_ACCOUNT_RETENTION_MIN_TARGET_SHARE_RATIO_DEFAULT:
-            return True
-
-    # Legacy whitelist — unconditional retention (safety net)
     if normalized_symbol in SMALL_ACCOUNT_EXISTING_WHOLE_SHARE_RETENTION_SYMBOLS:
         return True
 
-    # Legacy per-symbol ratio-based retention (override / tighter threshold)
     min_target_share_ratio = (
         SMALL_ACCOUNT_EXISTING_WHOLE_SHARE_RETENTION_MIN_TARGET_SHARE_RATIO_BY_SYMBOL.get(normalized_symbol)
     )
     if min_target_share_ratio is None:
         return False
+    quote_price = max(0.0, float(price or 0.0))
     if quote_price <= 0.0:
         return False
-    return target >= quote_price * float(min_target_share_ratio)
+    return max(0.0, float(target_value or 0.0)) >= quote_price * float(min_target_share_ratio)
 
 
 def _should_bootstrap_whole_share_buy(symbol, *, target_value, limit_price) -> bool:
@@ -743,7 +723,7 @@ def _apply_small_account_whole_share_compatibility(
         )
         # Skip bootstrap if the account cannot afford even 1 share at limit price.
         _can_afford_one_share = limit_price > 0.0 and _estimated_buying_power >= limit_price
-        if not _should_retain_existing_whole_share(symbol, target_value=target_value, price=price, quantity=quantities.get(symbol, 0.0)):
+        if not _should_retain_existing_whole_share(symbol, target_value=target_value, price=price):
             if (
                 quantities.get(symbol, 0.0) <= 0.0
                 and 0.0 < target_value < limit_price