feat(fitness): multi-objective combined = alpha*IS + (1-alpha)*OOS opt-in

Aggiunti due meccanismi per selection multi-objective:

1) Helper compute_combined_fitness(fit_train, fit_oos, alpha):
   formula = alpha*IS + (1-alpha)*OOS, fallback a IS se OOS è None/NaN.

2) RunConfig.eval_oos_during_loop (default False) + fitness_combined_alpha
   (default 0.5). Quando True E wfa_train_split attivo, ogni genome con
   fitness IS > 0 viene rivalutato su test_ohlcv DURANTE il loop GA e la
   fitness usata per tournament_select/elite_select è quella combinata.
   2x costo backtest engine (richiede 2 evaluation per genome).

3) CLI flags --eval-oos-during-loop e --fitness-combined-alpha.

Motivazione: il run phase2-7-max7y-v2-wfa-001 ha mostrato che il top by
fitness_IS (4e1be9fa, ratio OOS 0.31) NON è il top per performance OOS reale
(634111992702, ratio 1.42, ret_OOS +105% / 2.2y). Selezionare durante GA
con combined fitness orienta l'evoluzione verso strategie OOS-robust by
design invece di filtrarle a posteriori.

Backward compat: default eval_oos_during_loop=False → comportamento
invariato per run senza il flag.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

scripts/run_phase1.py

@@ -81,6 +81,21 @@ def parse_args() -> argparse.Namespace:
         default=5,
         help="Walk-forward: quanti top genomi rivalutare OOS (default 5)",
     )
+    p.add_argument(
+        "--eval-oos-during-loop",
+        action="store_true",
+        help=(
+            "Multi-objective: eval ogni genome anche su test_ohlcv durante "
+            "il loop e usa combined = alpha*IS + (1-alpha)*OOS per selection. "
+            "Richiede --wfa-train-split. 2x costo backtest engine."
+        ),
+    )
+    p.add_argument(
+        "--fitness-combined-alpha",
+        type=float,
+        default=0.5,
+        help="Multi-objective: peso IS (1-alpha = OOS). 1.0=solo IS, 0.5=bilanciato, 0.0=solo OOS",
+    )
     return p.parse_args()
 
 
@@ -144,6 +159,8 @@ def main() -> None:
         fitness_adversarial_soft_penalty=args.fitness_soft_penalty,
         wfa_train_split=args.wfa_train_split,
         wfa_top_k=args.wfa_top_k,
+        eval_oos_during_loop=args.eval_oos_during_loop,
+        fitness_combined_alpha=args.fitness_combined_alpha,
     )
 
     run_id = run_phase1(cfg, ohlcv=ohlcv, llm=llm)

src/multi_swarm/ga/fitness.py

@@ -32,6 +32,26 @@ from ..agents.adversarial import AdversarialReport, Severity
 from ..agents.falsification import FalsificationReport
 
 
+def compute_combined_fitness(
+    fitness_train: float,
+    fitness_oos: float | None,
+    alpha: float = 0.5,
+) -> float:
+    """Combina fitness IS e OOS in uno scalare per selection multi-objective.
+
+    Formula::
+
+        combined = alpha * fitness_train + (1 - alpha) * fitness_oos
+
+    Se ``fitness_oos`` è ``None`` o NaN, ritorna ``fitness_train`` (fallback).
+    alpha=1.0 → solo IS (= comportamento default). alpha=0.0 → solo OOS.
+    alpha=0.5 → bilanciato.
+    """
+    if fitness_oos is None or fitness_oos != fitness_oos:  # noqa: PLR0124  (NaN check)
+        return fitness_train
+    return alpha * fitness_train + (1.0 - alpha) * fitness_oos
+
+
 def compute_fitness(
     falsification: FalsificationReport,
     adversarial: AdversarialReport,

src/multi_swarm/orchestrator/run.py

@@ -61,6 +61,12 @@ class RunConfig:
     # dei top genomi sui restanti. None/0 = no WFA (eval full ohlcv).
     wfa_train_split: float | None = None
     wfa_top_k: int = 5  # quanti top genomi rivalutare OOS
+    # Multi-objective selection: se True, ogni genome viene valutato anche su
+    # test_ohlcv durante il loop e la fitness usata per tournament/elite è
+    # combined = alpha*IS + (1-alpha)*OOS. Richiede wfa_train_split attivo.
+    # 2x costo backtest engine.
+    eval_oos_during_loop: bool = False
+    fitness_combined_alpha: float = 0.5  # peso IS (1-alpha = OOS)
 
 
 def run_phase1(
@@ -176,6 +182,31 @@ def run_phase1(
                     hard_kill_findings=cfg.fitness_hard_kill_findings,
                     adversarial_soft_penalty=cfg.fitness_adversarial_soft_penalty,
                 )
+                # Multi-objective: se attivo, eval OOS subito e combina via alpha.
+                if (
+                    cfg.eval_oos_during_loop
+                    and test_ohlcv is not None
+                    and len(test_ohlcv) >= 100
+                    and fit > 0
+                ):
+                    try:
+                        fals_oos_inloop = falsification_agent.evaluate(
+                            proposal.strategy, test_ohlcv
+                        )
+                        adv_oos_inloop = adversarial_agent.review(
+                            proposal.strategy, test_ohlcv
+                        )
+                        fit_oos_inloop = compute_fitness(
+                            fals_oos_inloop, adv_oos_inloop,
+                            hard_kill_findings=cfg.fitness_hard_kill_findings,
+                            adversarial_soft_penalty=cfg.fitness_adversarial_soft_penalty,
+                        )
+                        fit = (
+                            cfg.fitness_combined_alpha * fit
+                            + (1.0 - cfg.fitness_combined_alpha) * fit_oos_inloop
+                        )
+                    except Exception:  # noqa: BLE001
+                        pass  # fallback: usa solo IS
                 repo.save_evaluation(
                     run_id=run_id,
                     genome_id=genome.id,