feat(backtest): add Sortino, Gain-to-Pain and dollar profit factor

Three portfolio metrics computed where their inputs already live in
_simulate_portfolio: Sortino off the existing daily return series, Gain-to-Pain
off a monthly aggregation of the equity curve, profit factor off closed-trade
dollar P&L.

Gain-to-Pain follows Schwager — sum of ALL monthly returns over the absolute
sum of the negative ones. The profit-factor-shaped variant,
sum(positive)/|sum(negative)|, sits exactly 1.0 higher for every input since
sum(all) = sum(pos) - |sum(neg)|; the test asserts against both so the wrong one
cannot pass. Sortino divides by len(rets), the full-sample lower partial moment,
not by the count of down days, which would shrink the denominator and inflate
the ratio.

No MAR field: calmar is already CAGR / max drawdown, the same number under the
other name (docs/research/effective-risk-floor-ab.md).

All three keys are emitted unconditionally even when None — the UI reads an
absent key as "report predates these metrics", so presence is a contract.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

app/services/backtest_service.py

@@ -2625,6 +2625,19 @@ def _simulate_portfolio(
     diag = sharpe_diagnostics(rets)
     sharpe = diag["sharpe"]
 
+    # Sortino: the same numerator as Sharpe over downside deviation about a zero
+    # target. The denominator divides by len(rets) — the full-sample lower partial
+    # moment — NOT by the count of down days, which would shrink the denominator
+    # and inflate the ratio. n >= 3 matches sharpe_diagnostics so the two appear
+    # together or not at all. No down days is +inf, reported as None.
+    sortino = None
+    downside = [r for r in rets if r < 0.0]
+    if len(rets) >= 3 and downside:
+        mean_ret = sum(rets) / len(rets)
+        dd = math.sqrt(sum(r * r for r in downside) / len(rets))
+        if dd > 0:
+            sortino = round(mean_ret / dd * math.sqrt(252.0), 2)
+
     # Per-calendar-year returns off the equity curve — shows whether every year
     # contributed or one exceptional stretch carried the result.
     yearly: list[dict] = []
@@ -2650,8 +2663,40 @@ def _simulate_portfolio(
         ),
     })
 
+    # Gain-to-Pain off the same curve, on MONTHLY returns: Schwager's ratio is
+    # defined monthly and the daily variant is not comparable to published
+    # figures. Distinct loop variables from the yearly pass above — that one exits
+    # with last_eq at final equity, so reusing its names silently corrupts the
+    # first month. The monthly series itself is not emitted: 36-120 floats per
+    # strategy per lookback would bloat the single stored report blob.
+    monthly: list[float] = []
+    month_start_eq = curve[0][1]
+    month_last_eq = curve[0][1]
+    cur_month = date.fromordinal(curve[0][0]).replace(day=1)
+    for o, eq in curve:
+        m = date.fromordinal(o).replace(day=1)
+        if m != cur_month:
+            if month_start_eq > 0:
+                monthly.append(month_last_eq / month_start_eq - 1.0)
+            cur_month = m
+            month_start_eq = month_last_eq
+        month_last_eq = eq
+    if month_start_eq > 0:
+        monthly.append(month_last_eq / month_start_eq - 1.0)
+
+    # Schwager: SUM OF ALL monthly returns over the absolute sum of the negative
+    # ones. Not sum(positive)/|sum(negative)| — that is profit-factor-shaped and
+    # sits exactly 1.0 higher for every input, since sum(all) = sum(pos) - |sum(neg)|.
+    monthly_pain = -sum(r for r in monthly if r < 0.0)
+    gain_to_pain = round(sum(monthly) / monthly_pain, 2) if monthly_pain > 0 else None
+
     pnls = [t["pnl"] for t in trades]
     wins = sum(1 for p in pnls if p > 0)
+    # Dollar-based, over closed-trade P&L. Distinct from the R-based profit_factor
+    # in _robustness_stats; the two never share an object.
+    gross_win = sum(p for p in pnls if p > 0)
+    gross_loss = -sum(p for p in pnls if p < 0)
+    profit_factor = round(gross_win / gross_loss, 2) if gross_loss > 0 else None
     reason_counts = {
         reason: sum(1 for t in trades if t["reason"] == reason)
         for reason in sorted({t["reason"] for t in trades})
@@ -2706,7 +2751,13 @@ def _simulate_portfolio(
         "total_return_pct": round(total_return_pct, 1),
         "cagr_pct": round(cagr_pct, 1) if cagr_pct is not None else None,
         "max_drawdown_pct": round(max_dd_pct, 1),
+        # calmar IS MAR here (CAGR / max drawdown) — one field, two names.
         "calmar": round(calmar, 2) if calmar is not None else None,
+        # Emitted unconditionally even when None: the UI treats an ABSENT key as
+        # "report predates these metrics", so presence is a contract.
+        "sortino": sortino,
+        "gain_to_pain": gain_to_pain,
+        "profit_factor": profit_factor,
         "sharpe": sharpe,
         "sharpe_se": diag["sharpe_se"],
         "psr": diag["psr"],