Forward test: record forecasts before their outcomes exist.

`snapshot` writes each tracked model's forecast quantiles at the seven
backtest horizons from the latest price to data/forecasts/<origin>.csv. It
refuses stale data (older than two days) and duplicate dates, so snapshots
can't be reconstructed after the fact; committing them dates them.
`forward` scores every recorded forecast whose target date has passed,
reusing the backtest's scoring (now factored out as evaluate.score).

Tracked: random_walk, drift_rw, cycle, powerlaw, plus powerlaw_ou,
powerlaw_ou_param and cycle_on_powerlaw, which development data couldn't
settle. `just weekly` runs update, snapshot and forward.

First snapshot: 2026-09-23 (BTC $84.4K). The first outcomes are due
2026-10-23.

btcmodel/__main__.py

@@ -6,6 +6,8 @@ Command line entry point.
     python -m btcmodel backtest --holdout  score models on outcomes after DEV_CUTOFF
     python -m btcmodel forecast            forecast from the latest price
     python -m btcmodel ab [NAME ...]       run A/B experiments on development data
+    python -m btcmodel snapshot            record tracked models' forecasts for the forward test
+    python -m btcmodel forward             score recorded forecasts whose targets have passed
 """
 
 import argparse
@@ -14,7 +16,7 @@ from pathlib import Path
 import numpy as np
 import pandas as pd
 
-from . import data, evaluate, experiments, plots
+from . import data, evaluate, experiments, forward, plots
 from .models import MODELS
 
 FORECAST_REPORT_HORIZONS = (182, 365, 730, 1095, 1460)
@@ -36,6 +38,8 @@ def main() -> None:
     commands.add_parser("forecast", help="forecast from the latest price")
     ab = commands.add_parser("ab", help="run A/B experiments on development data")
     ab.add_argument("names", nargs="*", help="experiments to run (default: all)")
+    commands.add_parser("snapshot", help="record forecasts for the forward test")
+    commands.add_parser("forward", help="score recorded forecasts")
     args = parser.parse_args()
     if args.command == "ab" and (unknown := set(args.names) - set(experiments.EXPERIMENTS)):
         parser.error(f"unknown experiments: {', '.join(sorted(unknown))}")
@@ -50,6 +54,10 @@ def main() -> None:
         run_forecast(models, args.output)
     elif args.command == "ab":
         run_ab(args.names or list(experiments.EXPERIMENTS), args.output)
+    elif args.command == "snapshot":
+        run_snapshot()
+    elif args.command == "forward":
+        run_forward(args.output)
 
 
 def run_backtest(models, output: Path, holdout: bool) -> None:
@@ -112,6 +120,46 @@ def run_ab(names: list[str], output: Path) -> None:
     (output / "ab" / "verdicts.txt").write_text(table + "\n")
 
 
+def run_snapshot() -> None:
+    prices = data.load_prices()
+    path = forward.snapshot(prices)
+    recorded = [(m, f) for m, f in forward.load_snapshots() if f.origin == prices.index[-1]]
+    medians = pd.DataFrame(
+        {m: [plots.price_formatter(p) for p in np.exp(f.quantile(0.5))] for m, f in recorded},
+        index=[evaluate.horizon_label(h) for h in evaluate.HORIZONS],
+    ).T
+    print(
+        f"recorded {len(recorded)} models' forecasts from {prices.index[-1]:%Y-%m-%d} in {path}\n"
+    )
+    print("medians:\n" + medians.to_string())
+
+
+def run_forward(output: Path) -> None:
+    prices = data.load_prices()
+    scores = forward.score_snapshots(prices)
+    due = forward.next_due(prices)
+    if scores.empty:
+        when = f"; the first is due {due:%Y-%m-%d}" if due is not None else ""
+        print(f"no recorded forecast has reached its target date yet{when}")
+        return
+    out = output / "forward"
+    out.mkdir(parents=True, exist_ok=True)
+    summary = evaluate.summarize(scores, step_days=forward.SNAPSHOT_STEP_DAYS)
+    report = (
+        f"forward test: {scores.origin.nunique()} snapshots from {scores.origin.min():%Y-%m-%d}, "
+        f"outcomes through {prices.index[-1]:%Y-%m-%d}\n\n" + evaluate.format_summary(summary)
+    )
+    if due is not None:
+        report += f"\n\nnext outcome due {due:%Y-%m-%d}"
+    print(report)
+    (out / "report.txt").write_text(report + "\n")
+    scores.to_csv(out / "scores.csv", index=False)
+    summary.to_csv(out / "summary.csv", index=False)
+    plots.skill_chart(summary, out / "skill.png", "forward test: skill by horizon")
+    plots.calibration_chart(summary, out / "calibration.png", "forward test: interval coverage")
+    print(f"\nwrote {out}/")
+
+
 def run_forecast(models, output: Path) -> None:
     prices = data.load_prices()
     horizons = np.arange(1, max(FORECAST_REPORT_HORIZONS) + 1)