feat: add strategy and backtest services

- strategy_service.py combines opportunity + portfolio signals into
  unified buy/sell/hold recommendations
- backtest_service.py runs walk-forward backtests on historical datasets
  with virtual cash and positions
- CLI adds `strategy` and `backtest` commands with `--decision-interval`
  and other tuning parameters
- Add tests for both services and CLI dispatch
- Update CLAUDE.md with new architecture docs
- Optimize model weights via opportunity optimizer

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

tests/test_backtest_service.py

@@ -0,0 +1,129 @@
+"""Tests for backtest_service."""
+
+from __future__ import annotations
+
+import json
+import tempfile
+import unittest
+from pathlib import Path
+from typing import Any
+
+from coinhunter.services import backtest_service
+
+
+class BacktestServiceTestCase(unittest.TestCase):
+    def _klines(self, closes: list[float], start_ms: int = 0, volumes: list[float] | None = None) -> list[list[float]]:
+        volumes = volumes or [1.0] * len(closes)
+        return [
+            [start_ms + i * 3600000, c * 0.98, c * 1.02, c * 0.97, c, v, 0.0, c * v, 100, 0.0, 0.0, 0.0]
+            for i, (c, v) in enumerate(zip(closes, volumes))
+        ]
+
+    def _config(self) -> dict[str, Any]:
+        return {
+            "opportunity": {
+                "entry_threshold": 1.5,
+                "watch_threshold": 0.6,
+                "min_trigger_score": 0.45,
+                "min_setup_score": 0.35,
+                "overlap_penalty": 0.6,
+                "top_n": 10,
+                "scan_limit": 50,
+                "kline_limit": 48,
+                "weights": {},
+                "model_weights": {},
+            },
+            "portfolio": {
+                "add_threshold": 1.5,
+                "hold_threshold": 0.6,
+                "trim_threshold": 0.2,
+                "exit_threshold": -0.2,
+                "max_position_weight": 0.6,
+                "max_positions": 5,
+            },
+            "signal": {
+                "lookback_interval": "1h",
+            },
+            "market": {
+                "default_quote": "USDT",
+            },
+            "trading": {
+                "commission_pct": 0.001,
+            },
+        }
+
+    def _make_dataset(self, closes_by_symbol: dict[str, list[float]], start_iso: str = "2025-12-28T00:00:00Z", sim_start_iso: str = "2025-12-30T00:00:00Z", sim_end_iso: str = "2026-01-01T00:00:00Z") -> Path:
+        from datetime import datetime, timezone
+        start_ms = int(datetime.fromisoformat(start_iso.replace("Z", "+00:00")).timestamp() * 1000)
+        klines: dict[str, dict[str, list[list[float]]]] = {}
+        for symbol, closes in closes_by_symbol.items():
+            klines[symbol] = {"1h": self._klines(closes, start_ms=start_ms)}
+        dataset = {
+            "metadata": {
+                "created_at": "2026-01-01T00:00:00Z",
+                "quote": "USDT",
+                "symbols": list(closes_by_symbol.keys()),
+                "plan": {
+                    "intervals": ["1h"],
+                    "kline_limit": 48,
+                    "reference_days": 2.0,
+                    "simulate_days": 1.0,
+                    "run_days": 1.0,
+                    "total_days": 4.0,
+                    "start": start_iso,
+                    "simulation_start": sim_start_iso,
+                    "simulation_end": sim_end_iso,
+                    "end": sim_end_iso,
+                },
+                "external_history": {"provider": "disabled", "status": "disabled"},
+            },
+            "klines": klines,
+        }
+        fp = tempfile.NamedTemporaryFile(mode="w", suffix=".json", delete=False)
+        json.dump(dataset, fp)
+        fp.close()
+        return Path(fp.name)
+
+    def test_run_backtest_produces_summary(self) -> None:
+        config = self._config()
+        closes = list(range(20, 92))
+        path = self._make_dataset({"BTCUSDT": closes})
+        try:
+            result = backtest_service.run_backtest(config, dataset_path=str(path), initial_cash=10000.0)
+            self.assertIn("summary", result)
+            self.assertIn("trades", result)
+            self.assertIn("equity_curve", result)
+            self.assertIn("parameters", result)
+            summary = result["summary"]
+            self.assertIn("initial_cash", summary)
+            self.assertIn("final_equity", summary)
+            self.assertIn("total_return_pct", summary)
+            self.assertIn("max_drawdown_pct", summary)
+            self.assertIn("win_rate", summary)
+        finally:
+            path.unlink()
+
+    def test_run_backtest_missing_simulation_dates_raises(self) -> None:
+        config = self._config()
+        path = self._make_dataset({"BTCUSDT": list(range(20, 92))}, sim_start_iso="", sim_end_iso="")
+        try:
+            with self.assertRaises(ValueError):
+                backtest_service.run_backtest(config, dataset_path=str(path))
+        finally:
+            path.unlink()
+
+    def test_run_backtest_tracks_equity_curve(self) -> None:
+        config = self._config()
+        # Need ~72 candles to cover 2025-12-28 through 2026-01-01 (warmup + simulation)
+        closes = list(range(20, 92))
+        path = self._make_dataset({"BTCUSDT": closes})
+        try:
+            result = backtest_service.run_backtest(config, dataset_path=str(path), initial_cash=10000.0)
+            self.assertTrue(len(result["equity_curve"]) > 0)
+            first = result["equity_curve"][0]
+            self.assertIn("time", first)
+            self.assertIn("equity", first)
+            self.assertIn("cash", first)
+            self.assertIn("positions_count", first)
+        finally:
+            path.unlink()