gribsons/quant-tradingbot
0
1"""2api_server.py — FastAPI backend for QuantBot.3 4Exposes the backtesting engine over HTTP so the Next.js frontend5(or any HTTP client) can run backtests without touching Python directly.6 7Deploy to Railway:8 Procfile: web: uvicorn api_server:app --host 0.0.0.0 --port $PORT9"""10 11from __future__ import annotations12 13import logging14import math15import sys16import os17 18import numpy as np19import pandas as pd20from fastapi import FastAPI, HTTPException21from fastapi.middleware.cors import CORSMiddleware22from pydantic import BaseModel, Field23 24# Local imports (this file lives at the quant_tradingbot root)25from config import (26 AnalyticsConfig,27 BrokerConfig,28 EnsembleConfig,29 MACrossoverConfig,30 MomentumConfig,31 QuantBotConfig,32 RiskConfig,33 RSIReversionConfig,34)35from data.fetcher import DataFetcher36from data.processor import DataProcessor37from backtest.engine import BacktestEngine38from risk.manager import RiskManager39from strategies import STRATEGY_REGISTRY, get_strategy40 41logging.basicConfig(level=logging.INFO)42logger = logging.getLogger(__name__)43 44app = FastAPI(title="QuantBot API", version="1.0.0")45 46# ---------------------------------------------------------------------------47# CORS — allow the Vercel frontend (and localhost dev)48# ---------------------------------------------------------------------------49 50ALLOWED_ORIGINS = os.environ.get(51 "ALLOWED_ORIGINS",52 "http://localhost:3000,http://localhost:3001",53).split(",")54 55app.add_middleware(56 CORSMiddleware,57 allow_origins=ALLOWED_ORIGINS,58 allow_origin_regex=r"https://.*\.vercel\.app",59 allow_credentials=True,60 allow_methods=["*"],61 allow_headers=["*"],62)63 64 65# ---------------------------------------------------------------------------66# Request / Response models67# ---------------------------------------------------------------------------68 69class BacktestRequest(BaseModel):70 symbol: str = "SPY"71 start_date: str = "2020-01-01"72 end_date: str = "2024-12-31"73 interval: str = "1d"74 strategy: str = "ensemble"75 # Broker76 initial_cash: float = Field(100_000.0, ge=1_000)77 commission_bps: float = Field(10.0, ge=0, le=100)78 slippage_bps: float = Field(5.0, ge=0, le=100)79 # Risk80 max_position_pct: float = Field(0.20, ge=0.01, le=1.0)81 max_drawdown_pct: float = Field(0.15, ge=0.01, le=1.0)82 max_open_positions: int = Field(5, ge=1, le=20)83 position_sizing: str = "fixed_pct"84 # Ensemble-specific85 adx_trend: float = 25.086 adx_range: float = 20.087 vote_threshold: float = 0.588 min_agreement: int = 289 atr_stop: float = 2.090 atr_target: float = 4.091 92 93class EquityPoint(BaseModel):94 date: str95 equity: float96 97 98class TradeRecord(BaseModel):99 entry_date: str = ""100 exit_date: str = ""101 symbol: str = ""102 direction: str = ""103 entry_price: float = 0.0104 exit_price: float = 0.0105 shares: float = 0.0106 pnl: float = 0.0107 pnl_pct: float = 0.0108 commission: float = 0.0109 slippage: float = 0.0110 exit_reason: str = ""111 112 113class BacktestResponse(BaseModel):114 strategy_name: str115 symbol: str116 metrics: dict117 equity_curve: list[EquityPoint]118 trade_log: list[dict]119 120 121# ---------------------------------------------------------------------------122# Helpers123# ---------------------------------------------------------------------------124 125def _safe(v):126 """Make a value JSON-safe (handles inf / nan)."""127 if isinstance(v, float):128 if math.isnan(v) or math.isinf(v):129 return None130 return v131 132 133def _df_to_equity_points(ec: pd.DataFrame) -> list[dict]:134 """Convert equity curve DataFrame to list of {date, equity} dicts."""135 if ec.empty:136 return []137 result = []138 for idx, row in ec.iterrows():139 result.append({140 "date": str(idx.date()) if hasattr(idx, "date") else str(idx),141 "equity": round(float(row["equity"]), 2),142 })143 return result144 145 146def _df_to_trade_records(tl: pd.DataFrame) -> list[dict]:147 """Convert trade log DataFrame to list of dicts, JSON-safe."""148 if tl.empty:149 return []150 records = []151 for _, row in tl.iterrows():152 rec = {}153 for col in tl.columns:154 val = row[col]155 if isinstance(val, (pd.Timestamp,)):156 rec[col] = str(val.date())157 elif isinstance(val, float) and (math.isnan(val) or math.isinf(val)):158 rec[col] = None159 elif isinstance(val, (np.integer,)):160 rec[col] = int(val)161 elif isinstance(val, (np.floating,)):162 rec[col] = round(float(val), 6)163 else:164 rec[col] = val165 records.append(rec)166 return records167 168 169# ---------------------------------------------------------------------------170# Routes171# ---------------------------------------------------------------------------172 173@app.get("/")174def root():175 return {"name": "QuantBot API", "version": "1.0.0", "docs": "/docs", "health": "/health"}176 177 178@app.get("/health")179def health():180 return {"status": "ok", "strategies": list(STRATEGY_REGISTRY.keys())}181 182 183@app.get("/strategies")184def list_strategies():185 return {"strategies": list(STRATEGY_REGISTRY.keys())}186 187 188@app.post("/run", response_model=BacktestResponse)189def run_backtest(req: BacktestRequest):190 logger.info("Backtest request: %s %s %s→%s", req.strategy, req.symbol, req.start_date, req.end_date)191 192 # Build config193 cfg = QuantBotConfig(194 broker=BrokerConfig(195 commission_pct=req.commission_bps / 10_000.0,196 slippage_pct=req.slippage_bps / 10_000.0,197 initial_cash=req.initial_cash,198 ),199 risk=RiskConfig(200 max_position_pct=req.max_position_pct,201 max_drawdown_pct=req.max_drawdown_pct,202 max_open_positions=req.max_open_positions,203 position_sizing=req.position_sizing,204 ),205 ma_crossover=MACrossoverConfig(),206 rsi_reversion=RSIReversionConfig(),207 momentum=MomentumConfig(),208 ensemble=EnsembleConfig(209 adx_trend_threshold=req.adx_trend,210 adx_range_threshold=req.adx_range,211 vote_threshold=req.vote_threshold,212 min_agreement=int(req.min_agreement),213 atr_stop_multiplier=req.atr_stop,214 atr_target_multiplier=req.atr_target,215 ),216 analytics=AnalyticsConfig(),217 start_date=req.start_date,218 end_date=req.end_date,219 interval=req.interval,220 )221 222 try:223 # Fetch data224 raw = DataFetcher().fetch(req.symbol, req.start_date, req.end_date, req.interval)225 except Exception as exc:226 raise HTTPException(status_code=422, detail=f"Data fetch failed: {exc}")227 228 try:229 proc = DataProcessor().process(raw)230 except Exception as exc:231 raise HTTPException(status_code=422, detail=f"Data processing failed: {exc}")232 233 # Build strategy kwargs234 kwargs: dict = {}235 if req.strategy == "ensemble":236 kwargs = {"config": cfg.ensemble}237 elif req.strategy == "ma_crossover":238 kwargs = {"config": cfg.ma_crossover}239 elif req.strategy == "rsi_reversion":240 kwargs = {"config": cfg.rsi_reversion}241 elif req.strategy == "momentum":242 kwargs = {"config": cfg.momentum}243 244 try:245 strat = get_strategy(req.strategy, **kwargs)246 except Exception as exc:247 raise HTTPException(status_code=422, detail=f"Strategy error: {exc}")248 249 try:250 rm = RiskManager(cfg.risk, cfg.broker.initial_cash)251 result = BacktestEngine(cfg).run(strat, proc, req.symbol, rm)252 except Exception as exc:253 logger.exception("Backtest failed")254 raise HTTPException(status_code=500, detail=f"Backtest failed: {exc}")255 256 # Sanitise metrics257 safe_metrics = {k: _safe(v) for k, v in result.metrics.items()}258 259 return BacktestResponse(260 strategy_name=result.strategy_name,261 symbol=result.symbol,262 metrics=safe_metrics,263 equity_curve=_df_to_equity_points(result.equity_curve),264 trade_log=_df_to_trade_records(result.trade_log),265 )266