"""Synthetic library replay only; never canonical VR research or actual trading.""" import argparse import contextlib import hashlib import io import json import sys from pathlib import Path HERE=Path(__file__).resolve().parent LOCAL_ENGINE=HERE/'_bats_private/engine' if LOCAL_ENGINE.exists():sys.path.insert(0,str(LOCAL_ENGINE)) import backtesting from backtesting import Backtest, Strategy from backtesting.lib import crossover import numpy as np import pandas as pd SETTINGS=dict(schema_version=1,classification='synthetic-library-fixture',fast=3,slow=5,completed_bars_before_exit_request=3,eligible_entry_first_bar=5,eligible_entry_last_bar=49,quantity=1,initial_cash='10000',commission_fixed_per_side='0.10',commission_relative='0',spread='0',margin='1',trade_on_close=False,exclusive_orders=True,finalize_trades=False,selection='one frozen rule; no optimiser or parameter ranking',economic_validation=False,production_authority=False) def digest(data):return hashlib.sha256(data).hexdigest() def canonical(value):return (json.dumps(value,sort_keys=True,separators=(',',':'),ensure_ascii=True)+'\n').encode('utf-8') def sma(values,window):return pd.Series(values).rolling(window,min_periods=window).mean().to_numpy() def load_data(data): frame=pd.read_csv(io.BytesIO(data)) assert list(frame)==['bar_id','bar_open_utc','Open','High','Low','Close','Volume'] assert len(frame)==64 and frame.bar_id.tolist()==list(range(64)) index=pd.to_datetime(frame.pop('bar_open_utc'),utc=True,errors='raise') assert index.is_monotonic_increasing and index.is_unique assert all((index.diff().dropna()==pd.Timedelta(minutes=1))) assert np.isfinite(frame[['Open','High','Low','Close','Volume']].to_numpy()).all() assert (frame[['Open','High','Low','Close']]>0).all().all() assert ((frame.High>=frame[['Open','Close']].max(axis=1))&(frame.Low<=frame[['Open','Close']].min(axis=1))&(frame.Volume>=0)).all() frame.index=pd.DatetimeIndex(index);return frame def replay(data,settings=SETTINGS): assert settings==SETTINGS,'Only the frozen teaching configuration is supported' assert backtesting.__version__=='0.6.6' frame=load_data(data) class TeachingRule(Strategy): def init(self): self.fast=self.I(sma,self.data.Close,settings['fast'],plot=False) self.slow=self.I(sma,self.data.Close,settings['slow'],plot=False) self.audit=[] def next(self): i=len(self.data)-1;action='none' if self.position: if i-self.trades[0].entry_bar+1>=settings['completed_bars_before_exit_request']: self.position.close();action='request_whole_exit' elif not self.orders and settings['eligible_entry_first_bar']<=i<=settings['eligible_entry_last_bar'] and crossover(self.fast,self.slow): self.buy(size=settings['quantity'],tag='fixture-entry-'+str(i));action='request_long_entry' self.audit.append(dict(bar_id=i,available_close=float(self.data.Close[-1]),fast=float(self.fast[-1]),slow=float(self.slow[-1]),action=action)) engine=Backtest(frame,TeachingRule,cash=float(settings['initial_cash']),commission=(float(settings['commission_fixed_per_side']),float(settings['commission_relative'])),spread=float(settings['spread']),margin=float(settings['margin']),trade_on_close=settings['trade_on_close'],exclusive_orders=settings['exclusive_orders'],finalize_trades=settings['finalize_trades']) with contextlib.redirect_stderr(io.StringIO()):stats=engine.run() state=stats['_strategy'];assert not state.trades and not state.orders,'Fixture must finish flat without forced liquidation' trades=[] for _,r in stats['_trades'].iterrows(): entry=int(r.EntryBar);exit=int(r.ExitBar) trades.append(dict(intent_bar=entry-1,entry_fill_bar=entry,exit_request_bar=exit-1,exit_fill_bar=exit,whole_quantity=int(r.Size),entry_price=float(r.EntryPrice),exit_price=float(r.ExitPrice),net_cash=float(r.PnL),commission_cash=float(r.Commission),tag=str(r.Tag))) curve=[dict(bar_id=i,equity_cash=float(r.Equity)) for i,(_,r) in enumerate(stats['_equity_curve'].iterrows())] return dict(schema_version=1,classification='synthetic-library-fixture',engine=dict(name='backtesting.py',version=backtesting.__version__,pandas=pd.__version__,numpy=np.__version__,python=sys.version.split()[0]),source_code_sha256=digest(Path(__file__).read_bytes()),data_sha256=digest(data),settings_sha256=digest(canonical(settings)),settings=settings,bar_count=len(frame),first_decision_bar=state.audit[0]['bar_id'],decisions=state.audit,trades=trades,equity_curve=curve,final_cash=float(stats['Equity Final [$]']),net_cash=float(stats['Equity Final [$]'])-float(settings['initial_cash']),economic_validation=False,broker_fills=None,production_promotion=False,limitations='Invented bars and times; next-bar-open library fills and fixed fees; no queue, slippage, financing, shorts, tax, real market, untouched economic holdout or strategy certification. Tutorial fill policy is not the canonical VR observation-window convention.') def main(): parser=argparse.ArgumentParser(description=__doc__);parser.add_argument('csv',type=Path);parser.add_argument('--output',type=Path);args=parser.parse_args() value=replay(args.csv.read_bytes());payload=json.dumps(value,indent=2,ensure_ascii=True)+'\n' if args.output:args.output.write_text(payload,encoding='utf-8',newline='\n') else:print(payload,end='') if __name__=='__main__':main()