btengine

An event-driven, multi-instrument backtesting engine for FX/CFD strategies, using Oanda v20 candlestick data.

• Event-driven loop — bar-by-bar, no look-ahead bias.
• Multi-instrument portfolio with shared account equity.
• Class-based strategies against a small built-in indicator library.
• Backtest only — Oanda is used purely as a historical data source.

Setup

python -m venv .venv && source .venv/bin/activate
pip install -e ".[dev]"
cp .env.example .env   # then add your Oanda practice token

Get a free practice-account token at https://www.oanda.com/demo-account/tpa/personal_token.

Usage

Download and cache candles:

bt-download --instruments EUR_USD,GBP_USD --granularity H1 --from 2023-01-01 --to 2024-01-01

Run a backtest on cached data:

bt-run --strategy sma_crossover --instruments EUR_USD,GBP_USD --granularity H1

Visualizing results

Add --plot for a static equity/drawdown PNG, or --dashboard for a self-contained interactive HTML report (candlesticks + buy/sell markers + indicator overlays + equity, drawdown, and trade-analytics histograms):

bt-run --strategy sma_crossover --instruments EUR_USD --granularity H1 \
       --plot equity.png --dashboard dashboard.html

Open dashboard.html in any browser — Plotly.js is embedded, so it works offline. Strategies surface indicators on the price chart by calling ctx.record(instrument, name=value) (see sma_crossover).

Built-in strategies

--strategy	Description
sma_crossover	Fast/slow SMA crossover (records the SMAs for the chart).
opening_range	Opening-range breakout via the SessionStrategy base.
session_breakout	Breakout of the 09:30–10:00 opening range; overlays OR + Asia levels.
asia_reversion	Fades the London break of the Asia range back to the 50% level, with a 1%-risk stop/target bracket; flat by 16:00 ET.

Bracket orders & risk sizing

Strategies can place resting stop-loss / take-profit brackets (GTC, OCO, reduce-only) and size by risk:

units = ctx.units_for_risk(inst, stop_price=stop, risk_pct=0.01)  # risk 1% of equity
ctx.enter(inst, Side.SELL, units, stop_loss=stop, take_profit=target)  # OCO bracket

The stop and target rest across bars until one fills (cancelling the other), and only ever reduce the position. See asia_reversion for a full example.

Writing a strategy

Subclass Strategy and implement on_bar:

from btengine.strategy.base import Strategy
from btengine.indicators import sma

class SmaCrossover(Strategy):
    def on_bar(self, ctx):
        for inst in ctx.instruments:
            close = ctx.history(inst)["close"]
            if len(close) < 50:
                continue
            fast, slow = sma(close, 20), sma(close, 50)
            if fast.iloc[-1] > slow.iloc[-1] and not ctx.position(inst):
                ctx.buy(inst)
            elif fast.iloc[-1] < slow.iloc[-1] and ctx.position(inst):
                ctx.close(inst)

Testing

pytest

The engine test suite runs fully offline against synthetic data.

Architecture

MarketEvent -> Strategy -> SignalEvent -> Portfolio (sizing) -> OrderEvent
            -> SimulatedBroker (spread/slippage/commission) -> FillEvent -> Portfolio

See PLAN.md for the full design.