Overnight options backtest (close to next open)

This folder contains a backtest that compares these overnight positions:

• Straddle: call + put at the same (ATM) strike
• Strangle: OTM call + OTM put at different strikes
• Equity baseline: long shares from close to next open

For each trading day, the backtest enters at the close and exits at the next open.

What gets plotted and written

Outputs go to outputs/:

• pnl.png: three panels
  • Cumulative P&L in dollars
  • Free cash after entry (account equity minus capital required)
  • Trade-by-trade P&L in dollars
• trades.csv: one row per trade per strategy
• performance.csv: trailing percent returns (1d, 1m, 1y, etc.) per strategy line

Pricing model (options)

Historical option prices are not pulled from an options data feed. Instead, the backtest prices the call and put using Black–Scholes (European, no dividends) with volatility coming from either:

• Rolling realized volatility from the underlying’s daily closes, or
• A fixed implied volatility from the config

This means the option P&L is model-based and will differ from live fills and real option markets.

Requirements

• Python 3.7 or higher
• Dependencies: yfinance, pandas, numpy, matplotlib, PyYAML
• For Python 3.7: typing_extensions and pytz are automatically installed if using pip
• Install all dependencies: pip3 install -e . or pip3 install yfinance pandas numpy matplotlib PyYAML typing_extensions pytz

Running

From OvernightStraddleBacktest/:

Show the plot window:

PYTHONPATH=src python3 -m overnight_straddle.main --config config.yaml

Run without opening a plot window (still writes PNG and CSVs):

MPLBACKEND=Agg PYTHONPATH=src python3 -m overnight_straddle.main --config config.yaml --no-show

Realtime (paper) runner

src/overnight_straddle/realtime_short_straddle.py is a long-running script that:

• Starts with a cash balance of 50000 (configurable)
• Near the close, records a paper fill for a short straddle
• At the next open, records a paper close
• After closing, posts the last trade P&L and current equity to a Discord webhook

It uses yfinance option chain quotes for paper fills (bid/ask/last). It does not place live orders.

Run it:

export DISCORD_WEBHOOK_URL="(your webhook url)"
PYTHONPATH=src python3 -m overnight_straddle.realtime_short_straddle --config realtime_config.yaml

config.yaml reference

tickers

• List of tickers to backtest, e.g. ["SPY"]. Each ticker is run independently.

data

• source: yfinance or csv
• interval: only 1d is expected for this strategy
• csv_folder: used only if source=csv. Expects data/<TICKER>.csv with columns Date,Open,High,Low,Close.

backtest

• lookback_trading_days: number of most recent trading days to include
• end_date: optional end date (YYYY-MM-DD). null means use the most recent data.

run

• strategies: any of straddle, strangle (the equity baseline is controlled separately)
• position:
  • long: buy at close, sell at next open
  • short: sell at close, buy back at next open
• include_equity_baseline: if true, also runs long equity (close → next open)

calendar

• allow_weekend_holds:
  • true: includes Friday close → Monday open
  • false: skips trades where the next trading day is more than 1 calendar day away

strategy

• dte_days: assumed calendar days to expiry for the option used at entry
• include_weekends_in_time_decay:
  • true: Friday close → Monday open uses the larger calendar time gap for theta
  • false: treats any close → next open as a fixed overnight duration for theta
• strike_rounding: rounds strikes to the nearest increment (e.g. 1.0 means $1 strikes)
• strangle.call_otm_pct: call strike set to (S \cdot (1 + call_otm_pct))
• strangle.put_otm_pct: put strike set to (S \cdot (1 - put_otm_pct))
• contract_multiplier: typically 100 for US equity options
• contracts_per_trade: number of straddles/strangles per trade per day

strategy.vol

• method: rolling_realized or fixed
• lookback_days: lookback for realized volatility (daily closes)
• fixed_iv: used if method=fixed (and as a fallback)

strategy.rates

• risk_free_rate: annualized rate used by Black–Scholes

strategy.costs

• slippage_bps: bps applied on entry and exit notional (options and equity)

equity

• shares_per_trade: used by the equity baseline

margin

• short_options_pct_underlying: used only for plotting “capital required” on short option trades. It is a proxy for how much of the account is tied up when holding a short straddle/strangle.

performance

• initial_balance: starting account balance for percent return calculations (default 100000)
• horizons_trading_days: trailing return windows in trading days (e.g. 1m=21, 1y=252)

plot

• output_dir: output directory (relative to config file)
• save_png: whether to write pnl.png
• show: whether to display the matplotlib window (can be overridden by --no-show)

stress_test

This is a sensitivity analysis for the option strategies (straddle/strangle). It does not use real option data. It re-prices each trade using Black–Scholes under a grid of assumptions, then writes additional CSVs and a chart.

• enabled: runs the stress test if true
• iv_shift_entry_points: list of absolute volatility shifts applied at entry (e.g. 0.05 means +5 vol points)
• iv_shift_exit_points: list of absolute volatility shifts applied at exit
• option_half_spread_pct: half-spread applied to option premiums:
  • buys use mid * (1 + half_spread)
  • sells use mid * (1 - half_spread)
• extra_slippage_bps: extra bps slippage applied on top of strategy.costs.slippage_bps
• plot: if true, writes outputs/stress.png showing total return vs exit IV shift

Disclaimer: This project was created with the assistance of Cursor AI