- Joined
- 19 August 2007
- Posts
- 24
- Reactions
- 0
Backtesting option strategies is extremely difficult, but the rewards justify the effort (excuse my enthusiasm, but you may have noticed that this a pet topic of mine based on past posts ).
I agree with Howards view that your can't use OHLC option price history for backtesting, however you can get closing bid/ask data for all option chains in the US from www.historicaloptiondata.com. I maintain a database for same in Australia.
Here are three methods for backtesting option strategies in order of representativeness (and difficulty):
1. Use HV as a proxy for IV as Howard mentioned, then use a pricing model such to estimate option prices based on the underlying share price, div, ex div date, strike, date, expiry date, interest rate. This doesn't account for periods where HV and IV diverge such as around earnings time, after a one off HV spike, post div, or for IV skews.
2. Use historical IV and a pricing model as above. This is much better but it still doesnt allow for IV skews.
3. Use actual historical option bid/ask data. Despite the occasional problem with dodgy eod mkt-maker quotes, this is as about close as you can get to reliable backtesting.
You have to be very specific in how you setup the backtest (i.e. what option strategy, which stocks, strikes(in-at or out of the money-and how far), expiry date range).Other things to consider are how you handle bid/ask spreads, liquidity, brokerage, early exercise and capital restructurings. The results can be very interesting, often going against conventional thought. The most important thing is you can reliably estimate a systems expectancy, win%, win/loss and base your position size upon this.
Don't give up on the idea, it is worth the effort. Happy to help with info where I can.
VT
I agree with Howards view that your can't use OHLC option price history for backtesting, however you can get closing bid/ask data for all option chains in the US from www.historicaloptiondata.com. I maintain a database for same in Australia.
Here are three methods for backtesting option strategies in order of representativeness (and difficulty):
1. Use HV as a proxy for IV as Howard mentioned, then use a pricing model such to estimate option prices based on the underlying share price, div, ex div date, strike, date, expiry date, interest rate. This doesn't account for periods where HV and IV diverge such as around earnings time, after a one off HV spike, post div, or for IV skews.
2. Use historical IV and a pricing model as above. This is much better but it still doesnt allow for IV skews.
3. Use actual historical option bid/ask data. Despite the occasional problem with dodgy eod mkt-maker quotes, this is as about close as you can get to reliable backtesting.
You have to be very specific in how you setup the backtest (i.e. what option strategy, which stocks, strikes(in-at or out of the money-and how far), expiry date range).Other things to consider are how you handle bid/ask spreads, liquidity, brokerage, early exercise and capital restructurings. The results can be very interesting, often going against conventional thought. The most important thing is you can reliably estimate a systems expectancy, win%, win/loss and base your position size upon this.
Don't give up on the idea, it is worth the effort. Happy to help with info where I can.
VT
