Backtesting Option Strategies by John A Sarkett
Looking for a quick way to backtest option strategies? You can use Excel to count occurrences of historical prices outside a price band you set to develop an expectation for future price movement. Here’s how.
SO much of the option trader’s life has been enhanced over just the past 10 or so years. We have radically lower commissions — $0.15 to $1.50 per contract, plus various ticket charge plans; we have tighter markets, with high-volume contracts just one penny apart; we have better software with risk curves, what-if scenarios, and charting, with brokers and vendors continuing to enhance their offerings; and finally, we have easier access to futures and world markets for intermarket hedging.
Only one aspect of the option trading experience remains, more or less, as it was years ago — option strategy backtesting. If I put on this Spx or Spy or Rut or Etf or Xyz butterfly, condor, calendar, or double diagonal, what are my chances of success based on the recent past? That is the biggest question for every option trader.
For a back-of-the-envelope type calculation, some use deltas as shorthand for the probability of going in-the-money. For example, an option with a 10 delta will have roughly a 10% chance of doing so. Actual probability math is much more complex (see my December 1997 S&C article), but the answers usually come out close enough. Some software calculates “probability of expiration” or “probability of touching” (strike) for you. Thinkorswim does this, for example.
Other traders take on a much more time-consuming task of establishing the trade in their option software, then clicking through, day by day, to see how the trade progresses. This allows you to insert “adjustments” and see how these fare. Many option traders have invested many weekend hours in this kind of research.
Industrial strength option packages provide this opportunity; OptionVue is still the gold standard, although there are new offerings, too. Typical time involved for just one ticker: one year’s worth of data can be analyzed in 30 to 60 minutes. The data accessed is typically real, historical option data. (In the case of missing data, some software programs will insert theoretical prices.)