Backtesting Issues in ONE (Part 2)

Last time, I presented some screenshots detailing quirky things I have been seeing in OptionNet Explorer (ONE) pertaining to data. Today I will continue the discussion.

My initial thought was that I should not be seeing zero extrinsic value across a range of strikes with significant time to expiration. If the benefits of early put exercise (i.e. earning risk-free rate on the proceeds) outweighs the extrinsic value, then the put should be exercised early. Since extrinsic value is surrendered upon exercise, put buyers can exercise ITM options without penalty once extrinsic value is gone. They can then implement the proceeds elsewhere.

If what I am seeing in ONE were accurate, then early exercise should be common. Based on my experience and what I hear from others, it is not.

Wait! I now realize I made a major oversight here. Looking back at the screenshots, do you see it?*

The more general case of seeing constant (“sticky”) nonzero extrinsic value across a wide range of strikes also seems wrong.

I e-mailed ONE Support on this issue and got some meaningful information. I strongly praise DJ who has been very helpful and usually prompt with his response time. He writes:

     > The data isn’t corrupt but you will notice wider spreads on some
     > instruments, particularly SPX when using EOD quotes. These were
     > the exact quotes provided by CBOE so they are accurate and exactly
     > what you would have seen in the market at that point in time when
     > viewed at the EOD…
     >
     > I note what you refer to as the “sticky” nature of some of the
     > quotes but again, they are correctly derived from the market data.
     > Our beta version shows both bid and ask prices (as well as mid
     > price), and you will notice up to $3 spreads on some strikes at
     > the EOD. This can inevitably impact the calculation of both IV and
     > extrinsic value, depending on where the mid price resides and
     > how the spread widens.

Provided we believe DJ, which I do, the data seen in ONE are accurate.

I remain troubled because these inconsistencies seem to violate option arbitrage. As an example with the underlying at 100, suppose the 105 and 110 puts are priced at $5 and $10, respectively: both having zero extrinsic value. I could then sell the credit spread for $5 and have a position that would at worst lose nothing (options expire ITM) and at best profit $5 (options expire worthless). This edge would be exploited repeatedly until it went away (selling causes extrinsic value to decrease while buying causes extrinsic value to increase).

A range of options having constant nonzero extrinsic value should similarly violate option arbitrage. With the underlying at 100, suppose the 105 and 110 puts are priced at $8 and $13, respectively: both having $3 extrinsic value. I could then sell the spread and have a max potential profit of $5 (options expire OTM) with a max potential loss of breakeven (options expire ITM). Again, this edge would be exploited repeatedly until it disappeared.

I question whether flawed numbers like this can be used to backtest, and I remain puzzled as to why I am seeing it.

I will continue next time.

* — SPX has cash-settled, European options that are not subject to early exercise.