Musings on Put Credit Spreads (Part 8)

Today I will continue to ramble on about some possibilities of backtesting the put credit spread.

I could stick with the naked put and buy a unit or two for a fixed percentage of the naked put premium. After backtesting a put spread I was thinking about including insurance like this as a ratio backspread. Maybe what I am really talking about here are two distinct backtests—naked puts and long options—that I overlay and sum to generate an equity curve. If this is valid then I would be killing two birds with one stone because the put spreads would not have to be backtested.

Alternatively, I could opt to buy the same number of longs as I short in the back month. The extra time could synthetically substitute for the additional contract especially because for any given series of shorts, two longs would be in place. I might not include this additional bonus for backtesting purposes since each trade would consist of the naked put plus a farther out-of-the-money option in the following month. Nevertheless, I feel a sense of comfort anytime I have reason to believe actual results would be better than those backtested.

Practically speaking, a problem here is the availability of deep-out-of-the-money strikes in the database. This is especially the case in the early years (2001-4). I should call OptionVue and ask why the strikes displayed are limited and what it would take to be able to see 5-10 more strikes automatically generated by the software.

As it is now, I need to add those strikes manually and then set the prices to fit some logical vertical skew. The skew I employ is 1.0%-1.2% IV per 10 underlying points or, in subjectively perceived crash conditions, 1.5%-2.0%. Of course, too, 10 points is much larger percentage of the underlying early in the database compared to later.

After the last eight posts, I’m not sure I have solved my problem with backtesting put credit spreads. I do think everything is on the table in front of me to mull over and consider, though. I will go forth in my writing from here.

Musings on Put Credit Spreads (Part 7)

That last sentence strikes me as quite profound.

For the sake of backtesting, given these two choices I should go with the one offering a decent sample size. Discretionary traders make a [temporary?] living trading based on small sample size. Trading system salespeople use small-sample-size examples as presentation material (e.g. fallacy of the well-chosen example). I want to limit the chance of something unexpected happening, though, and a small sample size puts me at risk.

From a marketing standpoint it can’t hurt to be net long puts. I cannot reliably backtest efficacy because the number of cases where they would have saved my bacon is so few. Especially because I have to believe “my worst drawdown is always ahead,” though, long puts are insurance against something horrific. I feel confident in saying that while the broad market has never gone to zero, if people are exposed to huge risk should that happen then they would be reluctant to invest. On the flip side, while the broad market has never gone to zero, if being net long puts would result in windfall profits should that happen then people would feel quite secure about investing.

It couldn’t hurt to have both, either. Maybe I should choose a conservative profit-taking target and also buy an extra long option to serve as catastrophic insurance.

Getting back to the matter of the spread backtesting itself, my deliberation led me to believe going wider on the spread might be necessary to overcome slippage. The greater credit received would somewhat offset slippage but result in a lower ROI and it wouldn’t help in a market crash where slippage would be gigantic either way. It’s a bit sad if I have to widen the spread simply for the sake of backesting but I may have to since I don’t know how spreads have traded in real time.

I could aim for a fixed credit but that might result in variable short deltas and probability of profit. As the underlying increases in price, a fixed option premium is available farther OTM. I am assuming this would pertain to spreads too although with the long option offset, I really don’t know for certain until I look at some historic data.

Musings on Put Credit Spreads (Part 6)

3AM Eternal! My last entry was not actually posted at 03:00 and that was the April Fools joke. These days I sleep in the middle of the night unless it’s a rare weekend. Classic song, though…

In any case, I left off considering some possible ways to avoid the slippage problem.

Another possibility would be to incorporate slippage as a(n) discount (exaggeration) to profits (losses). A ballpark number I have in mind is 3%. If I sell a 10-point spread for $0.60 and close it for $0.10 then I make ~5% on it. If I discount that to 2% then I make $0.20, which is about $0.14 slippage per spread traded (leaving $1/contract for commissions).

I should expect execution to deteriorate in fast-moving markets. Since I don’t vary execution in backtesting, I should be liberal with slippage to get a realistic average. I’m not sure $0.14 per spread quite accomplishes that.

A few years ago I backtested 100-point put spreads. I intended to let the long option expire worthless, which cut slippage. The exception was when the market moved down significantly causing the long option to increase in value. With a profit-taking rule I could realize a gain before the market reverses causing the option to expire worthless. This would also prevent me from realizing a huge gain in the rare situation no reversal takes place. While a material difference exists between the two, it may be a moot point if the small sample size lacks enough consistency for future relevance.

More specifically, over the last 15 years a deep-out-of-the-money put would have gained material value roughly 11 times. These include: Sep 2001, summer 2002, June 2006, Jul 2007, Jan 2008, fall 2008, winter 2009, May 2010, summer 2011, Aug 2015, and Jan 2016. Of these 11, I’m more certain about seven (significant volatility increases). I could divide and say it happens every 16-24 months but that is an artificial construct. It happens so infrequently that I should not be surprised to go multiples of 16-24 months without experiencing it.

Inclusion of a profit-taking rule will realize more consistent profits at the cost of a few huge gainers. I cannot determine whether this is roughly equal because of the small sample size, though. This is a rare instance where 15+ years of data is probably not enough. It might amount to the difference between having a quantifiable profit-taking rule and the undefined potential of purchasing a lottery ticket.

Musings on Put Credit Spreads (Part 5)

This post does have a subtle April Fool’s joke. Do you see it?

To finish up with my clarification of Part 2:

> Dream of the Holy Grail remains alive and demand to
> purchase the education in hopes of profitable trading
> is directly proportional to marketing efficiency.

If I have spent money to join a trading program or community then confirmation bias may predispose me to believe in it. As a consequence, I may search for evidence that it works and ignore any suggestion to the contrary. This could also involve future leaks (discussed in last post). Marketing efficiency (e.g. fallacy of the well-chosen example) may determine whether I become sold on the program in the first place.

Since I am completely disorganized in this series of blog posts, let me toss in a mention of Charles Cottle. In Parts 2 and 3 I discussed how complex spreads are often combinations of simpler units. Cottle talks a lot about this in his writing.

Now then, I will finally return to discussion about a methodology for backtesting put credit spreads. I will sell the first option under 9.90 delta or below to account for plasticity in the matrix greeks. The option to buy will be 10 points OTM. I will use $64 slippage per spread round turn.

Looking at two arbitrary dates, in 2010 at 637.55 with 46 DTE the vertical spread just described fetches $0.65 whereas in 2004 at 397.44 with 47 DTE this spread fetches $0.40. The differential between the two is not as much of a concern as the fact that slippage alone will completely negate the credit.

Houston, we have a problem!

One solution would be to decrease the slippage. Lately I have been finding $0.10 for a $3.00-$4.00 option is not unreasonable. On a spread that could be $0.20, which is $42 round trip (34% less).

A second solution would be to use the “last” prices instead of the “mid.” If the option traded at that price before then it seems reasonable I should be able to get that price again. One problem would be if I were selling (buying) a spread and the “last” quotes reflect purchase (sale) of that same spread, which would result in an artificially increased (decreased) price. I would never be able to tell what side the “last” was on. Another problem is that the “last” can be very stale if the option has low volume.

I will continue piecing together these backtesting guidelines in the next post.