split-strike synthetic PP portfolio?

A place to talk about speculative investing ideas for the optional Variable Portfolio

Moderator: Global Moderator

Post Reply
atrchi
Full Member
Full Member
Posts: 58
Joined: Sat Mar 31, 2012 2:52 pm

split-strike synthetic PP portfolio?

Post by atrchi » Sat May 30, 2020 1:23 pm

Has anybody ever tried to run or backtest a PP portfolio modified to hold synthetic split-strikes instead of ETFs?

No I'm not talking about Bernie Madoff's split-strikes fraudulent fund.

A split-strike synthetic is where you sell an out-of-the-money PUT Option to finance the purchase of an out-of-the-money CALL Option at the same option price (but a different strike price). The resulting position resembles the risk/return profile of the underlying ETF during big moves, but returns exactly $0 when the market is range-bound within your chosen range.

Now here comes the brainy insight: Because of Implied Volatility Skew on stocks, the same-priced options are such that the CALL is closer to the current price and therefore has a higher probability of making money than a loss on the sold PUT. This strategy should have a statistical edge as long as the Implied Volatility Curve (IV vs Expiry) shows a smirk, which anecdotally has been the case since after the 1987 crash (I don't have access to historical option prices to verify this claim).

Example:

Basic PP as of the close on Friday- SPY $304.32, TLT $163.59, GLD $162.91

Instead of buying SPY for $304.32, you SELL DEC 31 '20 SPY 271 PUT @ $13.95 and BUY DEC 31 '20 SPY 320 CALL @ $13.73 (*)

The edge of this strategy is that the market right now values the EOY probability of SPY=$271 (an 11% drop) the same as SPY=$320 (a 5.2% gain).
i.e. "the market" is willing to give you any gains above 5.2% in exchange for you committing to assume any losses beyond 11%, from now to the end of the year. This is just intuitively wrong, and a better deal than holding SPY and committing to assume all losses in exchange for all potential gains.

To complete the PP:

- Instead of buying TLT for $163.59, SELL DEC 18 '20 TLT 150 PUT @ $3.35 and BUY DEC 18 '20 TLT 177 CALL @ $3.35

- Buy GLD ETF straight-up, because the Volatility Skew is inverted and not conducive to a split-strikes strategy:
e.g. if instead of buying GLD for $162.91, you SELL DEC 31 '20 GLD 146 PUT @ $2.90 and BUY DEC 31 '20 GLD 195 CALL @ $2.76, the market is only offering all gains above 19.7% in exchange for assuming all losses in excess of 10.4%. (incidentally, that makes it beautifully suitable for a Cost-Free Collar!)

Perform standard PP rebalance on 12/31, repeat every 6 months (i.e. on 1/1 trade the June or July 2021 options).

What would be the long-term result?

(*) Additional explanation for people who don't understand options:
- If the SPY finishes the year between $271 and $320, you make $0 on this part of the investment
- If the SPY finishes the year $271 or lower, you take losses on the value below $271 (same as retroactively paying $271 for the ETF)
- If the SPY finishes the year $320 or higher, you take gains on the value above $320 (same as retroactively paying $320 for the ETF)
- Just to avoid confusion, I don't differentiate "paper" gains and losses from realized gains and losses. Everything is mark-to-market.
- Assume no mid-course adjustments, hold every option until expiration
- Assume the mid-point for the option prices. One should never use a market order to buy or sell options.
Kbg
Executive Member
Executive Member
Posts: 2815
Joined: Fri May 23, 2014 4:18 pm

Re: split-strike synthetic PP portfolio?

Post by Kbg » Mon Jun 08, 2020 7:59 pm

I have no idea and backtesting something like this requires specialized data and software. If you don't have them, it's a pure guess.

The bold is my actual answer. My speculative answer is this is probably not a great idea with the PP. While from an options perspective, you have a point, from a portfolio perspective IDK.

Let's use your numbers and wag it a bit. You are essentially breaking even on the trade to put it on. That in and of itself has some worth, but I digress. The big problem I see is that you need to buy two calls (assuming nothing changes between now and when you need to roll the position which is a bad assumption but we need to make it) and SPY needs to move 10.4% in a year (5.2x2) for you to make any profit at all.

Test question: Do you think it coincidence that 10.2% is reasonably close to what the SPY makes in an average year?

Test answer: It is not.

Learning point 1: Options on broad popular indexes are EXTREMELY efficient.

Learning point 2: The cost of options on broad popular indexes are generally the expected move for the duration of the option bought.

With options you necessarily have to do a couple of things to be profitable. You must express an opinion as to the underlying's future course when putting on the position and your opinion must be almost precisely right. If one is good at both, options are a good venue choice. If one is not good at BOTH, they are an extremely tough venue.

In summary, there is no such thing as a free lunch.

Also, since 1987...do you think the smirk is actually irrational? I think it's highly rational. 1987 is pretty much the year I started investing and nothing in my life suggests it is not. Asian Contagion, Tech Bubble, Real Estate Bubble, COVID...and some lesser ones in between that were not as bad and short lived.
atrchi
Full Member
Full Member
Posts: 58
Joined: Sat Mar 31, 2012 2:52 pm

Re: split-strike synthetic PP portfolio?

Post by atrchi » Sat Jul 11, 2020 9:43 pm

Kbg wrote:
Mon Jun 08, 2020 7:59 pm
Also, since 1987...do you think the smirk is actually irrational? I think it's highly rational. 1987 is pretty much the year I started investing and nothing in my life suggests it is not. Asian Contagion, Tech Bubble, Real Estate Bubble, COVID...and some lesser ones in between that were not as bad and short lived.
This was actually a very insightful comment. Now I can sort of answer my own question:

- The strategy would only beat a straight PP if the implied IV skew smirk is irrational, specifically, steeper than it should be.

- It'll lose out if the skew is irrational in the other direction, i.e., not steep enough to compensate for future skewed distributions of returns.

- It's a wash if the skew is perfectly rational, subject to random variance over any given period.

I'd still be curious to know what someone has experienced if they have in fact tried this.
Post Reply