USO Short Put 45 DTE Cash-Secured Options Backtest

9 Comments

1. Tomaz
   August 17, 2019 @ 9:36 am

   I know most of your SPY backtested results show that buy and hold SPY is best, but I have some strategies with options that beat buy and hold, specially on risk basis and all produce much higher sharpe ratio which means all of them can be leveraged up to have same drawdown and standard deviation as buy and hold spy but in this case much higher returns. I am about to implement some mix of those, but maybe you can also backtest them.

   Links where you can find strategies. But the theme is quite the same for all. Be invested by selling monthly ATM put, but with different twists. One strategy sell only 67 % when VIX is above 20, and sells 150 % when wix is below twenty. The other very promising strategy is to only invest when SPY makes new high. Either directly in SPY or what I would implement is selling 200 % of ATM monthly put if SPY closes at new high or 10 month high. One should have some lookback to get in the markets after 2008 or so selloff so 10 month high should be ok. Sure there will be some whipsaws when you invest just at the top, but you would avoid big drawdowns if you only stay invested when markets make new high. It would keep you in market almost all the time and avoid big drawdowns, which is what improves long term results and makes much lower volatility.. The rest of the time you either stay in cash or in bonds..
   Another strategy which MEB talks about in the paper below is to sell ATM monthly SPY put and purchase TAIL protection for 10 %.. And when he combines it to 60 % short ATM spy + 40 % tail risk protection + the rest of the money in cash or short term bonds, you get like 6-8 % returns but what is most important less than 10 % drawdown and only 5 % standard deviation.. Which means one could lever such portfolio x 2 to get above 10 % returns and still have very shallow drawdowns and SD…

   If you are interested, please check links below, I think you could find some useful stuff for yourself and for your portolio and for us readers it would be great if you could test some mix of the below papers to maybe get even better results..

   https://mebfaber.com/wp-content/uploads/2017/10/TAIL-Final-10.03.17-B.pdf
   http://econompicdata.blogspot.com/2016/02/avoiding-bear-markets-to-improve-risk.html
   http://econompicdata.blogspot.com/2016/05/the-smoother-path-putwriting-at-high.html
   http://econompicdata.blogspot.com/2015/09/the-case-for-put-writing-further.html

   Reply

   • JEI
     August 17, 2019 @ 4:28 pm

     What do you mean 67%, or 150%?

     Reply

   • spintwig.com
     August 19, 2019 @ 12:36 am

     Long time no see Tomaz. Glad to have you back.

     I was finally able to refresh the SPY short put study and add Sharpe ratio as a statistic. Just taking a glance at the short put and short vert put spread, several have higher Sharpe ratios than b/h SPY. The challenge with levering those strategies is margin requirements. In theory one could lever such strategies as needed to generate the desired result. However, in practice there’s a ceiling due to reg-t that prevents the necessary leverage from being applied.

     As for the links, starting with the first one, I’d like to better understand how they’re coming up with the data on page 15. They spent 14 pages leading up to making a point on a strategy, only to show a single table and speak nothing to methodology. For instance, how did they backtest the data to 1986? Option data isn’t available prior to 2005 for SPY. Also, and I may have missed this, it’s not clear what benchmark they’re using to measure performance. Investing 90% of the portfolio in 10-year treasury and the remaining 10% to buy 5% OTM options (are these call or puts being bought – paper doesn’t say) will generate a return, yes, but compared to what? Overall question: what specific goal, more than specific than “mitigate drawdown”, is the author is attempting to accomplish?

     Link two talks about market timing by selling positions after a 10% drawdown and holding in bonds until another signal and uses MSCI world index as the index for reference. Plausible on paper, difficult to implement in practice. Two things coming to mind: 1) realizing capital gains / tax drag for those in the US and 2) SPY drawdown profile and volatility is very different than MSCI – see the table in my spintwig vs ERN post . The 10% drawdown signal would have occurred 78! times since 1945 and lasted, on average, for only a single month. While talking about int’l, I just wrapped up some preliminary research on EEM option backtests. Really interesting results! That study is scheduled to be published on 8/30. Stay tuned 🙂

     Skipping ahead to link 4, interesting claim the risk-adjusted returns of short put options on SPY is higher when VIX is lower yet the absolute returns are higher when VIX is lower. Based on this, they propose increasing capital allocation to short puts when VIX is low and lowering allocation when VIX is high. This happens to be the opposite of how I traded, allocating little-to-no capital when VIX is low and hustling when >17.5. This will be tricky to backtest since the tool I use only allows for IV rank, not VIX, for trade entry. Let me look into this a bit more.

     Link 3 references 2 and 4 and proposes combining these concepts into a single strategy as a modified Cobe PutWrite (PUT) strategy on SPY. Someone implementing this would need to be looking at the market daily and managing positions continuously. Need more info on how the historic performance was determined.

     Reply

     • Tomaz Korosec
       August 19, 2019 @ 8:03 am

       Thanks Spintwig. I really like your blog and backtests so I am throwing some ideas here and there. Maybe we develop some good strategy out of many and get a system with good enough returns and low drawdowns.. Holly grail put options selling course anyone? 😀

       Faber paper is all over the place, I agree. I had to look hard to see what kind of strategy they suggest and when I had a question about it sent to Mebane, he just replied that it is too much hassle to replicate manualy, one shoudl just buy their etf TAIL and PUTW and the rest in bonds..

       But TAIL is is just buying 5 % OTM puts not calls. This is for sudden sell-off protection which happens from time to time. So you like loose 20 times when those puts expire worthles but when dramatic sell-off happens they “explode” in value getting you 1:20 returns.. Or something like that, in theory.. Is the same as SWAN funds use. Results are not bad, what you get using strategy like that is more normal distributon. You cut off left tails and right tails. Means you underperform in huge bull markets but at the same time you get much better results during bear markets which makes everything much smoother and easier to invest. I would say strategy like that makes even more sense for someone nearing retirement. You can barely afford looing 50 % of your portfolio when you will soon enough start drawing from it. And it makes sense for me, because I am risk averse 🙂 Not yet near retirement at 39, but would really hate to see my portfolio got cut in half and I doubt I could sleep from that time one 😀

       What makes all those studies flawed is you ask me is the fact that all of them use bonds for money that is not alocated or when not invested in stocks and we all know we had like 30 years bonds bull market. So from 2019 to 2039 results will for sure be much worse for a strategy like that. Bonds added a lot of returns, even money market yielded like 4-5 % back then.. Now? …

       As to VIX and put selling, I agree with you completely. My thinking was the same as yours, high vix, sell more because premiums are more juicy and you can get more money or go further OTM. But on the other hand I do understand their logic as well. When is VIX low? When markets run higher month after month. So one should take this gift and sell more premium. Markets just churn higher, vix it getting lower and lower and if you want to make decent returns, you have to sell more. Ok you will eventually get caught in reversal and loose a bit more, but other 90 % of positive months should more than make up for this..
       Then when things turn ugly specially 2008 ugly, one is better of limiting risk.. Sure you can get hefty premium at VIX = 20, but VIX could go to 30 and then 50… If you limit risk and exposure during such volatile and dramatic times it is beneficial.

       It is the same when I was thinking of selling 5 % OTM monthly SPY puts. The logic was that I always have 5 % protection this was. Even more because of premium received. So I would rarely loose money in any month. Anything less than -5 % month is a winner. On the other hand one can sell ATM monthly and lose much more in negative month, but I think again this logic is false just like with VIX. You will loose more, but because of upward drift if one sells ATM it gets much further ahead. So with ATM I could collect 2.000 USD per month, going 5 % OTM I can collect only 1.000 USD. In 36 months with ATM I could have like 30 winning months and collect 30.000 USD more premium, other 6 months I loose, but not that much more to make it worthwhile selling OTM.. But selling OTM is more apealing because you have less loosing months and it is easier on the psyche.. From returns standpoint I think sellint ATM makes sense. You gain more and have about 2 % safety buffer in it, depends on the VIX. So even when SPY ends just where it began you still make 2 % monthly return and for any month down 2 % or less you loose nothing..

       What I was also thinking could be beneficial is selling monthly ATM, but rolling it up each week. One would only roll up to have about 50 delta all the time. This would mean one always have 2 % drawdown help but gains much more in huge up months. Beause if you sell ATM and it is 3 % up week, you still have 3 weeks left but now your ATM option is only 30 delta option and any further advance you make less and less. By rolling up one could follow the market up.. Sure on big reversals this again looses more than if you left option alone without rolling, but in upwards markets you would also gain I think a lot more having constant 50 delta exposure.. Now how to test this strategy I have no idea. Would be hard I think 🙂

       I also kindy like idea of only investing if SPY makes new high with look back period, getting out sooner than 10 % drawdown (close below 10 month EMA for example), but then getting back in the markets when drawdown is 30% or more.. This would be like trend following and mean reversal combined.. Because a lot of money is made after huge declines (2008) and also in trend following when markets just advance month after month.. But I know easier said than done. You can get back in after 30-40 % drawdown and markets plunge more, or maybe drawdown is only 28 % and you miss the rally… Kinda like last year. If one took advance of february and december sell off one could make a lot of money, but it was had to see such big and fast recovery back then.. 🙂

       Reply

2. Tomaz
   August 17, 2019 @ 9:47 am

   If I may allow a second comment. Could you backtest selling futures options? A paper came out in 2015 which showed that IV in futures is much more overstated than actual volatility on equities, bonds and currencies. Like 3 times as much. Let me copy their findings below, which is stunning:

   “They found that shorting volatility offers a very high and statistically significant Sharpe ratio: 0.6 equities, 0.5 fixed income, 0.5 currency, 1.5 commodities, and 1.0 for a global VRP composite strategy (which is dramatically higher than the 0.4 Sharpe ratio for the market beta premium). However, they also cautioned that shorting volatility strategies is not a free lunch in that they come with occasional, but substantial tail risks”

   For the last year and a half I mostly sell put options on crude oil and confirm that from time to time they are incredibly juicy. You can sell like 1 month out put on crud oil which is 25 % lower than current price and get like 500 USD. That is 1 % return. Contract if crude reaches that level is worth 42.000 USD and you get 500 USD. Try that with SPY… If you sell SPY 20 % below the current price you get pennies. You have to sell almost ATM to get 1 % return. Is crude more volatilty than SPY? Sure. But that much more? And at that level? Yeah, last year crude fell from 75 to 45 before bouncing back, but at levels it is now, I think every sell off below 50 is a buy and selling 40-42 puts one or two months out gives you a nice return. At that level I am willing to just take a delivery of crude futures and be long oil at 42.. With OPEC limiting production one can be quite sure oil will not be below 45 for long time..

   Anyway what I wanted to point out is that selling premium works much better with commodities futures options. One just have to make sure to limit exposure to 1 contract only because of leverage and to stay well below the current price to have a huge safety margin. But because you know how to backests and you like selling premium I think part of your focus should be on commodities futures rather than selling premium on SPY or any equity.

   Oh and you backtested USO. Try also on XOP 🙂 XOP has some hefty premiums sometimes. I just sold some one month put options a few days ago well below the current price and still received a hefty premium. Again much bigger than on XLE, let alone SPY, but ok XOP is a high beta ETF.. 🙂

   Reply

   • spintwig.com
     August 19, 2019 @ 12:45 am

     But of course —

     Do you have the link to this quote? Curious what a “VRP composite strategy” is.

     This may be true, but USO is also significantly more risky. This risk is reflected in the premium you’re collecting.

     I’ll have to see if the backtesting platform supports futures symbols. Will keep XOP in mind as a candidate for future research.

     Reply

3. JEI
   August 17, 2019 @ 4:27 pm

   So I guess my question is… why would you sell puts on an underlying that is falling falling falling? The reason it “works” on SPY is because we’re generally in a bull market. Seems like your strategy with USO should be selling calls.

   Reply

4. Uri
   October 27, 2019 @ 10:00 am

   Hi, thank you for your great work! It will be possible to make a similar backtest of USO for GLD and/or SLV?

   thanks

   Reply

   • spintwig.com
     October 29, 2019 @ 6:48 pm

     Thanks Uri!

     Sure, I’ll add those to the list.

     Reply