Search the Community

There are many potential ways to manage a short put trade, so in this article I’ll share some backtested research to look at the differences between a few methodologies. In SMPW, we benchmark our performance against an ETF that attempts to replicate a popular index, CBOE S&P 500 PutWrite Index (PUT). PUT uses a simple approach of selling front month S&P 500 puts and holding them until expiration. 33 years of historical data is available on CBOE’s website to see the results of this straightforward approach. I like to think of PUT as a broad measure of the “beta” of put writing, similar to an index like the Russell 2000 for US Small Cap stocks. We’ll test this methodology on 7 different underlying assets from 2007-2019 (the data period available in ORATS Wheel). We’ll also test entering at 45 days until expiration (DTE) with an exit at 21 DTE. Lastly, we’ll test a 30 DTE entry with exits occurring when 75% of the credit received has been earned or 5 DTE, whichever occurs first. We’ll look at both excess annualized returns, net of estimated transaction costs, as well as risk adjusted returns with the Sharpe Ratio. Sharpe Ratio is a popular risk adjusted return measurement that is calculated as annualized excess return divided by annualized volatility. Results Interpreting The Data There are many ways to interpret what this data is telling us. I prefer to increase the sample size when reviewing parameter choices by averaging results across multiple underlying assets. In this case, 7 symbols were tested over a period of 13 years, with entries assumed to occur every 7 days, creating a sample size of more than 4,500 total trades. A large sample size helps minimize the impact of any outlier trades that may have occurred during the sample period that might otherwise skew results in a way that could lead to false conclusions. Overall, it doesn’t look like there was a significant difference in results based on the trade parameters over this time period. This is good, as we prefer to see broad parameter stability. The 45 DTE to 21 DTE method produced average results that were slightly worse than the other 2 methods, which is interesting considering this approach is recommended by a popular options trading educator and brokerage firm. In SMPW, we enter our short puts around 30 DTE and look to exit when we’ve made 75% or more of the credit received or about 5 DTE, whichever occurs first. With lower priced ETF’s that represent International equities we typically wait to exit winning trades until they are worth a nickel or less, as certain brokers allow you to exit these positions commission free. The logic, which is generally supported by the data in the chart, is that rolling winning trades ahead of expiration when we’ve made most of the potential profit maymodestly increase returns over the long term since we expect the equity premium to persist. Exiting losing trades a few days before expiration slightly reduces the risk of large losses due to the negative gamma of a short option that increases as expiration approaches. Conclusion: The Power of Diversification My final point is meant to highlight the power of diversification. Looking specifically at the 30 DTE to 5 DTE results, we see an average Sharpe Ratio of 0.59. I had the ORATS Wheel combine together all 7 symbols into an equal weighted portfolio, and the result was a Sharpe Ratio of 0.76...a 29% relative increase. Diversification is a generally accepted way to either A. increase returns for the same risk or B. maintain the same return with lower risk. Diversification can be achieved in many ways, and it’s one of the most compelling opportunities for “craftsmanship alpha” in the portfolio construction process that is used in our SMPW strategy. Jesse Blom is a licensed investment advisor and Vice President of Lorintine Capital, LP. He provides investment advice to clients all over the United States and around the world. Jesse has been in financial services since 2008 and is a CERTIFIED FINANCIAL PLANNER™ professional. Working with a CFP® professional represents the highest standard of financial planning advice. Jesse has a Bachelor of Science in Finance from Oral Roberts University. Jesse manages the Steady Momentum service, and regularly incorporates options into client portfolios. Related articles: Combining Momentum And Put Selling Combining Momentum And Put Selling (Updated) Steady Momentum ETF Portfolio Equity Index Put Writing For The Long Run Can You "Time" The Steady Momentum PutWrite Strategy? How Steady Momentum Captures Multiple Risk Premiums Put Selling: Strike Selection Considerations

I love the software, and it has saved me literally months of keying in data, so I have no issue telling others about it. Hopefully, this will also provide some more insight into the Anchor Strategy and how it was developed. In 2011 when Anchor was first being developed, testing the strategy was a Herculean task. I did not have access to any files of old option data, backtesting software or any of the programs that are available today to assist in the process of backtesting. When we had a proposed strategy, we had to painstakingly go through ThinkOrSwim’s “ThinkBack” feature that has day end option pricing. An excel spreadsheet would be created, several years of data would be typed in by hand, and then the data would be tested. If a result was not what we wanted, and, for example, we wanted to move the subject option from a 50 delta position to a 33 delta position, then it would take two more days of work just to key in the new option prices. Every minor change to the test was an effort of days. Then we were introduced to ORATS, which has made testing simpler, and helped create the current version of Leveraged Anchor that we are currently trading. It also makes answering member questions about the adaptability of the Anchor strategy much simpler. For instance, in the past few months, three different readers have asked “Can I use the Anchor Strategy or the Leveraged Anchor Strategy to trade gold through the GLD ETF?” My initial thought was “Well, it should be possible,” but thoughts and feelings aren’t exactly evidence. So, I dove into ORATS to actually test the theory. The first thing to do is to determine the cost of the long hedge, which can be done without ORATS by simply pulling up today’s options prices and looking – if the cost of the hedge is more than the average yearly gain in GLD, then there is no chance Anchor will work. For example, if GLD’s average gain over the last two years was 5% per year, but the average cost of an annual hedge was 8%, then there would be no point in checking further as hedging would be cost prohibitive. On the day of writing this article, the GLD option we would use to hedge would be the September 30, 2020 option. With GLD currently trading at 138.72, the five percent out of the money put is the 132 put, which is trading for $2.24, or less than two percent of the price of the underlying. This makes it “seem” like GLD might be a good candidate for Anchor or Leveraged Anchor because it appears to have a cheap long-term hedge compared to the performance of GLD over the past two or three years.Passing the first quick test, we need to build out a model and see if the strategy works over a longer time frame. The next thing to determine is if it is possible to pay for that hedge in a given year by selling short puts against the cost of the long hedge. If selling short puts does not have a positive annual return on average, the strategy will not work. This is where ORATS comes in handy. We start with the software by going into “Wheel” and clicking on the orange button in the top right corner that says “Create New.” This takes us to the following screen: We enter the test parameters on this page. For the initial test, we used the same parameters as for the Leveraged Anchor Strategy short puts: Symbol: GLD Strategy: Short Put Days to Expiration: Target: 24 Min: 21 Max: 30 Strike Selection: Target .55 Min .49 Max .59 Exit Profit Loss %: Min: blank Max 1.66 Exit DTE 1 Then click “submit.” Processing can take several minutes, so be patient with the system. When done, it spits out the results: The good news is that there is a positive annual average return. It’s even better that the positive return is in excess of the annual cost of the hedge. However, for our testing purposes, the real value comes in the monthly returns the software provides: We can now use those monthly returns to start modeling a Leveraged Anchor Strategy in excel. I simply cut and paste the data into my spreadsheet, where I can then manipulate it. Our next step is to find out the returns of the deep in the money long calls (the leveraged portion of Levered Anchor). Here, we simply go back to “Create New” in Orats and change our strategy inputs to: Symbol: GLD Strategy: LongCall Days to Expiration: Target 366 Min 365 Max 430 Strike Selection Target .95 Min 0.94 Max 1 Remember, we use a minimum of 365 to ensure long term capital gains treatment on the long holdings. Since we’ll always be holding this to expiration, the rest of the fields are left blank Click “Submit” And we get the following results: However, this doesn’t look right. It says we’re only in the market 85.39% of the days, and we should be in the market close to 100% of the time. This most likely means our “days to expiration” field wasn’t big enough and there might not have been an option available. So we re-run the test, changing the max days from 430 to 460. After doing this, similar results come out and we still are not in the market enough. At this point, I go to the individual month returns to determine what is wrong. After digging into the data that was returned, we realize that the software doesn’t support GLD options prior to October 2009, so we had a long period of just being in “cash.” This issue can be fixed by editing our test in the “Date Range” category and changing the start date to 2009-10-01, which gives us: These results are not near as good, showing an annual return of just under two percent. In fact, this seems quite low, so we need to do a data check. Since this is a deep in the money call position, returns should be near what stock performance has been over the same period. GLD’s performance is easily found on yahoo finance or any other stock data site. Since November 2009, the current prices of GLD has increased from 115 to 138 or 20% (2% per year). We’re trading a 95 delta position, so we would expect our returns to be 1.9% per year and got 1.82%. In other words, these returns are likely correct (and I note that gold has not performed well over the last decade, at all). We then copy the same monthly data into our excel spreadsheet. For our last piece, we need to run the returns for the long puts (the hedge): Symbol: GLD Strategy: Long Put Days to Expiration: Target 365 Min 320 Max 430 Strike Selection (stockOTMPercentage) Target .95 Min: .93 Max .98 Date Range: 2009-10-01 to Current Exit DTE 21 Exit Profit Loss % Min .33 The last category (Exit Profit Loss %) is the only one I don’t like. Since we roll the long hedge when the underlying has gone up around 7.5%, it’s really difficult to determine what the loss on the long put would be at that point, as its variable depending on how many days are left in the long hedge, current market volatility, and other factors. For instance, if the market in GLD moves up 8% in the first week after opening the position, the long put may only lose 50% of its value. We would typically roll the position then, but the software would not. Whereas if the market goes up 10% in a week with a month to expiration, it may lose 90% of its value, but the hedge would have rolled a bit too early in the software. However, we are not using the software to do a to the penny backtest at this point, rather testing the theory. So the 33% is really more a “guess” for purposes of this initial test. If it passes, when conducting a “full” Anchor Test, I leave the Exit Profit Loss category blank, go back through manually from the long call data and flag exactly when the rolls will occur. This is a slight increase in labor but not much. (Though if ORATS were to build in a feature that had entries/exits based on moves of the underlying, that would be a great feature addon). And with all of the above information, we now have the return information for the short puts, the long calls, and the long puts, and can manipulate it as necessary to determine an “optimum” level of leveraged, hedging, and whatever else would like to include in implementing Anchor on GLD. Once that process is done, you can see how the strategy performs, which unfortunately, is “not that well.” Changing leveraged ratios around really doesn’t help much. Digging into the data demonstrates why – when there’s a big downturn in the price of GLD, the hedge does not move accordingly unless expiration is near. For instance, in September 2011, when the price dropped around 15%, the long puts only went up about 3.5%. (Differences in long term vs short term volatility). Anchor always performs the worst on instruments that remain “mostly flat.” Over the last decade, GLD has moved less than two percent per year on average. There are periods where Anchor would have worked quite well on GLD, for instance 2009-2012 and even into 2013. However, since mid-2013, the price of GLD has remained in a tight range, which isn’t good for Anchor. As this is not a historically unusual trading patter for GLD, it basically rules out using the Anchor strategy on it. Logically this makes sense, as historically gold has been a good inflation hedge. With annual inflation over the last 5 years or so being under control, one would not expect the price of GLD to move much – and it hasn’t. ORATS simply helped verify these conclusions. Christopher Welsh is a licensed investment advisor and president of LorintineCapital, LP. He provides investment advice to clients all over the United States and around the world. Christopher has been in financial services since 2008 and is a CERTIFIED FINANCIAL PLANNER™. Working with a CFP® professional represents the highest standard of financial planning advice. Christopher has a J.D. from the SMU Dedman School of Law, a Bachelor of Science in Computer Science, and a Bachelor of Science in Economics. Christopher is a regular contributor to the Steady Options Anchor Trades and Lorintine CapitalBlog. Related articles Anchor Trades Portfolio Launched Defining The Anchor Strategy Market Thoughts And Anchor Update Leveraged Anchor Is Boosting Performance Anchor Trades Strategy Performance Revisiting Anchor (Thanks To ORATS Wheel) Revisiting Anchor Part 2 Leveraged Anchor Update Leveraged Anchor Implementation Leveraged Anchor: A Three Month Review Anchor Maximum Drawdown Analysis Why Doesn't Anchor Roll The Long Calls?

Selling calls for a credit to help offset the cost of the hedge is, more often than not, a losing strategy over time in the Anchor strategy. It tends to hurt performance more than help it; About a month is the ideal period for selling short puts over both in bull and bear markets. This tends to be the ideal trade off between decay, being able to hold through minor price fluctuations, and available extrinsic value. Since options come out weekly, we’ll be using a 28-day period; Rolling on a set day like Friday is not the most efficient method of rolling the short puts. Rather having a profit target of between 35% to 50%, and rolling when that target is hit, leads to vastly improved outcomes. Waiting until profits get above 50% tends to start negatively impacting the trade on average. This month we’re going to look at another technique which has the possibility of increasing Anchor’s performance over time – namely reducing the hedge. Reducing the Hedge The single biggest cost to Anchor is the hedge. Depending on when the hedge is purchased, it can cost anywhere from 5% to 15% of the value of the entire portfolio. In large bull markets, which result in having to roll the hedge up several times in a year, we have seen this cost eat a substantial part of the gains in the underlying stocks and/or ETFs. There is also the issue of not being “fully” invested and this resulting in lagging the market. If the cost of the hedge is 8%, then we are only 92% long. In other words if our ETFs go up 100 points, our portfolio would only go up 92 points. A large hedge cost also has a negative impact at the start of a bear market as well due to the losses on the short puts. If the market drops a mild amount, particularly soon after purchasing the hedge, the losses on the short puts will exceed the gains on the long puts, negatively impacting performance. This loss is less noticeable as the long hedge gets nearer to expiration and/or market losses increase as delta of the long hedge and the short puts both end up about the same. However, as was seen a few years ago, if the market drops slightly, then rebounds, those losses on the short puts are realized and any gains on the long puts are lost when the market rebounds. If there was a way to reduce the cost of the hedge, without dramatically increasing risk, the entire strategy would benefit. A possible solution comes from slightly “under hedging.” Testing over the periods from 2012 to the present and from 2007 to the present has revealed if we only hedged 95% of the portfolio, returns would be significantly improved. Let’s take a look at the data from the close of market on September 14, 2018, when SPY was at 290.88. If we were to enter the hedge, we would have bought the September 20, 2019 290 Puts for $14.96. If we have a theoretical $90,000 portfolio, it would take 3.1 puts to hedge (we can’t have 3.1puts so we’ll round down to 3). At that price, three puts would cost $4,488 or 5% of the portfolio (almost historically low). However, if we were to say “I am not upset if I lose five percent of my portfolio value due to market movements; I am just really worried about large losses,” we could buy the 275 puts instead of the 290. The 275 puts are trading at $10.61 – a discount of thirty percent. This means we need less short puts to pay for the position, paying for the position is a simpler process, and rolling up in a large bull market is cheaper. Yes it comes at a cost – risking the first five percent – but given the stock markets trend positive over time, this pays off in spades over longer investment horizons. Even if you are near retirement, any planning you do should not be largely impacted by a five percent loss, but the gains which can come from (a) having a larger portion of your portfolio invested in long positions instead of the hedge (meaning less lag in market gains), (b) having less risk on the short puts in minor market fluctuations, and (c) paying for the hedge in full more frequently more than offset that over time. We will implement this in the official Anchor portfolios by simply delaying a roll up from gains. The official portfolio is in the January 19, 2019 280 puts. We’d normally roll around a 7% or 10% gain (or around SPY 300), instead we’ll just hold until we get to our five percent margin. OR when we roll the long puts around the start of December, we’ll then roll out and down to hit our target. Note – if you do want to continue to be “fully” hedged, you can do so. There’s nothing wrong with this, you just sacrifice significant upside potential and will be continuing to perform as Anchor has recently. If we had implemented this change in 2012, Anchor’s performance would have been more than five percent per year higher. This is not an insignificant difference. Related articles: Defining The Anchor Strategy Market Thoughts And Anchor Update Leveraged Anchor Is Boosting Performance Anchor Trades Strategy Performance Revisiting Anchor (Thanks To ORATS Wheel)

It immediately became clear this could be used, not only for put selling testing, but to test Anchor over longer periods of time than previously done and try to find other areas to improve the strategy, which has been a core part of SteadyOptions for quite some time. After two weeks of testing, much of what Anchor has evolved to over the past few years was validated, but we also identified some areas for improvement that should increase performance of the strategy. In this article, and a future one next week, I will discuss the conclusions from our expanded optimization, back testing, and review. Later articles will dive into the implications for the leveraged versions of the Anchor Strategy, as well as the benefits of expanding Anchor by diversifying into IWM, QQQ, DIA, and potentially other indexes. I. Selling Calls for Credit Anyone who has been following Anchor recently knows we have been on a quest to find other ways to pay for the hedge. The biggest drag on the strategy is the hedge, and any way we can improve paying for the hedge cost helps. In this decade long bull market, paying for the hedge has been particularly problematic, more so in recent years. For the past few months, we’ve been structuring and testing a variety of call selling strategies in an effort to extract a few more basis points of performance out of Anchor. Initial paper trading had us optimistic, as well as the manual testing done over the previous nine months. Further, the CBOE maintains covered call indexes, which seemed to indicate that the strategy should work. We were optimistic enough to start tracking it on the forums in the leveraged anchor accounts. What a thorough back testing demonstrated was that selling naked calls on indexes, SPY in particular, is a losing strategy, or at best a breakeven one, since 2007. This is true over shorter periods of time as well, such as since 2012. If calls, three weeks and one standard deviation out are used, since 2012, the strategy would have lost a 1.6% per year. If data from 2007 is used, so as to capture 2008 and 2009, the strategy still would have lost 0.20% per year. Trying further out in time over periods such as 28 days, 45 days, or 60 days were all losers. What about putting in stop losses or profit targets – also all losers. In fact, only through extreme curve fitting, was I able to identify any possible profitable naked call selling strategy at all, and only if you use the data set from 2007 to the present – even then performance would only have gone to 0.55% per year. Then if you remove 2008 from the data set, results immediately went back to negative performance. Changing from an at the money position, to a 30 delta position did not help much either. Since 2007, selling a 30 delta one month call, would have netted you only 0.22% per year – essentially flat. Changing the delta to 10 or 60 did not help either. This result was initially puzzling, as the covered call index (BXM) is up over 50% over the last five years and 75% over the last ten – until those results are broken down. BXM is not naked call selling, it is covered call selling. Given over the last five years, the S&P 500 is up about 75%, which means call selling is responsible for 25%, or more, of BXM’s losses. If covered call selling was profitable, there would not be this drag. By eliminating the gains from the long stock positions, the returns go to negative – as indicated by our testing. The conclusion? Simple Anchor will not be selling calls as a way to gain additional income to help cover the cost of the hedge and such strategies will be removed from the leveraged versions of Anchor. II. 14 vs. 21 vs. 28 Days for the Short Puts A few years ago, we switched from selling puts either one or two weeks out to three weeks till expiration. Doing so gives the strategy more time to “be patient” in the event of small market moves down and not realize losses that did not have to be realized simply due to normal market fluctuations. When making the selection on how many days “was optimal” we used the past 18 months of actual Anchor data for back testing purposes. ORATS confirmed that over that 18 month period, 21 days was the optimal time to use. However, with more data at hand, we have been able to confirm that 28 days is a significantly better time period over history than a 21 day period. In testing, we used two data sets on SPY – from January 2012 to present and from January 2007 to present. January 2012 was picked because after that point, SPY weeklies were fully available. January 2007 was picked because that’s the furthest back in time the software’s data went. From January 2012 to the present, selling puts 21 days out, with a profit target of 30%, would have returned, on average, 8.87% per year with a Sharpe Ratio of 1.63. From 2007, a return of 5.32% was realized with a Sharpe Ratio of 0.42. Merely by increasing from 21 days to 28 days, those numbers increase to 11.08% and 2.14 for 2012 to the present and 8.52% and 0.93 for 2007 to the present. This is a massive increase. It was enough of an increase to make us question the results. If this was a more “optimum” period, as theorized, such results should hold across other, similar instruments. For both QQQ and IWM, the results held. 28 days achieved much better returns on a put selling period than 21. We also learned that rolling the short puts on a set day (Friday), anytime you can for a gain, is not close to an optimal roll period. Significant improvements can be gained by ensuring the profits are somewhere between 25%-35% prior to rolling. Interestingly, waiting till profits are in excess of 50% began to have a negative impact on results. (Profits are defined as a percentage of credit received. So if we receive $1.00 for selling a put, a thirty percent gain would occur when the price declined to $0.70) Anchor’s current put selling strategy has us making an adjustment each Friday. If there’s a gain in the position, we roll, and if not, we hold – simple rules. Unfortunately, by adhering to “simpler” rules in an effort to make the strategy easier to manage for everyone, we cost ourselves significant performance. If we rolled when a profit target was reached, regardless of day, as opposed to on a Friday at any profit point, we would have increased our returns to 11.08% and Sharpe Ratio to 2.14 from the 6.87% return and 0.7 Sharpe Ratio we have experienced since 2012 on put selling. Using the same put selling ratio we have, that means by doing Friday roles, instead of a profit target, we’ve cost ourselves between 1.5% to 2.0% per year in total performance. III. Cautionary Notes One thing to be careful with software such as ORATS is over mining and getting confused by the noise. For instance, why not pick 25% or 33% profit target instead of 30%? Once you get that granular, randomness becomes a factor. One year 25% might work significantly better and another 33% and yet another 30%. Given that there are only 12-24 trades per year, which profit target hit on that tight of a range is a bit of “luck.” For instance, if we sold a put for $1.80, a 25% profit would have the price dropping to $1.35 and at 33%, $1.21. Prices move that much intraday frequently, so trying to target the “exact” price to do everything is a fool’s errand. The inability and/or inaccuracy in trying to overly optimize is a concern in getting exact performance numbers. It is not a concern in identifying major trends. If profit targets between 1%-20% all underperform profit targets from 25%-35%, over multiple periods, clearly we should be using 25%-35%. Similarly if all results are worse over 50%, then 50% is too high. Using different instruments (IWN and QQQ) provided further verification for this process, providing a higher degree of confidence. There is also a concern that any back testing is simply “curve fitting,” and that is one hundred percent true. The data we have discussed is pulling out the optimal curve. Some of this can be combated by using different time periods (e.g. 2007 to present and 2012 to present, or even smaller periods). If the conclusions founds hold over various time periods and various instruments (QQQ and IWM), it is more likely than not that the results are not merely curve fit, but instead due to consistent trends. However, as we all know, previous results are no guarantee of future performance – they merely increase our chances. In verifying our hypothesis about 28 days and profit target rolling, we went even more granular, across SPY, QQQ, and IWM. To our relief, the results generally stood. 28 day periods are more optimal than 7, 14, 21, 35 or 45 over the vast majority of time periods. There are years “here and there” were 21 days were better and one year on IWM were 35 days was a better period. But even in those years, 28 days was the second best performing. Over any multiple year period (from 2007 to the present), our conclusions held. The same held true for profit margins. Maybe one year 20% was the best and another 45% the best, but in all multi-year periods we looked at, using a profit margin of “around” 30% was much better than a static day roll with no profit margin requirement. Because of this, starting this Friday, we will be modifying Anchor’s rolling rules. Moving forward, we will be rolling to 28 days out and rolling once profits have gotten above 30%. This means we may be rolling on days other than Friday moving forward if profit targets are obtained. We could even roll several days in a row in a bull market. If profit targets are not hit, we will continue to hold, as the strategy currently operates. Next week we’ll discuss other Anchor modifications we will be implementing to ideally further improve Anchor’s average performance.