It seems you have JavaScript disabled.

Ummm.. Yeah... I'm going to have to ask you to turn Javascript back on... Yeah... Thanks.

The 200 Day Moving Average Is Your Friend at Trader’s Narrative

The work of our very own Wayne Whaley is being featured in the current edition of the Technical Analysis of Stocks & Commodities. His contribution to that fine magazine involves the utility of using the 200 day moving average as a guide to enter and exit the stock market.

First to illustrate why the 200 day moving average is deserving of all the attention bestowed upon it, he looks at the returns for 7 different moving average time periods (readers of this blog got a sneak peak into this study last month!):

200 average moving study

At an average annual return of 10.85% the 200 day moving average easily stands out among its peers. But it isn’t enough to topple the 11.5% return provided by a humble buy & hold strategy (with the boost provided by dividends).

Wayne concludes:

When measured solely on total return, the 200-day moving average crossover trading strategy has some utility as a trend-following system, but its total return (9.2%) is comparable to the return of the S&P (8%) over this time period and actually under the S&P buy & hold plus dividends (11.5%) return.

This particular trend-following system has some utility in a long-term trading model but is not the net total solution to trend following and should only be used in combination with other forecasting tools.

While that may be true, I wonder if using leverage would allow for a market beating strategy using the 200 day moving average. Even if we use 1.5 or 2 times leverage (and are willing to stomach the volatility) the returns would handily exceed the buy and hold with reinvested dividends. This would be the quintessential trend trading system and about as simple as you can get.

Father Knows Best
And before you turn up your nose at such a basic trading plan, consider that it really does have merit. One particular stock market newsletter bases its whole strategy on the 39 week moving average. Technically that’s 5 days short of a 200 day moving average but close enough it seems. For the curious, that newsletter would be Richard Fabian’s “Successful Investing”. Unfortunately, his son, Doug Fabian, who took over the newsletter in 1992 has strayed from this simple strategy. His subscribers are poorer for it.

Ned Davis Research Knows Best
In a similar vein, NDR was commissioned to do some backtesting for a strategy that switched between the 200 day moving average and money market funds. For the period from late 1979 to May 2007 (when the study was done) this would have provided you with an 11% return and even more impressive, it would have delivered such return with less volatility than just buy and hold. Also interesting, a long term moving average strategy would perform better during volatile time periods like the decade we just went through than smooth ones like the 1990’s. Find out more about Fabian, the NDR study and the 200 day moving average as a signal from this article by Hulbert.

A related area ripe for exploration would be the “Golden Cross” (where the 50 day moving average closes above the 200 day moving average. This similarly simple strategy also has ample historical evidence of profitability.

This gives me even more confidence that exploring the use of leverage with such a strategy would be the smart way to go. Oh, before I forget, you can read Wayne’s complete article here.

Enjoyed this? Don't miss the next one, grab the feed  or 

                               subscribe through email:  

11 Responses to “The 200 Day Moving Average Is Your Friend”  

  1. 1 burt

    I have posted a day-by-day graph of the equity of using this system in the 1st post of my blog

  2. 2 wayne


    Note that the 10.85% return you quoted for the 200 day moving average was only for the long side of the 200 day strategy. You had to go to Tbills on the shorts to get the returns up to 9.2% annually, which would be comparable to buy and hold, with much less volatility, a smaller Beta, max drawdown, Sharp ratio etc.

    Something worth exploring sometime in the future for one of us number crunching dudes, is that, notice that the average return going short the 50 day moving average was -10.15%, which suggest you really actually want to go long when the price pulls back to the 50 day average. Especially if you are still above the 200 day moving average. There has to be a trading system in that somewhere, but I haven’t gotten around to thinking about it in detail, yet. Maybe a reader can suggest something or do some additional studies..

    Also, the 200 day moving average worked well over the time period tested. If we go into the a secular bear market as many believe, many time honored strategies will struggle, because they were optimized over a different type market. It is possible that the shorts could turn profitable if we go into an extended bear market.

    Lot’s of permutations of this study can be done. For example, a problem with the simple moving average is that the day being dropped off the end of the moving average is as important as the day being added, which goes against common sense. So the first response many statisticians will have is to use some type of exponential or linearly weighted moving averages. But the more parameters you enter into the system, the more prone the system is to being curve fitted to the time period tested. ‘Simple’ is usually better going forward

  3. 3 Mike C

    A related area ripe for exploration would be the “Golden Cross” (where the 50 day moving average closes above the 200 day moving average. This similarly simple strategy also has ample historical evidence of profitability.

    Someone got the data and modeling skills to test this? Wayne? Babak? I’d love to see these results against just the 200 DMA. Just eyeballing the chart, the golden cross and death cross seem to have been particularly effective from 1995 to present which is my entire investing lifetime.

  4. 4 Mike C

    What would make for an interesting test IMO would be to combine valuation and trend together to essentially test 4 quadrants (similar conceptually to what Hussman does, but using a different valuation metric and different technical indicator then whatever his “market action” is).

    So quadrant 1 would be undervalued and trending up (over 200 DMA). Maybe use Shiller’s P/E and use the median as the dividing line. In quadrant 1 you would be heavily leveraged.

    Quadrant 2 would be undervalued and trending down (under 200 DMA). Maybe 50% invested

    Quadrant 3 would be overvalued and trending up (over 200 DMA). Maybe 100% invested.

    Quadrant 4 would be overvalued and trending down. 0% invested. I suspect the bulk of the 2000-2002 bear market and the 2007-2009 bear market would have fallen into Quadrant 4.

    Just thinking out loud here, but I keep thinking that some combination of valuation and trend is superior to looking at just one alone. Incidentally, right now we would be in Quadrant 3 overvalued and trending up.

  5. 5 wayne

    Mike, I can test the golden cross without much difficultly and present the results soon. I can even do variations on the 50 day, (say 10-100) to see what would work best. Again no guarantee that is the way the mkt will work in the future. I suspect that since the 50 day is used in the Golden Cross, that someone has already optimized to the 50 day.

    Your second suggestion, based on tying to valuation, all falls back to your ability to define whether your indicator is operating in a bear market or in a bull market before the fact, and that makes all the difference in your forecasting ability. Who is to say if we are overvalued or undervalued. For example, I am not sure there is total agreement on the fact that we are overvalued, The overvaluation camp will argue that bear mkts end with PEs in single digit, while the undervaluation camp will present their regression curves that show PEs or earnings yields as a function of interest rates and suggest that we are at least fairly valued. Especially given that earnings have exceeded forecast 4 quarters in a row and LEI trends suggest that could continue to be the case for sometime.

    Also, I deviate some from the subject matter here some, but most of the charts I have seen on valuation, you would actually come out better buying when the market is overvalued and selling when the market is undervalued. I suppose that is because the market has discounted the future earnings trends by moving to high or low PEs prior to the fact.

  6. 6 Mike C


    Appreciate the response. To your point, perhaps I didn’t articulate it well but I really wasn’t trying to identify anything before the fact or argue what constitutes overvalued or undervalued. What I was suggesting is taking as your starting point for testing that the market is “overvalued” if the Shiller P/E is above its median value or “undervalued” if below its median value.

    I have mixed feelings about conflating P/E ratios with interest rates to determine over or undervaluation. Cliff Asness in a paper took this methodology (really just the Fed model) to task, and I’d say that Hussman has torn it apart pretty good as well:

    That said, I think it might have even been you that posted something intriguing on P/E ratios and interest rates on this blog, but if I recall you only went back to around 1960 and it just so happens that everything prior to 1960 completely breaks down with respect to P/E ratios and interest rates. Vitaly Katnelson has done some work on this. It is beyond the scope of this post, but there isn’t a sound theoretical foundation for lower interest rates to result in higher P/Es and the empirical evidence is mixed depending on whether you look at the 50 years after 1960 or the 50 years before.

    So as a valuation metric to determine over or undervaluation that leads me to one of the cyclically adjusted metrics like Tobins Q or Shiller’s P/E.

    have seen on valuation, you would actually come out better buying when the market is overvalued and selling when the market is undervalued. I suppose that is because the market has discounted the future earnings trends by moving to high or low PEs prior to the fact.

    Ahhhh…but the devil is in the details here. When you say “come out better” what holding time frame are you referring to? 1-day, 1 week, 6 months, 12 months, 3 years, 10 years? My gut instinct tells me that for any time frame under 5 years, probably 7-10 years valuation is completely irrelevant, but it does matter for 10-year holds. Buying stocks when they were single digit P/Es in 1932, 1942, 1949, 1974, 1982 all worked out well for the subsequent 10-years.

    What I am suggesting and I really don’t know is that by combining the rare instances when the market is BOTH cheap by the low P/E standard regardless of interest rates and the trend is up you are capturing possibly the few instances where maybe subsequent 1-5 year performance is usually stellar.

  7. 7 wayne

    Mike C.

    It is a pleasure to converse with someone who has done his homework. Before I spend the rest of the evening studying the links that you have thoughfully provided, I will make a few comments without the benefit of doing any research to support them (I leave my data files at the office to resist the temptation). It is this person’s opinion that if one chooses to go far enough back in time, you can find periods where all market timing systems fail miserably. I believe that is probably the reason why those market timing institutions that have been in business for several decades, never let the market get far away from them and abide by strict risk management procedures. They have the wisdom to know that all systems will likely not do as well going forward as the time period in which they were tested on. The one possible exception being sound trend following systems that will never show mind boggling backtested results but possess inherent characteristics which ensure that they will never get blown out of the water going forward.

    I am in agreement with you on most all of your comments. I am however, fairly confident that I could use statistical sampling techniques to show that, as you alluded, since 1960, there is an undeniable inverse relationship between PEs and interest rates. And yes, if you want to go back to the 1950’s, the relationship breaks down. Your links may suggest ow, but I am fairly confident of such. I want to recall there were a couple of years in the 50’s where there were single digit PEs with low interest rates (below 5%). I recall seeing the charts and thinking it was the buying opportunity of the century. So one has to decide, do I want to rely on the strong evidence that exist over the last 50 years, or choose to rely on statistics prior to that. I honestly can not say what the correct answer is. I suppose it depends on which serves your purpose. Certainly if the market or economy starts showing characteristics (such as deflationary) that haven’t existed for decades, it is imperative that one go back in time for enough to draw the appropriate correlations. My studies have been more a function of the data I had available to me on the day I was motivated to, or found the time to study them.

    Which leads me to change the subject a bit, I want to point out do readers that I do not do market research for a career, I trade (or attempt too) for a living. So I have never felt the motivation to keep 50-100 year databases on items such as earnings and interest rates. For the past 25 years, I have done numerous studies for no one but myself, that simply resulted in a few lines of code in a model that I was working on and to then be forgotten. When I stumbled onto Traders Narrative, I thought from time to time, I would pass along some of my research and it would serve as somewhat of documentation process. The idea exhanges are also stimulating for me since I am a retired system analyst in Huntsville, Al., a city filled with rocket scientist, but not an abundance of economist or market experts.

    And yes, that is correct, I live in Huntsville Al. and have a son who attends UAH, the school where a fired teacher attempted to kill the entire biology dept last week.

    So let me to check out your links.

  8. 8 wayne

    Mike C,

    I have not had enough time yet to study all of the links you provided, but I did give them a perusal for the timebeing. Four comments for now.

    First, Hussman’s chart actually confirmed much of what we conjectured and that was that there has been a very strong correlation between rates and earnings since 1960. A strong 50 year correlation, that did not exist prior to that.

    Secondly, let’s just use common sense. Think about this example for a minute. You are 55 years old, you have been penniless all your life and you just inherited a million dollars, the vast majority of which you would like to invest for your retirement in 10 years.

    Case A.
    The stock market is selling at a PE of 20 or an earnings yield of 5%, you can get 15% (1980ish) on a 10 year CD. What do you do with your money?

    Case B.
    Similar to Case A. the stock market is selling at a PE of 20 or an earnings yield of 5%, you can get 3% (2009ish) on a 10 year CD. What do you do with your money?

    I used an extreme case to make a point and I don’t want to insult you with the simplicity of this example, but in Case A, you would obviously put the majority, if not all of your new found wealth in interest bearing securities, and in case B, you would obviously put some larger percentage in equities. Regardless, assuming some incredibly unusual economic circumstances that differentiate Case A from Case B, the stock market is going to be more attractive to investors in Case B than in Case A, despite the fact that in both cases the market was selling at 20 times earnings. Thus making it hard for me to accept the idea that the market in Case A that was selling at a PE of 20 carries the same valuation attractiveness to investors as Case B that also trades for 20 times earnings.

    There are a myriad of factors that play into the relative valuation of the market, but I find it difficult to argue that as long interest bearing secuities are the primary source of competition to equities for investors assets, that next to

    1) the prospects for changes in future earnings relative to current earnings.

    2) the level of risk aversion that investors have, depending on what they have experienced economically (exp, a depression, fear of future joblessness, etc) in their lifetime,

    that the level of competing interest bearing securities is the third most important factor in determining the valuation of the market and should be given some consideration in defining the valuation of the market.

    the third comment, which pales in importance to the above two, is that the relationship between interest rates and earnings should not be perceived to 1 to 1 over the entire yield curve as is done in the Fed model. I can expand on this at some point in the future.

    The fourth comment, which is tertiary in significance is that the 10 yr treasury bond used in many of the comparisons is not necessary representative of the entire yield curve that competes with equities for investment dollars.

    Mike, I have some other comments, which I would prefer to share with you privately. If you have some way of getting me your email address, either on here or via Babak, I will pass them along to you.

  9. 9 Mike C


    Thanks again, and I’ll echo back at you that it is a pleasure who brings data and logical reasoning to the conversation. Getting close to my bedtime, so I’ll try to keep this brief. Your common sense example of Case A and Case B does highlight that interest rates matter from a competitive standpoint and certainly from at least that perspective interest rates do matter to equity valuation. I’m just trying to ascertain how much.

    I recall 2006-2007 when both Tobin’s Q and Shiller’s P/E were literally screaming from the rooftop the market was very overvalued yet the Fed model based on forward earnings or even trailing earnings relative to interest rates said stocks were “cheap”. I could point you to many bloggers including professional money managers making repeatedly and prominently making that argument but I won’t embarass them. That was just before the worst bear market in 60 years and a 60% drawdown peak to trough.

    Now did the market drop 60% because it was “overvalued” or because there was a credit crisis/financial panic and stocks were actually reasonably priced in 2007? Ultimately, I find that a nonsensical question because to my way of thinking the purpose of valuation is to tell you when the risk of stocks are high particularly to any unanticipated negative economic issues. I find it interesting/notable that the market finally bottomed once the Shiller P/E crossed below its long-term median valuee:

    You raise an interesting question about timeframes for doing historical analysis and modeling. Is 1900-1950 even relevant to today’s economy? Not sure what you’ve read or studied on Long Wave/Kondratieff cycles, but according to that theory the economy moves in 70-80 year cycles based partially on the massive buildup and then reduction in credit/debt. According to the theory, we are in the Kondratieff winter which is most closely paralleled by the 1930s/1940s which would seem to suggest eventually we will see those 1940s type stock valuations again even with low interest rates. Perhaps one more buying opportunity of the century will come sometime in the next 5-10 years?

    Ok, time for bed. My e-mail is

  10. 10 MachineGhost

    There is no magic in moving averages (or fundamental ratios aka investor sentiment). Such studies and with paramater optimization have been done to death in many academic studies and several mainstream books, all pointing to no statistical significance. Come on, this stuff is all at least 40 years old, pre-dating the age of computers!

    Furthermore, Ned Davis Research is on the board of advisors to Hussman. Why reinvent the wheel? Hussman will pwn you on anything you waste your time fooling around with. Pay the man and get on with your life… there are other areas in finance ripe for harvesting.

    I’ll tell you the “secret” to the business cycle. Mutliple the value of pi by 1000 and that is the number of calendar days from the major low to a major low, excluding cycle inversion. Halves will mark minor highs or lows on the upleg and downlegs. There is no magic here either.

  11. 11 wayne

    Machine Ghost,
    If you will read 8 and 9 above, you will note that I present the results with some degree of skepticism as well. But modeling probabilistic models is what I have done all my life and choose to do, whether it be military applications or the odds for the market on the 8th Friday in the month of August with a full moon. I appreciate others research but prefer to lean on my own. I don’t believe anyone is forcing you to read the studies and I trust your trading must be going incredibly well. Is there a particular resson you don’t use your real name on your post?

Leave a Reply