Comparing variations of a 200 dma trend following strategy


As you probably know, buying and holding the S&P when it is above its 200 dma yields an average return that’s similar to buy and hold, but with smaller drawdowns.

Scouring the internet for ways to improve the 200 dma, I came across 2 suggestions:

  1. Use a 40 weekly moving average instead of a 200 daily moving average. Supposedly, using weekly data or “only sell after the S&P breaks down below its 200 dma for 3 consecutive days” will improve returns because it decreases the number of false SELL signals.
  2. Use the direction of the 200 dma. I.e. buy and hold the S&P when the S&P’s 200 dma is going up. Sell and shift to 100% cash when the S&P’s 200 dma is going down. (Since the moving average’s direction changes less frequently, the model’s returns will supposedly improve because there are fewer false SELL signals.)

40 weekly moving average

Model’s rules

Buy and hold the S&P 500 when it is above its 40 weekly moving average
Sell and shift into 100% cash when the S&P is equal or below its 40 weekly moving average

Click here to download the data in Excel

  1. From October 2, 1950 – January 7, 2019: buy and hold yields an average annual return of 7.34%
  2. From October 2, 1950 – January 7, 2019: this model yields an average annual return of 6.72%

As you can see, the model’s returns are worse than that of buy and hold. Using longer time frames before buying and selling isn’t helping. While there are fewer false SELL signals, the gaps between the 40 weekly moving average and your SELL price increase (because weekly volatility is always higher than daily volatility).

Direction of the 200 dma

Model’s rules

Buy and hold the S&P 500 when the S&P’s 200 dma is going up (i.e. the 200 dma today is higher than where it was yesterday).
Sell and shift into 100% cash when the S&P’s 200 dma is going sideways or down (i.e. the 200 dma today is <= where it was yesterday).

Click here to download the data in Excel

  1. From October 19, 1950 – January 7, 2019: buy and hold yields an average annual return of 7.4%
  2. From October 19, 1950 – January 7, 2019: this model yields an average annual return of 6.26%

As you can see, the model’s returns are worse than that of buy and hold. The problem with using the 200 dma’s direction is that the direction lags the 200 dma itself. I.e. the “direction” gives you even later buy and sell dates than a breakout/breakdown from the 200 dma does.

Conclusion

As is often suggested, popular trading strategies suggested by gurus (like the ones above) do not beat buy and hold.
I cannot tell you how many articles and videos I’ve seen claiming “here’s the REAL secret way to use the XYZ indicator”. And then when you run the backtest, you realize that this “secret method” does no better than the original.

2 comments add yours

  1. Hi Troy,
    Myself I have difficulty believing that the difference between the regular model with SSO and the safer 200 DMA model is only 30% vs 25.5% per year.
    I mean what about bad/late fills, in reality I don’t see it as possible to do this in realtime, so do you look at this on a closing day basis?
    I just see so many back and forth over the line, so for me practically it would not be feasible to manage, and again I am surprised that the negative effect on performance isn’t bigger – even if you look theoretically…
    Also the regular model in itself has a uniquely good historic performance, in the 3 instances that the model missed a deep pullback, the safer model will buy back later than the regular model, that should hamper performance additionally, no? Obviously in other instances there were reslly no need for the safer version?
    Also, and that is rightly likely outside of your scope, but both comissions and taxes will have a negative additional effect.
    Thanks!

    • The “25.5%” assumes no bad/late fills and no transactions fees. Of course in the past transaction fees would be higher, but now transaction fees are negligble. Might hurt performance by a few basis points each year.

Leave a Comment