Supplement to: This simple trend-following model has crushed buy and hold

3 comments

This study is intended as a supplement to the NAS Trading article, “This long term trend model has dominated buy and hold” posted at NAS Trading and on Seeking Alpha.

In this article, we evaluated a moving average system using monthly S&P500 index data going back 60+ years. One weakness of the study is that we used an index rather than an investable security. Lets look at the results of the same system, but using SPDR S&P 500 ETF (SPY) data from 1994 to present (which is all we can use given this securities more limited history). Here are the results using a 10MA:

As you can see (and similar to the 10 month lookback system) this approach as historically dominated buy and hold. One important concept in trading system development is to focus on concepts rather than specific rule-sets. Ideally, you want to be able to test a basic idea using a wide variety of different specific trading rules and parameters and find that they all work in a similar fashion.

Lets look at the results using an 8 – 16 month moving average, stepping two months at a time:

All of the parameters we have tested do reasonably well, once again much like our original “lookback” approach.

If you have made it this far, I think you would benefit from signing up for the NAS Trading free email list. You will receive supplemental studies and have the opportunity to stay informed about developments at the NAS Trading Subscription Service, which will have high value, specific trading and investing ideas that are available exclusively to members. Sign up for the email list by clicking here.

Good investing.

Solar689-1

This doesn’t work well when a walk-forward test is applied. This means that a period of 10 is curve fitted.
http://twitter.com/RobertBaillieul Robert Baillieul

Love this strategy. I use a similar system but apply it to a basket of markets (i.e. domestic equities, international equities, long term bonds, commodities and REITs).

You might be interested in reading Mebane Faber’s paper (if you haven’t already), where he backtests this system on a portfolio of different markets.
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=962461
Daniel

Thanks for your analysis. Are you able to share your
calculations i.e. spreadsheet?

A few questions, do you consider taxes (long term and short
term) on dividends, interest and commissions on buying/selling?

I have invested in Trend following programs i.e. Dunn
Capital, yet I have found fidelity claiming that there is no sound data
supporting trend following. Fidelity appear to be very data driven, given your
analysis why would they reach such a different conclusion?

Managed futures index is down 3% (according to Attain) for
2012. If I took your 10 month lookback for just 2012 it would have returned a +
8% (based on a buy in feb 2012 excl dividends).

Is there any ETF which replicates your 10 month lookback?

Have you tried using the 10 Year Shiller trailing PE as a
signal? I tried 15 for buy and 30 for sell. I don’t have any program to
calculate this but my back of the envelope calculation was this returned higher
than the 10 month lookback.

Lastly, do you know any program available to use for
simulating the backtesting which include peak to valley, drawdown period etc…?

Thanks.

Daniel