Economic Indicators , Economy


Update: I updated the charts and some of the performance results on March 28, 2017. I found and corrected an error in the performance calculations. Conclusions remain the same.

It’s finally time to start turning all the economic indicator stuff I’ve been posting on into something useful for investors. In this post I’ll introduce the SPY-COMP indicator and how it works as tool for entering and exiting investments.

The mechanics of the SPY-COMP system are similar to the SPY-UI system I’ve posted on previously. The only difference is that the new system uses a composite of the top 6 economic indicators instead of only using a single indicator, the unemployment rate. There are many ways to build a composite but we keep it simple. Each indicator covers a different aspect of the economy so we have the composite indicator trigger whenever any one of the 6 indicators triggers. Then, just like the SPY-UI system, the composite triggers the monitoring of the SPY crossing it’s 200 day SMA to signal a buy or a sell signal for investments. Simple. Just like the SPY-UI system but using the COMP indicator.

Now, let’s take a look at some results. Let’s compare buying and holding the SPY to the SPY-UI system and the SPY-COMP system. When the SPY-UI and the SPY-COMP system are out of the market the funds are held in cash (earning 0%) in these simulations. The addition of any kind of bonds only improves the systems’ results. First, we’ll compare drawdowns from 1973 through 2016.

Both the SPY-UI and the SPY-COMP system reduce drawdowns significantly vs the SPY alone. Max monthly drawdown for the SPY during this period was 49% vs 26% for SPY-UI and 16% for SPY-COMP. The biggest difference between SPY-UI and SPY-COMP is that SPY-COMP captures some of the large non-recessionary drawdowns in the SPY. Now, let’s look at total returns. The chart below is a log chart of total returns for each system from 1973 through 2016.

Total annualized returns over this period for the SPY are approximately 10.2%. For SPY-UI total annualized returns are 12.4% and for the SPY-COMP system they are 12.8%. Again, these returns do not include any bond returns for the SPY-UI and SPY-COMP system while they are out of the market. For example, the SPY-COMP system was out of the market from Nov 2007 through May 2009. During that time an investment in intermediate government bonds (IEF) would have returned 14.5%. Over the 1973 to 2016 time period the SPY-UI and the SPY-COMP system were out of the market 16% and 22% of the time respectively.

Note: I say returns are approximate because I had to build my own monthly total return series from the Shiller SP500 data . I assumed dividends for the index were paid out equally over 12 months for all years which is an approximation.

One way to look at these systems is that they are trying to distinguish between volatility to pay attention to and volatility to ignore. And these systems do ignore some significant volatility. Easy for an algorithm. Not so easy for humans. For example, let’s look at the last two years of the SPY-COMP system.

There was no significant weakness in the economic indicators in 2015 and 2016 but there was still some significant volatility with the SPY crossing below it’s 200 day SMA a couple of times and having a max monthly drawdown of about 8%. The system was saying this was volatility to ignore. But of course, it’s never that easy. If you paid attention to daily fluctuations and the financial news the experience was a lot more painful. Max daily drawdown from July 2016 to Feb 2016 was just over 13%. See below.

In this case the SPY-COMP system was signaling the correct information and investors were rewarded for being patient.

That just about does it. The SPY-COMP system improves somewhat on the already powerful SPY-UI system in terms of returns and drawdowns. Now that we have the SPY-COMP system in place we can use it in TAA and quant portfolios to further improve performance in those systems. More on that in future posts. The SPY-COMP system is now the benchmark system I will use going forward to compare with other investment systems. I’ll provide updates to the SPY-COMP system performance every so often. I’ll track the COMP in the top 6 indicator update that I’m doing during the middle of each month.


Tagged , , , , ,

34 thoughts on “ Building an investing system with the top economic indicators

  1. Great work Paul.

    I understand that if the SPY-COMP activates then you use the 200 day moving average to decide whether to be in the market or not. My question is : How are you specifically using the 200 day moving average … Are you checking against the 200 day average on a monthly basis (or a more frequent basis – i.e daily). Thank you!

    1. The results posted are for checking on a monthly basis, end of month. Once COMP triggers you can check more often if you like. Works just as well but with a few more false triggers.


      1. Paul, I might be missing something and correct me if I’m wrong but if the signals are taken on a monthly basis, the 1987 crash would have been unavoidable…By the time we receive the economic signal and wait for the monthly close the crash had already happened.

        1. Good point. COMP (the econ signal part) triggered in Sep 1987. So when the SP500 closed below it’s 200 day SMA you would have been out. That happened just about mid month. So, you’re right if you waited until month end you would have experienced Black Monday. On a total return basis though, including dividends, you would only have been down about 14.5% from the peak by the end of the month of Oct (daily drawdowns were much worse than the monthly drawdowns. After Oct the market went down a bit further which you would have avoided. Extremely painful. After the econ signal triggered you would have to be acting on the weekly/daily signal to avoid the big pain. Friday Oct 16 1987, the weekend before black Monday, the SP500 closed below it’s 200 day SMA.


          1. After the econ signal is triggered how often do you monitor your 200 day SMA or other signals?

            Thanks for all your work by the way, very useful information!


          2. The results I’ve posted look at all the signals on a monthly basis. In practice, I look at the econ signals on a monthly basis. Once the econ signal triggers I look at the SPY vs it’s 200 day SMA on a weekly basis. It produces slightly better results.


  2. This is really cool and important. I invest using GEM, and I got whipsawed in late 2015 and early 2016, and it was messy and stressful for me. Your SPY-UI and SPY-COMP will allow me to turn off GEM and avoid some whipsaws. Also, I have some cash, I’m waiting for a dip to push it into the market, and if SPY-UI and SPY-COMP are benign when I find the dip I want, I’ll invest with more confidence. Excellent.

  3. Thanks for thinking outside the box with this new economic indicator. The results are great. On average, how many trades per year should we expect to make using the SPY-COMP system?


    1. Great question Chad. In the 1973 to 2016 period (44 years), there were 43 trades. So about 1 per year. Low cost. Low slippage.


  4. Very insightful as usual Paul!

    I can see the immediate application for the quant systems, but looking forward to see how you use this with the TAA portfolios.

    1. Thanks Tony. Basically you replace the absolute momentum trigger in TAA portfolios with SPY-COMP and keep the relative moment ranking.


      1. Thanks for that. Would be interesting to see a backtest with this compared to the baseline system. Just thinking that there will be times when TAA is calling for an allocation in bonds, commodities, or real estate and it seems that the SPY-COMP indicator is best suited for equities. I am guessing you have already found that it works well for TAA, just would be interesting to see the data.

  5. Paul, as usual, great stuff. Would it possible to see when the COMP triggers monitoring of the SPY 200MA? My guess is… it triggers monitoring more often than UER alone, but the index stays above the MA outside of the recessionary periods (for the most part?). Another way to look at it… how often did SPY-COMP get you out of the market when a recession did not occur (how many false positives)? Overall the positive results shown here are great and speak for themselves, this minutiae is just a curiosity of mine.

    1. Since what the system is trying to do is avoid financial losses, just because it triggers outside a recession doesn’t mean it’s a false positive. That is the biggest difference between the COMP and UI system. The COMP system avoids some significant no recessionary market drop – 1987 being the biggest.


  6. Where on this site does it explain the SPY-COMP indicator, how it is configured, and used?

    1. In this post. The COMP is a composite of the top 6 econ indicators linked to in the post. When any one of them triggers the COMP signal triggers. Then you wait for confirmation from the SPY signal to take action.


  7. I can’t quite tell from the graphical data but I’m wondering: are all 6 indicators necessary for the SPY-COMP system? In other words, has each of the 6 triggered at least one alert in the past that was independent of the other 5?


    [Even if not independent, I guess each of the 6 would have value if it triggered at least one alert earlier than the other 5 did.]

    1. Hey Tom, great question. Well, 5 of the indicators add value. The yield curve is the only one that is of questionable value. UI and Leverage are the most important. But Permits, IP, and RFSFS are that far behind. I left yield curve in the COMP because it’s discussed everywhere, it’s in all the leading indicators, etc…and it doesn’t detract from the results either.

      In a recession once one of the 5 indicators (leaving out the yield curve) goes, they all tend to go. But they go in different orders depending on the cause of the recession. Outside of a recession, leverage is the best indicator.


    1. Haven’t done Sharpe’s yet. Without modeling the bond component of the strategy they’re not as useful.


  8. Great post Paul!
    Is there an equivalent model that can be built for other regions, say maybe Europe? It’d be great to have the same quality model applied to a diversified portfolio beyond just US stocks… if there’s similar data that can be used?
    Thank you again,

    1. Don’t know. But historically, using US recessions also impact global markets significantly.


  9. Hi Paul
    New to investing.
    How can I follow your Spy-comp strategy?.
    Is portfolio 123 the place to look at?.
    i see that one of the indic is under 0 so if Spy goes under 200 SMA you go out?.
    Kind Regards

  10. Hi !

    I can’t find out how you move in to the market again after the recession. Is it when one of the signals trigger or must they all be green ?

  11. Hi Paul,

    Excellent article! Question: why don’t you use SPY 12mo Abs mom in SPY-UI or SPY-Comp to get more return than the SPY 200 day SMA?


    1. Because it doesn’t work as well. Although the differences aren’t huge.


  12. Paul, I’m heartened to hear the differences between 200 MA and AbsMom are small. I use AbsMom based on what Gary Antonacci and Wes Gray write, and as far as I’m concerned, less trading is better, every time I make a big trade it puts me on edge.

    1. The biggest difference I’ve noted is that using the 200 day SMA with the econ triggers gets back into the market quicker once the econ all clear is given. SMAs weight recent prices more than abs momentum does.


Comments are closed.