When Models Work

“All models are wrong. Some models are useful.”

This post is the second part of a white paper that I wrote for Economic Pulse Newsletter subscribers a few months ago. If you’re interested in learning more about the newsletter see here. Also, for more info on all kinds of TAA models I highly recommend Allocate Smartly.

This is one of our main tenets at the Economic Pulse Newsletter. We know our base TAA model, the SPY-COMP system, will fail, as I showed in my last post. SPY-COMP doesn’t fail often, not too long, and not catastrophically. The model’s errors are few, small, and short lived. But we also know the model works. In this post, I will talk about the model’s successes. Basically, this is a discussion about the effectiveness of absolute momentum before even adding relative momentum as an additional layer. Again, this is something I think should be discussed more with respect to TAA strategies in general.

Distribution of Returns

Table 1 is the same table from the post on When Models Fail. It shows the distribution of monthly returns for various TAA models (SPY-COMP, DM-COMP, GEM) vs buy and hold (SPY, 60/40, and SPY/VEA 50-50).

Table 1: Distribution of Monthly Returns for Various TAA Models vs Buy and Hold

Looking at a positive view of these figures, we see the TAA models deliver positive monthly returns about 70% of the time and that the average monthly return is about 50% higher than the buy and hold equivalent. The models also cut down on significant negative monthly returns (the sum of all returns less than -2.5%, the sum of the last three rows in the table above) by 50% in the case of SPY-COMP vs SPY and by 25% in the case of DM-COMP vs the SPY/VEA benchmark.

Let’s dive a little deeper and look at how the model provides higher returns and lower risk.

System Successes

The biggest success of the models is very simple – keeping you invested in equities when the economic and market conditions are favorable to stocks and invested in bonds when economic and market conditions are not favorable to equities. Overall, since 1971, that translates into being invested in 100% in equities about 70% of the time and invested in bonds about 30% of the time. Given all the noise in financial media and the quantity of bad information that investors need to weed through, this is no small feat.

As we saw in the “When Models Fail” post, the model makes mistakes, but the error rate is very low and the size of the mistakes is small. Now let’s look at the amount and size of the successes.

First, let’s define what success means. I’ll start with the most obvious. Since equities decline the most during recessions, one of the premises of our system is to avoid recessions. Let’s look at how well it catches recessions. Since 1971, the SPY-COMP system has caught every recession, before a significant market downturn; i.e. no false negatives. Let’s look at each of those recessions and how the SPY-COMP investment model performed vs the SP500 during the recessionary period.

• 1973 recession: 38% ahead of SPY
• 1980 recession: 14% ahead of SPY
• 1981 recession: 41% ahead of SPY
• 1990 recession: 7% ahead of SPY
• 2000 recession: 94% ahead of SPY
• 2007 recession: 88% ahead of SPY

The outperformance of the model in these 6 recessionary periods is responsible for a big part of the model’s outperformance over the entire period. See the last post for a performance table.

Now, let’s look at all monthly market drawdowns of more than 10% outside the recessionary periods. How did SPY-COMP do relative to this market-based measure of success? Since 1971, there have been 13 periods, of varying lengths, with market drawdowns in excess of 10%. Six of those were the recessions. As noted above, the SPY-COMP system caught everyone. Of the other 7 periods only 3 of them really made any difference (>5%) to the performance of the model vs the SP500. SPY-COMP also performed well in those 3 instances.

• 1973 pullback: 6.7% ahead of SPY
• 1987 pullback: 20% ahead of SPY
• 2010 pullback: 5.9% ahead of SPY

But our system also triggered 18 times outside of those periods. Let’s take a look at the successes among those false positives. In 6 instances, it was ahead of the SP500. Of the 6, only 2 were significant with difference of more than 5%.

• 1971 false trigger: 8.15% ahead
• 2007 false trigger: 6.00% ahead

The other 4 successes were positive but less than 5% vs a buy and hold of the SP500.

Summary

The 11 periods of success that I outlined above account for almost all of the outperformance and lower drawdowns of the SPY-COMP system over buy-and-hold. As the model was designed, and as we should expect with any TAA model, most the significant outperformance occurs during recessionary periods and period of serious financial stress. And this is before applying any relative momentum to various asset classes to further enhance returns.

The impact of this outperformance is at least four-fold:

1. You obviously have more money at the end of the recessionary period which is especially important if you’re in the portfolio withdrawal phase of your life.
2. Just as important, you have more money remaining in your account to compound in the future.
3. You’re much more likely to stick with a system that isn’t losing a ton of money in any given period.
4. Your ability to sleep at night is way higher with an automatic system that keeps your investments out of harm’s way.

Simplicity, transparency, and performance that enables peace of mind. That pretty much sums up TAA investment strategies.