In my last post I updated the official numbers for the IVY timing model. The original version of the model consists of a diversified portfolio of 5 asset classes. In the research paper Mebane Faber also looks at several extension to the 5 asset class timing model that improve returns over the basic model. Here I discuss the first extension, its results, and a few barriers to actually implementing it in individual portfolios.
The extension to the basic IVY timing model is simply the addition of more assets classes. More diversification is usually better, as long as the asset classes are not perfectly correlated. And a tilt to factors that have been proved to outperform over time (size, value, and momentum) also will probably help returns. Also, the model goes more global. This new model consists of 13 assets classes with a minimum of a 5% allocation to any one asset class. The portfolio is as follows.
As the table above shows the 20% US equity allocation in the original model is broken up into asset classes based on small cap size, value, and momentum. The foreign equity allocation of 20% increases exposure to emerging markets. The 20% bond allocation noew includes allocation to foreign government bonds. The commodities allocation adds a more specific gold allocation. Finally, the REIT allocation remains the same. Lets look at how this expanded model performs versus the original.
In the table above, B&H 5 is the IVY 5 asset buy and hold portfolio, GTAA 5 is the original IVY timing model, and B&H 13, GTAA 13 are the 13 asset class versions. The addition of more asset classes clearly improves performance. In fact its diversification that adds the most performance. Timing improves on that performance and more importantly reduces risk significantly – higher sharpe ratios and much lower drawdowns. And in case you’re wondering, yes, it does increase safe retirement withdrawal rates (more on that in a later post).
Fantastic. What’s not to like? Well, actually implementing the 13 asset class IVY timing model presents some issues. First and foremost is which ETFs do you use? We’re on our own here. Unfortunately, you can’t implement this model exactly in its current form. There are no ETFs that break out momentum into large and small cap (if anyone finds one please let me know). Also, remember the growth ETFs are not the same thing as momentum ETFs. But we can come pretty darn close to a solution. Below is my list of ETFs for the 13 asset class model. I use Vanguard ETFs when ever possible because they’re usually the best, have the lowest expenses, and are commission free at discount brokers. Thanks to reader, Kevin, for his list of the 13 ETFs.
The other issue with implementing the 13 asset class model is that you need to monitor the 200 day SMAs and the signals on your own. There is no website you can go to and simply get the signals like you can with the basic 5 asset class model. I and others publish an update every month. And remember these are dividend adjusted SMAs which are slightly different that SMAs based on closing prices. Its easy to do the calculations but its more work.
Is it worth the extra effort? That’s something only you can answer. Is that 1.56% extra return a year over the basic IVY worth it? Maybe. Personally, I haven’t decided that its worth it….yet. I only implement the basic IVY timing model in our IRAs. For now, I’d rather put in extra time into my active strategies like quantitative investing, option selling, picking great individual dividend stocks, etc…But I’ll probably change my mind at some point in the future.