Beware Model Portfolios With Unrepeatable Results

Morningstar did a quick writeup on model portfolios. The article wasn't too insightful but there was an example of a model portfolio and then an example of how to customize that same model. I didn't see where the original model came from or where the customized version came from either but there's something to learn from it.

All I tried to do was simplify the portfolio, not do anything to improve it. 

The original model slightly outperformed my simplified version. The customized version was much better than my simplified version but they all lagged VBAIX and all the standard deviations were in the same neighborhood. This sort of modeling is not something I have ever done. Should a model portfolio have any sort of differentiation? They really don't but I am not sure what the right answer is but I might know the right question. 

If you are an advisor and you want to go down this road, I think you need to decide ahead of time whether differentiation is important to you for your clients. It may not be important to you, I'm just saying to figure that out ahead of time. If you are a do-it-yourselfer but are presented some sort of model portfolio, great if you have an opinion about whether you do or don't want differentiation. Then ask whoever is presenting the model why they don't do it the other way. So with the above model, I would ask why there is no differentiation. If I were in the model business, there would be differentiation and so I am saying to ask me why I don't look more like plain vanilla 60/40. 

Also, when looking at models I would suggest being on the lookout for any sort of result that might not be repeatable. We talk about that all the time. If you want to replicate private equity with one of the operating companies like Blackstone or KKR, that might work but the past results for those two stocks are so good that modeling them makes for what I would say is an unreliable backtest. Similarly, any sort or modeling with Bitcoin probably gives an unreliable result. According to Yahoo, in the last ten years Bitcoin is up 39000%. Even if it went to $5 million, that wouldn't be anywhere close to 39000%. What I think you can glean from models that include unrepeatable results are attributes related to volatility and correlation. Bitcoin is very volatile and the correlation to equities is all over the place being highly correlated at times, negatively correlated at other times as well as uncorrelated too. 

One of the above started out 2% Bitcoin, 98% VBAIX ten years ago, no rebalancing and the other is 100% VBAIX. Guess which is which. That result is not repeatable. Bitcoin could be a great hold from here, or not, but the past result is not repeatable. 

The other day, I mentioned the brand new Locorr Strategic Allocation Fund (LSAIX) which is 50% equities/50% managed futures. Locorr sent over a powerpoint presentation for the fund that included the following. 

Going back to 1980 isn't too helpful because the the strategies they are talking about weren't accessible to retail sized investors via funds but the idea is similar to what we've seen before regarding alternatives. 

I put no effort into choosing the alts other than diversifying fund providers. Portfolio 1 isn't the best performer all the time, it doesn't work that way, but it is hands down superior to plain old 60/40 with the S&P 500 and Aggregate Bonds.

The information, analysis and opinions expressed herein reflect our judgment and opinions as of the date of writing and are subject to change at any time without notice. They are not intended to constitute legal, tax, securities or investment advice or a recommended course of action in any given situation.