There is a recent post by Eric Nelson that uses AQR to make one really good point and then to, kind of oddly, pointedly not ask the next logical question. Here I ask and answer that question.
Eric reminds us that what the average investor in a fund gets is different from the return of that fund. That’s true and important. If they get in and out with poor timing they can sabotage themselves. Next, his thesis is that investors in “alternative” funds, funds meant to be diversifying to traditional investments and often using more hedged strategies, are worse timers. He uses Dimensional Fund Advisors (DFA) as the example of traditional investing 1 1 Close I only mean “traditional” here as long-only equity funds, with a beta near 1.0; not that the method of investing is particularly traditional (for the record, I think DFA’s method of investing is better than what might be conveyed by “traditional”!). and us (AQR, in case you’ve forgotten!) as the example of alternative investing. It’s nice to be an icon I guess… But, then it all goes south for us in Eric’s piece. He shows that investors hurt themselves in the example alternative funds (by getting in/out at the wrong times and thus having their aggregate dollar weighted experience underperform the actual fund) but not in the traditional funds.
Well, so far we don’t really disagree. His calculations seem about right 2 2 Close We believe his numbers come straight from Morningstar. So we dug into their methodology a bit. There are certain choices in the methodology that can impact the results. For example, they use monthly instead of more frequent data (in this piece we do as well but, unlike for risk and return, for flows the right answer is unambiguously daily), and in doing so, they assume all cash flows occur at the end of the month irrespective of when they actually occurred. Our estimates, which try to be a bit more precise, show that investors were a little better at timing both the AQR and DFA funds. But, as much of my point is that other peoples’ timing ability is not nearly as important as whether you think an investment is good or not, and whether you can control your own timing, I’m going to let it go. I’m obviously growing as a person. , and we spend a lot of time ourselves trying to talk people out of too much timing. We don’t particularly like it for styles, for markets, and we really hate it when it’s done at what we call momentum investing at a value time horizon (peeve #3). So, we probably agree with anyone advising investors to time less, particularly if it’s to avoid timing like a medium- to long-term return chaser, and even more particularly when it’s about timing us!
So, what’s the problem? Well, the problem is Eric stops there. He somehow stifles the curiosity we know he must have but doesn’t ask “so, what if someone didn’t time these funds?” Presumably that’s something that should interest a long-term, strategic buy-and-hold investor like Eric! I mean he had to have all the data loaded and ready to go. It just seems like a really obvious next question. Well, I won’t leave you hanging, I’m going to ask and answer it here.
Let me first say that, obviously, three years of monthly data 3 3 Close All calculations in this piece use the thirty-six monthly returns from 2014-2016. You get, of course, slightly different answers, but no change in spirit, if use other frequencies (e.g., daily). are not close to enough to draw any serious conclusions. 4 4 Close I’m referring here to making sure not to draw serious conclusions about return and risk, but it’s also true for Eric’s main point. In particular, looking at three years of data for a set of relatively new and rapidly growing funds isn’t exactly dispositive on the question of how well or poorly investors implicitly will time, or not time, these funds going forward. Basically, although I don’t make this the focus of my argument, three years of data on relatively new and growing funds is not enough for us to willingly consign our investors to the bin of “bad timers” just yet! Nor is the particular set of funds that Eric chose (and I will stick with them just to avoid any possibility of cherry picking) necessarily the exact set of funds that AQR, DFA, or Eric would literally recommend for an overall asset allocation. But, come on, once you’re picking on these funds you’ve got to look at how they actually performed, and not just at whether investors helped or hurt themselves with their timing, don’t you?
I will start out using just equal weighted averages of the funds selected by Eric as the AQR and DFA composites. Again, that’s also not what anyone would necessarily recommend (not these particular funds nor equal weighing). 5 5 Close In particular, equal weight is a bit odd for funds like ours that explicitly target different volatilities. I’ve also looked at weighting inversely proportionately to our targeted risk, essentially an equal weighting in risk not dollar space, and found results a bit better, but not substantially different, from what I report here. But I’m trying to work with Eric’s choices here in as neutral a way as possible.
Looking at it this way the annualized average return on the DFA composite was 3.9% (the 50% in two USA funds were much stronger than the 50% in international and emerging) with a realized volatility of 11.5% and a worst drawdown of -15.5%. Its Sharpe ratio came in at 0.33. Doing the same exercise with the AQR composite you get an annualized average return of 3.2% (darn, we lose by 70 bps!), a realized volatility of only 3.5%, a worst drawdown of only -1.7%, and a Sharpe ratio near triple that of the traditional composite at 0.90. 6 6 Close By-the-way, there’s a separate message here about diversification within alternatives. When you build a portfolio of lowly correlated strategies, you don’t expect each and every one to do well in any given short sample period (the whole won’t do well in every short period either but hopefully it’s more consistent). Some will zig, while others zag. Yet, you generally expect the combination to have better properties than funds viewed individually. That is the beauty of diversification. As a great example of this, the equal weighted AQR composite, as reported, held up well (particularly in an environment with cash yielding nothing) in terms of return and importantly did so with very low volatility and a very mild drawdown. Individual fund results were, of course, much more variable but, due to fairly low correlations among them, the total “alternative” portfolio was very well behaved.
Let’s have more fun (and be a bit more realistic). Imagine you put ½ your money into the DFA composite and ½ into the AQR composite, again equally weighting within both. Now you get a 0.55 Sharpe (vs. 0.33 for only traditional) with a -7.1% worst drawdown (vs. -15.5% for only traditional). Of course, you get just about the same return (as the DFA and AQR composites were only 70 bps apart). 7 7 Close If you want to get really silly and just ex post optimize over this way too short three year period, looking for the best possible Sharpe ratio, you would end up with 96% into the AQR composite and 4% into the DFA composite (both composites are still equally weighted within across all funds). This combination would have realized 3.3% annualized, with 3.5% volatility, and a maximum drawdown of -1.9% over the sample period. (Note as a robustness check the weights on each composite are qualitatively similar if you take out any one of the funds from the population Eric chose for each composite.) I’d sign up for that improvement over 100% traditional every time. 8 8 Close Of course I won’t necessarily get it. In fact, I hope this is obvious, but I would never highlight three years of decent returns and great diversification benefits, except to respond to a piece that used the same short period to wrongly recommend against these funds! It happened because the AQR composite kept up in return, was actually much lower volatility, and only 0.25 correlated with the DFA composite. Of course, delivering those characteristics is much of the point of alternatives.
Short periods shouldn’t be used for too much in finance (sorry to remind everyone as we drone on about this again, but three years only feels like a long time — it’s really a very short period). But, what’s really odd is to use a short period with great confidence for one thing while completely ignoring it for another. What’s even odder is to recommend against something based on average suboptimal behavior that anyone individually can control themselves. For instance, if the DFA funds investors exhibited suboptimal timing behavior (they don’t over this period) would Eric abandon them because of the behavior of others? Now, if he wants to rewrite his piece with a new theme of “too many investors jump in and out of alternatives too much; what they should do is find ones they believe in, add them to a traditional portfolio and then stick to it” we’d all be in harmony. Instead, he says “investors time these good things poorly so you should just avoid them” and not simply “stop timing these good 9 9 Close Yeah, I added the word “good.” Got me! Of course he, or anyone else, is free to argue these aren’t good things at all, and we will, not surprisingly for our own funds, take the other side. But much of the point of this piece is really that this is a completely separate argument. things poorly.” We just think that’s kind of weird!
I’ll end with the title question. Eric, presumably, again, you wouldn’t get in and out capriciously like you document for others. Given that, and given the three years of data you’re drawing your own conclusions from, how much should we pencil you in for? :)