Timing Luck In Investment Strategies

During their latest episode of the VALUE: After Hours Podcast, Hoffstein, Taylor, and Carlisle discussed Timing Luck In Investment Strategies. Here’s an excerpt from the episode:

Tobias: One of the things I learned from you, Corey, that I was very grateful for was that the timing luck idea, which is one of the problems with any back test. And then no matter how careful you are, whenever you’re balancing– In quantitative value, we did a mid-year rebalance using year end data, so it was six months of data, six months lag. So, most of that data should have been in the public domain by the time we traded. But it’s very sensitive to the date that you roll. If you roll close to March 2009, if that’s in your close to the bottom in March 2009, you get vastly better performance than if you roll in September 2009.

Corey: That’s Research Affiliates’ Immaculate Rebalance. I would argue the entire history of that firm changes if they didn’t rebalance in March. You’ve heard that story, right? I must have told that story before.

Tobias: Tell it again.

Corey: Oh.

Jake: Tell us again.

Corey: Okay. So, Research Affiliates. If you don’t know who Research Affiliates is, they are a massive asset management firm based out of Southern California, led by Rob Arnott, or historically led by Rob Arnott. I think he’s moved on. And there’s a woman who’s CEO there now. He might be chairman. Long story short, he publishes this methodology in 2005 called Fundamental Indexing. His argument is that companies in your index should not be weighted relative to market cap. They should be weighted relative to their fundamentals. He publishes this whole index methodology and says, he’s revolutionizing indices. There’s a very funny debate between him and Cliff Asness at this time, because what Cliff points out is all this is a value tilt, [Jake laughs] and Rob refuses to acknowledge it and refuse to acknowledge it for like a decade.

Tobias: You can’t get a trademark on value tilt. You got to [crosstalk] indexes.

Corey: Yeah. Anyway, so he publishes this concept. It’s basically a glorified value tilt. And in the index, he arbitrarily chooses to rebalance every March, and he rebalances once a year. And so, coming out of 2009, he rebalanced March 2009 and outperformed the benchmark by a ridiculous amount. You can actually look at this. There’s an ETF that was around at the time tracking his index. I believe the ticker is PRF. I believe it’s an Invesco ETF, if I’m not mistaken, RAFI 1000, and you will see the relative performance to the S&P 500 is ridiculous. I think it’s like an extra 1,000, 1,500 basis points in 2009, and since then has had the usual value struggles.

Now, what happened in 2010 is that some researchers from Robeco in the Netherlands, David Blitz and Pim van Vliet, and I’m blanking on the third gentleman, my apologies, wrote a paper that said, “Well, that’s interesting. But what happens if you take that same index methodology and instead of rebalancing in March, you’re rebalanced every June or you rebalanced every December or September?” And so, they actually ran those counterfactuals. What they found was that, if instead of rebalancing in June, Rob had Arbitrarily chosen to rebalance in September, he actually would have underperformed the S&P 500 in 2009.

Jake: Wow.

Corey: So, you go from outperforming by 10 percentage points plus to underperforming the market using an identical methodology, just rebalanced at a different point in the year. Now, to RAFI’s credit, and they were using Footsie, I believe, as their index publisher at the time, Footsie-RAFI worked together and implemented what the paper suggested, which was to do these overlapping portfolios, which basically– [crosstalk]

Jake: Take some of the timing risk out.

Corey: Yeah. To think of it in a simple way, it’s to say, “Look, you want to rebalance once a year?” That’s fine. Pretend like you’ve got four managers. One of them will rebalance in March, one will do June, one will do September, one will do December, and give a quarter of your money to each and then keep rebalancing across each of them. What that’ll do is that’ll get rid of the timing luck. This has become a, I wouldn’t even call it a passion of mine anymore. I’d probably call it an obsession, like, to the point it’s a little deranged-

[laughter]

Corey: how much I care about–

Jake: Religion.

Corey: Yeah, it’s a religion for me. I’m starting a cult around. So, RAFI goes on. And this lives on forever in their track record. It’s not like they restated their track record to say, “Actually, we [crosstalk] shouldn’t have outperformed by that much. It should have been less.” And of course, they raised an absurd amount of money and that’s history. And you go and you say, “Okay, an active manager got very lucky.” But I want to point out the benchmarks that everyone uses– Like, Toby, you get benchmarked to the Russell 1000 value, I’m sure, or the Russell 2000 value. Those rebalance once a year. Who’s to say that that index didn’t get absurdly lucky when it rebalanced? It’s a horrible metric, and yet, we use it as an entire industry. So, I’m going to stop because I will literally just keep talking for the rest of the hour about this.

Tobias: What’s the solution to it? You can either do the four overlapping portfolios or just rebalance more often.

Corey: Those are different solutions. So, what I would argue is, when you have an alpha signal, so call it value, you generally have an idea as to what the decay cycle of that alpha signal is. What do I mean by that? Momentum. When you trade momentum, momentum signals are very fast. It’s a high turnover signal. So, you need to rebalance that very frequently versus value tends to be reasonably slow. When you take T costs into account, transaction and trading costs, you probably don’t want to refresh your value portfolio every day necessarily. You wouldn’t want to trade it every day. You might only want to update it every 6 months or every 12 months to avoid noise trades at the fringes.

So, what you first want to do is figure out, what is the right rebalance frequency to maximize the horizon over which your alpha has predictive efficacy? You can still overcome T costs. And then once that is set– So, say you think the optimal rebalance frequency for value is every six months. What you would then do is say, “Okay, now how do I break up that rebalance schedule?” So, instead of just January and June, I might be also February and July and March and August. You keep breaking it into as many sub-schedules as you can that are equally spaced apart. Ideally, you do that as much as possible. Quants like AQR, for example, will take that and do it daily. [crosstalk]

Tobias: They’re just trading some incremental– [crosstalk]

Corey: They are trading a very small part of the portfolio every day. So, you could argue, if you’re re balancing once every six months, you basically want to rebalance 250 seconds of your portfolio every day. It’s a small amount. It’s probably not feasible for most people at home, but it would be the equivalent of saying like, if you’re doing this at home and you think it’s every six months, every month, you should turn over one-sixth of your portfolio, like rebalance one-sixth of it, keep the other five-sixth constant.

Tobias: That’s if you’re taking a quantitative approach, because there are some people who listen to this who are just value guys who are like, “Why don’t I just sell it when it gets overvalued?” Which is also know how to– [crosstalk]

Corey: You can do that as well, right? Absolutely. So, the way I would say that is like, when I go to turn over that part of my portfolio, it’s not like I’m just selling everything indiscriminately. I’m looking at what’s– There’s going to be things that stay there. It’s just that is the proportion of the portfolio that I’m looking at. Those holdings have been there for six months and it’s time for me to refresh whether those holdings should still be there or whether I should get rid of them. Now, to your point, Toby, I could buy something and a month later it could be massively overvalued. Do I just hold it for another six months? That’s where the nuance and subtlety comes in.

Tobias: Whatever you do will be the wrong thing to have done.

Corey: 100%.

Tobias: If you sell it, it goes on for a multidecade run.

Corey: That’s right.

Tobias: If you don’t sell it, it returns to where it was.

Jake: [crosstalk] [laughs]

Corey: Spot on.

You can find out more about the VALUE: After Hours Podcast here – VALUE: After Hours Podcast. You can also listen to the podcast on your favorite podcast platforms here:

Apple Podcasts