How Long Should You Look Back To Forecast Forward?

In their latest episode of the VALUE: After Hours Podcast, Brewster, Taylor, and Carlisle discuss How Long Should You Look Back To Forecast Forward?. Here’s an excerpt from the episode:

Bill: Did we talk about a mailbag question about how many years should you look back and through cycle analysis? Did I bring this up last week or did we get cut off?

Tobias: No, that’s good one. Let’s do it.

Bill: Yeah.

Tobias: Like John Wayne.

Jake: [laughs]

Bill: Somebody at my alma mater is being taught to do– I think they’re doing models with three years of backwards financials and then forecasting for five, and he was curious to have us riff on that a bit.

Jake: Yeah. I think there’s probably like the Lindy effect is a reasonable assumption here to invoke? The longer that look back, the further that you can probably project off of that, knowing nothing else about it. So, I would probably not try to forecast further forward than I am looking backwards as a general rule.

Bill: I like that.

Jake: It’s like taking a beam off of or a board off of side of a building. You don’t put just the edge of it on the building and then walk away out on the edge of it.

Tobias: [laughs]

Bill: Yeah, that’s fair.

Jake: It’s better to have just a little bit of it sticking out in your projection and a lot of it behind in the historical.

Bill: Yeah.

Tobias: We tested that in quantitative value and found no advantage over, but that’s an aggregated–

Bill: Hold up, hold up, finish your sentence. You didn’t finish your sentence. You found no advantage over– [crosstalk]

Tobias: Looking back through the Compustat data to ’63 taking an average of EBIT, earnings, whatever we were looking at on every different metric that we tested, which was about, I think, five, or six, or seven, something like that, there was no advantage consistently to adding more years. You get just as good a result in your estimation on the TTM as you do at the eight-year average.

Jake: Really? Interesting.

Tobias: It sounds nuts. But the problem is that some stocks get these funny years, where they just get something– It’s like a factor more than they’ve ever earned. But it doesn’t help you to average that over a few years. It still makes it stand out. I don’t know if that information is particularly helpful. I prefer not to use TTM figures as well, but that’s the case. There’s no advantage. There’s no consistent advantage. There were some instances of five years being better, eight years being better, but it wasn’t consistent across the lot.

Bill: Sort of like my beloved Qurate. Rest in peace.

Jake: [laughs]

Bill: You wouldn’t want to take 2020 and say, “Okay, well, I could just spread this over 10 years, because that’s a once in a hopefully lifetime event,” right?

Tobias: All the bits that were spun out from that, is it still down after all of the–? [crosstalk]

Bill: Yeah. It’s a shame. They’re working on it. We’ll see.

Jake: You see that a lot with– some companies will have one big revenue contract that’s coming due and it’s a lot of question marks like, “Oh, God, are they going to get this re-signed or not?” Oftentimes, it’ll sell off big time relative to the current earnings that it’s showing.

Bill: Yeah.

Jake: That happens a fair number of times. I could see where that wouldn’t really look much probably different between TTM and five years or something.

Tobias: Keith Hillman says, “Throw out best and worst years per decade.” Winterization. I don’t mind something like that. it could work.

Jake: Yeah, it’s not bad.

Tobias: I like the– [crosstalk]

Bill: Throw out the best four and the worst four every eight years.

Jake: Yeah.

Bill: [laughs]

Jake: It’s like the Olympic scoring system for that, right?

Bill: Yeah.

Jake: You throw out the highest and lowest scores, the Russian judge.

Bill: I think the thing that’s tough when you’re looking back to the global financial crisis is, economically, it was slow, but it’s all up into the right. So, how do you actually [crosstalk]

Tobias: That’s the other problem that we have. There’s no recession in the data going back. So, you have to go back a lot further to see– That was one of the things I used to do, how does this thing fare when it went through a recession, or does it completely shut the bear or does it sort of struggle through and it’s okay? Because if it’s one of those ones that falls over, then any debt becomes lethal in that scenario. But it’s been so long now that, to get a good comp, you go back to 2008, starting to get those businesses that reinvest a little bit. They’re vastly different businesses now.

Bill: Got to do a money supply overlay to find out real recession. You dig now that we solved the businesses.

Tobias: Yeah, it looks like we have.

Jake: Permanent plateau.

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