I read it.
Really? Because I read it, and it's not the same crap I've seen anywhere before, and I'm not sure how "survivorship bias" has anything to do with it.
I was actually very excited to read it, because it finally investigates the last refuge of the investing scoundrel. In the olden days, everyone just assumed that skilled investors could easily beat the market. Then data reveals that mutual funds can't do it, so the investing scoundrel retreats to "ohhh, yeah, they can't do it because they get too big and too public, but privately people can still do it. You're looking at the wrong data". Then the data reveals that hedge funds can't do it, so the investing scoundrel retreats to "ohhh, yeah, they can't do it with all their fat bonuses and trading costs, but individual buy-and-hold investors can still do it. You're looking at the wrong data. (and no one better come up with a way to analyze a group of individual buy-and-hold investors, because I have nowhere left to retreat to!)"
So hey, this was actually a pretty reasonable data-set to use to analyze that last refuge, in which many of the individual stock-picking investors at this forum are holed up. And damn, the paper revealed some pretty impressive results. Following the picks made by the people at this Value Investor Club didn't just result in like a 0.9% outperformance or underperformance that we're used to seeing in studies like this. They found numbers like a 10%/year outperformance. So even if there is some "survivorship bias" or other flaw to the study that cuts the actual outperformance in half, it still seems like a number that's unlikely to completely disappear.
The authors definitely seem biased (they work for an investment-management company), and there were some minor errors that left me a little uncomfortable (randomly switching voice from "we" to "I" within the paper; derisively pointing out an investor who suggested an investment in Lehman Bros after they'd filed for bankruptcy, even though, in fact Lehman had not yet filed by the time of his suggestion; and no curiosity about the fact that the outperformance seemed to slow/stop in the 3rd year after the idea was presented, and 3 years was as long as they looked). But there's nothing that seems like a giant error to me.
Though another minor thing that bothered me was that, in 2008, they were publishing 3-year performance of data from 2008. Huh? So I checked the link to the SSRN database included in the intro to the PDF to see if they just had the dates wrong in this draft or something.
Hmm, except the paper is no longer hosted at SSRN. I did a search for the primary author, and came up this,
Do Fund Managers Identify and Share Profitable Ideas?, a paper by the same authors and a similar title, published a year later (and last revised 3 years after that, in 2012).
Curious.
It uses the same data source (valueinvestorclub.com "ideas"), but uses a different analysis method. This time, the result is FAR different, and much more in that "0.9%" sort of range: "We find evidence of stock-picking skill among VIC members....However, the abnormal returns are restricted to small securities. For example, the average one-year value-weight calendar-time portfolio alpha estimate is 0.73% for buy recommendations....in the smallest quintile of firms."
So we went from pretty strong evidence of individual skill to, at best, the creation of yet another, tinier refuge to which the scoundrel can still retreat.
Given that they make no reference to their original paper in this new one, that leaves me with even less faith in their credibility, so I'm not terribly motivated to understand what was wrong with their initial methodology and why they switched to a new one. But given their admitted bias, I'm quite confident that this methodogy that produced less-impressive results is the more "correct" of the two.
JoJoK, as the proponent of these researchers, perhaps you'd like to explain?