Yesterday – the first Wednesday in February and thus the so-called National Signing Day – was the first day that high school seniors could sign letters of intent to accept an athletic scholarship to play Division I college football in the fall. It’s the culmination of a long recruiting process and crucial to the success of teams and coaches. It can get more than a bit ridiculous.
Some players announced their intentions using live animal props, or worse. One recruit picked Texas over Washington based on a coin flip. At least it wasn’t for the gear, officially anyway. And Snoop Dogg will be giving up his support for USC to cross-town rival UCLA because his son picked the Bruins, where he’ll join P. Diddy’s kid on the team. Cornerback Iman Marshall, a big-time USC signee, has a self-styled “commitment video” that’s particularly absurd.
But the coaches and the media outlets that cover college football recruiting (of which there are an astonishingly high number) take it all very seriously indeed. As the parent of a DI player (at Cal, see above), *I* took it very seriously.
These various publications generally rate high school players being recruited via a “star system” of from two to five stars, with five stars being reserved for top 50 players, four stars for the next 250 (numbers 51-300), three stars for the next 500, and two stars for players who are considered “mid-major” and thus not good enough for the top conferences and teams. Alabama’s current recruiting class is usually reputed to be the nation’s best, for the fifth straight year, averaging out to 4.08 stars. And while it’s not much ado about nothing, it’s much ado about a lot less than you’d think, and in a different way than you probably think.
Aaron Rodgers is the best quarterback on the planet. But he was a zero-star recruit out of high school with no scholarship offers. Peyton Manning was a five-star guy, but the top QB in his class was a guy named Josh Booty. J.J. Watt is the best defensive player in the NFL today but was only a two-star recruit (which goes a long ways toward explaining the tweet shown above). Drew Brees, Tony Romo and Philip Rivers were two-star guys too.
Only about 50 percent of five-star recruits, those alleged “can’t miss” prospects, ever get drafted by the NFL. A surprising number of them flame-out totally, sometimes before being meaningful contributors in college. On the other hand, only one five-star recruit (Cowboys OT Tyron Smith) was named to the NFL All-Pro first team this season. Indeed, the All-Pro team averaged out at just 2.88 stars and three first-teamers were zero-star recruits.
Not a single starter for either team in Sunday’s Super Bowl was a five-star recruit coming out of high school. Moreover, the Seahawks only featured four four-star recruits (most prominently Marshawn Lynch) while the Patriots had only three (including Rob Gronkowski and Vince Wilfork). Seattle QB Russell Wilson was only a two-star recruit, like nine other starters. Superstar DBs Richard Sherman, Earl Thomas, Kam Chancellor and Darrelle Revis were all three-stars, Sherman as a wide receiver. Stud Seattle defensive lineman Michael Bennett, perhaps the Super Bowl’s best overall performer, was a zero-star guy (and he wasn’t drafted either). The Seahawks’ starting lineup had an average rating of 2.4, while the Patriots’ came in at 2.3. Clearly, high school recruiting status isn’t all that great an indicator of ultimate success.
The NFL Draft isn’t a great predictor of success either. Johnny Unitas, Bart Starr and Joe Montana are on every list of the greatest quarterbacks of all-time. They combined to win 11 championships. Yet Montana was the 82nd player drafted in 1979, Starr the 200th in 1956 and Unitas the 105th in 1955. And if you think that’s before scouting got sophisticated, consider that Tom Brady was the 199th player chosen in 2006, adding four (after last week-end) championships to our list of unheralded and previously unappreciated great quarterbacks.
Brady’s Super Bowl opponent Russell Wilson, a champion a year ago, was only a third round selection in 2012 (for which the Seahawks were widely criticized). Bleacher Report gave the Seattle Seahawks draft that year an “F” as the worst of any NFL team and called Seattle’s selection of Wilson “by far the worst move of the day.” CBS gave the Wilson pick a “D” and the Seahawks draft as a whole a “C+.” Mel Kiper of ESPN awarded a “C-“ and Sports Illustrated offered a “C.”
To look at the issue from a different perspective, consider that tackle Nate Solder was the only first round draft choice on the New England Patriots’ offense in Sunday’s Super Bowl and that the list of first round failures is as long as the list of successes. The difference between Peyton Manning (an all-time great by any measure) and Ryan Leaf (a total bust who has just been released from prison) can be razor-thin in the picking (a majority of NFL GMs favored Leaf over Manning in the 1998 draft, when Manning and Leaf went 1-2) but monumental in the reaping, as we San Diegans know all too well. Moreover, every team has drafted players who wash out without producing much while undrafted players become meaningful contributors and more (Malcolm Butler, take a bow).
Despite all the effort and talent that goes into making good choices, college football recruiting and the NFL Draft both involve a huge component of luck…just like investing. There is a lot of skill involved in player evaluation and selection, of course. But luck is at least as important. Think about how much time, attention and effort is focused on recruiting, the Draft and getting things right. Still, each is largely a crap-shoot.
This maddening combination of luck and skill can be well-illustrated using the experience of the legendary gambler Billy Walters, a crucial member of the famous “computer group,” which used a careful and computerized process to make a fortune and, as a result, to revolutionize sports betting in the 1980s (I have written about this before). Today, Walters uses multiple consultants — mostly mathematicians, just like a hedge fund manager uses analysts — and still makes a ton of money. Walters’s process is to create his own betting line largely using statistical measures of the teams and then to bet when his view of a game is significantly different from available commercial betting lines.
Walters is staggeringly rich (according to 60 Minutes, he is “worth hundreds of millions of dollars”), but he claims a lifetime winning percentage of only 57 percent, as compared to the break-even of 52.38 percent (the winning percentage sports gamblers need to hit to offset paying out a 10 percent “vigorish” on their losing bets). Even so, while he has had losing months (randomness can overcome a good process for substantial periods of time), he has never had a losing year during this 30-year streak. But that winning streak only started after he made a major change in his approach. By his own account, Walters lost his shirt many times over before becoming focused, data-driven and careful to play the long game (making a profit while losing 43 percent of the time requires it, especially because the losing streaks can be very long indeed).
Common gambling practice had been (and often still is) to make fewer bets – to bet only on those games one is most sure of. But that approach is not thinking probabilistically. Walters makes as many bets as he can within the confines of his model (when he thinks the point spread is off by at least one-and-one-half points). Thus the key is lots of bets – whenever the data suggests a significant edge – with bet size being determined by the extent of the edge.
The betting market in Las Vegas isn’t much different from Wall Street. Fed by rumor, speculation and greed, teams, like investments, can grow hot or cold for no good reason. Fast-moving betting lines are remarkably similar to market bubbles. Walters insists that “[b]etting on a ball game is identical to betting on Wall Street.” He even claims that he has lost a lot of money in the markets and thinks the Wall Street “hustle” is far more dangerous than that in Las Vegas. It should be no surprise then that many prominent Wall Street-types have gambling histories, most prominently Ed Thorp. For more information, read Scott Patterson’s excellent book, The Quants. I even know a few.
Lady Luck is crucial to investment outcomes. There is no getting around it. Managing one’s portfolio so as to benefit the most from good luck and (even more importantly) to get hurt the least by bad luck are the keys to investment management. Doing so well is a remarkable skill, but not the sort of skill that’s commonly assumed, even (especially!) by professionals.
More to the point, if investment returns depend that heavily on luck and real investment skill is that elusive and rare, what should we do with our (or our clients’) money? For some answers, we turn to another sort of gambling — poker. Whether poker is deemed a game of chance or skill has significant legal implications for gamblers and those who earn a living from them. Moreover, it is clear that there is a good deal of skill involved in playing poker well. However, for our purposes, since poker is an excellent means to evaluate probabilities under uncertainty and provides a great deal of data, it is an excellent means for investors to learn about the markets, which are also governed by a great deal of uncertainty, meaning that the best investors must deal with probabilities amidst uncertainty extraordinarily well.
Poker involves tremendous luck and tremendous skill. By way of comparison, consider chess, which involves little luck (white plays first) and tremendous skill, Tic-Tac-Toe, which involves little luck (who goes first) but little skill (such that a precocious first grader can quickly get entirely up to speed), and Chutes and Ladders, which involves tremendous luck and little skill. More specifically, the luck component in poker is so high that the volatility of poker “returns” can make it maddeningly difficult for a better player to outperform even over substantial time periods (say many months of daily play, especially if the stakes are limited).
Looked at another way (as Nate Silver does in his fine book, The Signal and the Noise), a weaker player might be ahead of a stronger player after tens of thousands of hands. Silver estimates the likely outcomes for a very good (limit) Texas Hold ‘Em player over the course of 60,000 hands to range from up $275,000 to down $35,000 to 95 percent certainty. He also estimates that a player who has won $30,000 over his first 10,000 hands is still more likely than not to be a long-term loser (and short-term losers have mostly given up). That’s a lot of luck!
In the markets, the average investor underperforms due to costs. Poker is similar on account of the house’s rake. Yet most investors — like most poker players and most people generally, due to optimism bias — think they are better (and often much better) than the norm. With poker players, the truth can be beaten into them as their losses mount. Since the markets are biased upward (most underperformers have positive returns overall), investors tend to remain delusional for a much longer time. Some never recover.
In a “quasi-experimental” study, researchers set out to examine these questions in poker. They got together a group of both expert and novice poker players to play fixed games, meaning that the players received hands that the researchers had set up – without the knowledge of the players – to test how things would go under various scenarios. The results revealed that while the cards dealt (luck) largely predicted the winner, skill was crucial to reducing losses when players were dealt a bad hand.
That’s a true if unsurprising result as far as it goes. But the conclusion of the study (“that poker should be regarded as a game of chance”) is clearly overstated. The study has been rightly criticized for not looking at enough data. It’s surely true that over the short-term, luck dominates skill in poker. However, over longer and longer periods of time – a much larger data base of hands – a slight skill advantage will result in a positive win rate because no player will have better cards in the aggregate. In other words, given enough time, luck cancels itself out.
As Silver argues in The Signal and the Noise, especially when the skill differential is not great, the interesting question is how long it will take for skill to win out. Another interesting question is why skill wins out. I suspect that – consistent with the study – the primary reason is that the expert player makes fewer mistakes. Science seeks the truth by uncovering what is false. What’s left is likely to be true. So, if you are a beginner, you’d better have beginner’s luck or you might be broke by the end of the game.
Investing also combines skill and luck, but is perhaps more complicated because of the personal aspect (our psychology makes it hard even for professionals to invest optimally), the difficulty in parsing out the market’s “psychology” the way a good poker player reads tells, and the much larger numbers of variables involved in ultimate outcomes. There is decidedly less market skill on display in the markets than poker skill at major tournaments.
As Charley Ellis has pointed out, “in a random 12-month period, about 60% of mutual fund managers underperform; lengthen the period to 10 years and the proportion of managers who underperform rises to about 70%. Although the data are not robust for 20-year periods, the proportion of managers who fall behind the market for this longer period is about 80%. At least as concerning, equity managers who underperform do so by roughly twice as much as the ‘outperforming’ funds beat their chosen benchmarks.” Moreover, “[N]ew research on the performance of institutional portfolios shows that after risk adjustment, 24% of funds fall significantly short of their chosen market benchmark and have negative alpha, 75% of funds roughly match the market and have zero alpha, and well under 1% achieve superior results after costs — a number not statistically significantly different from zero.”
Despite those poor results and thus the difficulty in making a case for investment skill, our tendency is to give money to managers who have recently done well and pull money from managers who have recently done poorly. Consistent with many studies over a wide variety of time periods, this tendency applies to individuals, advisors, and “expert” consultants alike. Sadly, however, this past performance does not indicate future results.
As Kahneman and Klein have established, true expertise can be developed such that expert intuition and guidance are routinely reliable, but only in areas that involve little luck. However, as Kahneman emphasizes, when intuition cannot work (as with investing), we tend to substitute a simpler question for the real question at issue and answer it. We can thus oversimplify a difficult problem or even miss the point entirely.
That’s why investment professionals all acknowledge (as their marketing materials must, legally) that past performance is not indicative of future results and then begin their presentations by talking about their past performance (at least when that performance is reasonably good). Even so, being human, underperforming managers attribute their prior success to skill (which led to their hiring) and the more current problems to bad luck — classic self-serving bias.
In real-life poker, a high win percentage is negatively correlated with win rate. In other words, the players that win the most money don’t win the most hands. They win the biggest hands. As one of the world’s great poker players, Tom Dwan, told Nate Silver, “It’s important in most areas of life to come up with a probability instead of a yes or no. It’s a huge flaw that people make in a lot of areas they analyze.”
Therefore, the advantage to recruiting five-star prospects is not that they are “sure things.” It’s that they are more likely to succeed and to succeed in a big way than their (apparently) less talented competitors. Thus Nick Saban landed two five-star cornerbacks for Alabama yesterday, but they will have to compete with the two five-star cornerbacks signed last year, neither of which were starters this past season. They will also have to compete with other, lesser-regarded recruits, some of whom may well turn out to be better players. Saban recruits five-star players to improve his odds of success.
The extent to the luck factor guarantees lots of “noisiness” in the data. As Silver writes in The Signal and the Noise, in a highly competitive probabilistic enterprise, it is really hard to make money. It’s only possible to do so “around the margins.” Moreover, the more less-than-stellar players that are removed (or remove themselves) from the game, the harder it is to profit (what Michael Mauboussin describes as the “paradox of skill”). It is obvious that as the investment landscape has become increasingly professionalized and computerized, it has become ever harder to outperform.
As Morningstar’s John Rekenthaler argues, in order to be successful, active management requires (a) low costs; (b) patience; and (c) skin in the game. It also demands strategies and approaches that persistently work over time. These seem to include the smart-beta flavors of small-company, value, fundamental weighting, equal-weighting, and momentum. Concentration can work too if the manager is especially skilled (like Seth Klarman). Low volatility has worked for a substantial period and seems to persist, but its current popularity makes me nervous.
As Silver emphasizes in The Signal and the Noise, we readily overestimate the degree of predictability in complex systems. The experts we see in the media are much too sure of themselves (I have written about this problem in our industry from a slightly different too). Much of what we attribute to skill is actually luck.
In all probabilistic fields, like investing, football talent evaluation and gambling, the best performers dwell on process. That’s why Alabama football coach Nick Saban emphasizes what he calls a “process focus.” He “teaches his players to stop actually thinking about winning and losing and instead focus on those daily activities that cause success.” For him, the keys are good practice and careful execution.
In 1948, John Wooden came to the West Coast and created a dynasty with his UCLA basketball team. He defined mental toughness as having the ability to judge oneself on effort rather than results. Armed with a process focus, Coach Wooden eventually led his teams to 10 national championships in 12 years. He did this without ever saying the word “winning” to his players. But he did carefully teach them how to put on their socks and tie their shoes.
Maintaining good process is really hard to do psychologically, emotionally, and organizationally. But it is absolutely imperative for investment success (and for gambling success too). For investors, the lessons to be gained here relate to diversification, a carefully delineated and bounded process, clear execution rules, and stick-to-itiveness over the long haul. This doesn’t mean that quants should control everything. Old school analysis and judgment still matter, perhaps more than ever since the pile of available data has gotten so large. But it does mean that our conclusions need to be consistent with and supported by the data, no matter how bizarre the numbers or how long the streak (a roulette wheel at the Rio in Las Vegas landed on 19 an incredible seven consecutive times).
Many factors determine a college football player’s success, such as talent, mental acuity, health, work ethic, growth, fit within the system, good teammates, relationships with coaches and teammates, other commitments (academics) and distractions, and personal as well as family relationships and support. Since many of them are outside the control of the player or team, luck is a huge component to the ultimate success of the player evaluation process. To succeed, we need to put ourselves in the best possible position to deal with the randomness that is inevitable, in gambling, investing or player evaluation.