Index Investing 201: Applying the EMH in Practice

Photo: Imagn Images

I'm a dynasty league junkie with a problem; I love dynasty leagues, and I love the NFL, but have little interest in the college game. For 364 days of the year, this doesn't matter... but on the day of my rookie draft, I always felt like I was operating at a disadvantage. Lacking any first-hand knowledge, I had to rely on the opinions and rankings of others. Giving up that control always felt unbearable, but I had no alternative- I certainly couldn't remain competitive against guys with that kind of time and experience by half-heartedly trying to scout the players myself.

This status quo remained for several years until it was interrupted by an inspiration from an unlikely source. In the world of finance, there is a method that allows an unskilled manager with no knowledge of the companies in which he invests to achieve returns that equal, or even better, the returns of his highly-educated, highly-experienced, highly-skilled peers. I've adopted, (and adapted), many of those key principles into a philosophy towards ranking rookies that I call "index investing", and in this series I will walk through the philosophical underpinnings of the strategy, the process of implementing the strategy, and the real-world impact of the strategy. Today, we continue the series with...

INDEX INVESTING 201: Leveraging the emh to dominate the rookie draft

In part 1 of this series, I explained the "Efficient Market Hypothesis", laid out my reasons for believing that the NFL draft is an efficient market, and discussed what implications that belief has for rookie drafts in dynasty leagues. In this article, I will walk through the actual process of using the Efficient Market Hypothesis to generate a set of rookie rankings. In the third and final installment of this series, I will discuss what advantages this approach to rookie rankings provides.

In case you missed Index Investing 101, the Efficient Market Hypothesis states that a market is "informationally efficient", which is a fancy way of saying that it is very good at pricing assets based on the information that is available. In other words, no individual equipped with the same information will be able to create a set of rankings that will outperform the NFL consensus at a rate better than chance. Or, to put it another way, we know that some players in the draft will be stars and some will be busts, but we aren't equipped to tell ahead of time which will be which. If we are faced with a single player and tasked with determining whether he will prove to be a good value at his draft position or a bad value at his draft position, our chances of success will be exactly the same whether we watch 100 hours of game tape or simply flip a coin.

Obviously this has huge ramifications for dynasty leagues, and especially for rookie rankings. If watching tape, scouting players, or the various other actions we perform can't tell us ahead of time which players will be busts and which will be stars, then there's no need to do any of it. When competing against owners who do these things, we aren't operating at a disadvantage at all. Instead, we all have access to the exact same information from the start; we all know where a player was drafted, and all the hard work in the world won't improve upon that initial knowledge.

In fact, this makes perfect sense. As dynasty leagues have exploded in popularity, we've seen a proliferation of very smart, very passionate people who devote hundreds or even thousands of hours a year to trying to find out which players are better or worse than people think. These passionate fans, however, are hopelessly outclassed. Even the worst NFL scouting department employs many full-time scouts, each of whom spend at least as much time scouting players as even the most devoted of fantasy enthusiasts. Each front office backs up that scouting department with an array of other experts and specialists, such as doctors and psychologists. Each front office has unfettered access to every corner of a prospect's life, and has millions of dollars worth of resources to devote to compiling and synthesizing all of this information. Most importantly, each NFL front office is highly incentivized. NFL franchises want to get it right because there are tens of millions of dollars at stake. Professional scouts want to get it right because they know that their continued employment depends on their continued performance. Fantasy football owners are knowledgeable and passionate, but few of them have their ability to provide for their family riding on their ability to get it right. Given the immense disparity in resources and incentives, is it any wonder that fantasy experts struggle to unearth anything that scouting departments haven't already considered?

On the one hand, this realization can feel demoralizing, even defeating. The NFL is a billion dollar juggernaut capable of squashing my feeble efforts like a bug. On the other hand, it can also be liberating. NFL franchises have the time, the talent, and the resources to make the most informed call possible, and I cannot possibly hope to compete against them when it comes to evaluating talent, so I don't even have to try. It's one less thing to worry about in the predraft process- all I have to do is google "NFL draft results" and I can check off the "talent evaluation" box in my checklist. No more time wasted searching for overlooked gems or sleepers. No more hours spent fretting over red flags or warning signs. More time for myself, time that I can now devote to other, more productive means of improving my dynasty team.

Does this mean that I just print out a copy of the NFL draft results and make picks from it straight down the line? Of course not. I rank players based on how much I expect them to produce compared to their peers. While talent is the primary component in production, it is far from the only component; as the old saying goes, "Production = Talent + Opportunity". Even if I copy off of the NFL's homework in terms of talent evaluations, I still have to do some of my own work to generate a usable set of rookie rankings.

Walking Through the Process

To begin with, let's get a list of all skill position players selected in the first two rounds of the 2013 NFL draft:

Now, the first thing we can do is adjust for relative positional value. In the NFL, quarterback is by far the most valuable position. As a result, NFL franchises will reach and draft a quarterback higher than his talent alone would dictate. In most fantasy leagues, on the other hand, quarterback is just the third most valuable position, behind running back and wide receiver. So, let's start by adjusting the quarterbacks heavily downward to compensate for the inflated talent evaluation and to reflect the relative positional importance. Individual owners might vary in terms of how steep they make these adjustments, and there's no perfect answer, but my own adjustments are steep and result in a list that looks like this:

Now that the quarterbacks are sorted out, we move to the tight ends. Like quarterbacks, tight ends are not a premium position in fantasy football, so we should move them down with respect to their RB and WR peers. On the other hand, tight ends are also not a premium position in the NFL, so their draft position is unlikely to be as inflated as a quarterback's. If anything, we might expect a Tight End's draft position to mildly underrate his talent level, since he plays a non-premium position; many teams would prefer a slightly less talented receiver over a slightly more talented tight end, so if a receiver and a tight end are both drafted in the same spot, we can assume that the tight end is slightly more talented. Combine those two effect, and we'll see the tight ends move down our board (because the position isn't as valuable), but not as sharply as the quarterbacks (because their actual talent level is likely underrated instead of overrated).

Next, we want to address the running back position. Running backs are largely fungible at the NFL level, which means that, like with tight ends, draft position probably underrates a running back's true talent level. Further, running backs are generally the most valuable players in fantasy football, especially in leagues that do not award a point per reception. As a result, we should adjust the running backs strongly upward in our rankings. In a PPR league, I wind up with this:

Notice, for instance, that Giovani Bernard's position was important enough to overcome the small 10-pick advantage of DeAndre Hopkins, but not enough to bridge the nearly 30-pick gap from Tavon Austin. In a non-PPR league, these adjustments would be slightly steeper still- certainly enough for the trio of Ball, Lacy, and Michael to overtake Justin Hunter, for instance.

And that's it- we now have the rookies ranked based on what position they play and where they were picked. We never have to adjust the receivers, as they were adjusted automatically when we moved all of the other players around them. The receiver position basically served as a framework around which we slotted the rest of the players.

Actually, the order we perform the adjustments is unimportant; you can adjust any three positions in any order you choose, and the fourth position will always settle into place automatically. I used this particular order (QB, then TE, then RB) simply because it's the easiest for me- quarterback adjustments are quite steep and easy to figure, and it's always made sense to save receiver for last because it's the most numerous position. By saving receiver for last, I only have to individually adjust 11 of the 17 names. If I'd saved quarterback for last, I'd have to adjust 15 of the 17, instead. If you'd prefer to address the positions in a different order, though, then do whatever works best for you- with practice, you should wind up with the same list regardless of what order you perform the adjustments.

Fine-tuning the Rankings

Our general rankings are created, but the process does not end here. Up until this point, none of our changes have considered anything about the actual players in question- we've just been applying very general filters to shuffle draft position based on the relative importance of each position in the NFL compared to fantasy football. At this point, we leave the realm of the general and we enter the realm of the specific. Here, I will briefly touch on some specific considerations fantasy owners might take into account during these final fine-tuning adjustments:

Risk. The NFL tends to be relatively risk-adverse because mistakes cost millions of dollars, and salary cap implications can be devastating. Fantasy football owners, on the other hand, can take more chances. If a fantasy draft pick busts, most leagues allow for him to be cut at no additional cost to the owner. As a result, players who have elevated risk profiles should probably be moved up in rookie rankings. Some players who might fit the bill this year include Tavon Austin (far smaller than any previous top-10 pick), Cordarrelle Patterson (very limited experience at the major-college level), Eddie Lacy (medical concerns, questions about work ethic), and Christine Michael (character concerns). In the later rounds, a classic example is receiver Da'Rick Rogers, a talented player who went undrafted after failing multiple drug tests and transfering out of Division I football. We can safely assume that his talent level is higher than his draft position.

Usage. NFL franchises value things like blocking and special teams contributions, but most fantasy leagues do not. As a result, players who are exceptional blockers or potential special teams dynamos can be discounted some, under the assumption that part of their draft position reflects their contributions in this area. Meanwhile, players who are liabilities as a blocker or who do not play special teams can be slightly elevated, as their liabilities in these areas are not relevant to most fantasy owners. Players who fit either of these criteria include Tavon Austin (potential returner) and Cordarrelle Patterson (potential returner). In later rounds, a classic example is tight end Jordan Reed, a woeful blocker who was presumably drafted entirely for his skills as a receiver. We can probably assume that his receiving production will be disproportionate to his overall talent level.

Play style. Most fantasy scoring systems reward certain types of production over others. For instance, in PPR leagues, a running back that catches a lot of passes will carry disproportionate value compared to one who gets most of his yardage on the ground. In most scoring systems, quarterbacks are rewarded at least twice as much for rushing yardage as they are for passing yardage. In leagues that give bonus points for long touchdowns, faster players might receive a small bump over slower players. Players who fit some of these criteria include Giovani Bernard (a prolific receiver at the college level) or E.J. Manuel (a moderately mobile QB). In later rounds, a classic example is receiver Marquise Goodwin, an Olympic long jumper who runs a 4.2 40-yard dash. We can probably conclude that he has an increased chance of scoring long touchdowns, as a result.

Competition. Obviously a player can only be fantasy relevant if he manages to get on the field. In order to reflect the risk that a player never wins the starting job, players should be downgraded based on how tough the competition is on their team. Players who are impacted by this include Tyler Eifert (shares the field with the established Jermaine Gresham), Giovani Bernard (faces weak competition from BenJarvus Green-Ellis), Eddie Lacy (faces moderate competition from fellow rookie Jonathan Franklin), and Christine Michael (faces intense competition from established All Pro Marshawn Lynch). Meanwhile, at receiver, players like Tavon Austin, Justin Hunter, and Robert Woods find themselves in a deep competition among many unproven challengers, while Cordarrelle Patterson and DeAndre Hopkins face shallower competition which includes some well-established veterans, and Aaron Dobson finds himself on a wide-open depth chart. In the later rounds, a classic example is receiver Tavarres King, an intruiging prospect who finds himself sharing a depth chart with Demaryius Thomas, Wes Welker, and Eric Decker. We can infer that his production will likely lag behind his talent, especially in the near term.

Supporting cast. It should come as no surprise that players in good offenses tend to be more productive than players in bad offenses. A back running behind an elite offensive line is more valuable than a comparable back behind a terrible line, and a receiver catching passes from a Hall of Famer should produce more than a similar receiver catching passes from a journeyman or struggling rookie. A player like Montee Ball finds himself in an offense that has consistently produced top-10 caliber fantasy numbers from its backs, while Eddie Lacy joins an offense that hasn't produced a 100-yard rusher since week 5 of 2010. In the later rounds, running back Marcus Lattimore found himself drafted to a team that features the best young offensive line in football, while Stepfan Taylor was drafted just nine picks later by a team that might have the worst. We can then predict that Lattimore's production will outpace his talent, while Taylor's will lag behind.

New information. As I said, the Efficient Market Hypothesis only applies to information that was available at the time of the draft. Rankings must be adjusted to account for any new information that has come out since then. If the coaching staff is particularly complimentary of a player, he can have his ranking adjusted upward. If a player struggles during camps, he can have his ranking adjusted downward. A current example would be Cordarrelle Patterson, who has drawn rave reviews (and comparison to the movie "The Matrix") in his first training camp.

Putting it all together.

Let's take a look at how all of these adjustments might affect a particular player by focusing on Montee Ball, Eddie Lacy, and Christine Michael. Prior to applying specific adjustments, all three backs are valued pretty comparably, as all went within five picks of each other at the end of the second round. Initially, Lacy and Michael get a slight bonus for being riskier picks, enough to surpass Ball. None of the three is particularly known for his blocking prowess, and none was particularly accomplished as a receiver, so we see no relative adjustment based on usage type. The competition adjustment is much more interesting, as Montee Ball is competing for a job against 2012 3rd rounder Ronnie Hillman and veteran Knowshon Moreno, Eddie Lacy is competing against fellow rookie Jonathan Franklin and incumbent DuJuan Harris, and Christine Michael shares a depth chart with reigning first-team AP All Pro back Marshawn Lynch and backup Robert Turbin. As a result, Michael gets heavily downgraded compared to his peers, as his chances of winning a starting job are minuscule and it will likely be years before he provides any fantasy value. When we compare supporting cast, Montee Ball gets a strong upgrade, as Denver, John Fox, and Peyton Manning all have a history of getting fantastic production from the running back position. Eddie Lacy, on the other hand, is downgraded for joining an offense that hasn't had quality RB production since the 2009 season, (and hasn't even had a 100 yard rusher in nearly three years). Finally, while there hasn't been time for much new information to surface yet, Montee Ball does get downgraded some after failing to distinguish himself in early competition for the starting job.

So, taking all of these factors into consideration, what might our final rookie rankings look like? After making the necessary adjustments, and considering only players drafted in the first two rounds of the NFL draft, my rookie rankings for PPR leagues look like this: 

Obviously the process of getting from raw draft position to final rankings leaves a lot of room for interpretation. We have to make value judgments about just how much to weight positional adjustments, such as whether a quarterback taken in the middle of the first round is worth more than an running back taken at the end of the second. We have to make judgments about just how much to weight specific factors such as competition. We have to make subjective evaluations of the quality of a player's situation. Still, by using draft position as a starting point and anchoring all of our evaluations to that, our rankings maintain an objective foundation and comport much more with the NFL's own talent evaluations. In short, by basing our rankings off of draft position, we increase our chances of landing more talented players in our rookie drafts.

This wraps up the practical application portion of the series. In the third and final installment, I will lay out the advantages of creating rookie rankings using the Index Investing method and make the case for implementing these principles in your own dynasty leagues. If Part 1 was designed to answer the "What?" of Index Investing, and Part 2 was written to explain the "How?" of Index Investing, Part 3 will fill in the last remaining blanks by addressing the "Why?" of Index Investing. Stay tuned!