Game Theory’s Value in Fantasy Football
Jonathan Bales is the author of Fantasy Football for Smart People: How to Dominate Your Draft—the #1 ranked football book in the world. He writes for the New York Times. You can purchase his 2012 fantasy football rankings for $2.99.
After Robert accepted my invitation to dominate his Dollarnaire Fantasy Football League, I posted part of a chapter from my book Fantasy Football for Smart People here at BSO. That was Part I, and this is Part II.
Game Theory’s Value in Fantasy Football
A few years ago, I was discussing an upcoming fantasy draft with my dad (who is part of my dynasty league and was new to the game that year). He mentioned he was thinking of taking Bears running back Matt Forte with the second overall selection. At that time, Forte wasn’t going in the first few rounds of drafts (he was ranked No. 43 overall by ESPN that year).
Naturally, I asked him why he would draft Forte so high. “Because he’s going to score the second-most points,” he quickly replied.
Seems straightforward enough, and if we use a traditional draft strategy (and even a complex one such as VORP), Forte might be the guy to whom the numbers lead. If my dad had him ranked far ahead of the third option such that Forte’s value made him an outlier among running backs, conventional fantasy football draft strategy says to draft him.
But we all know that isn’t right. You don’t win championships by selecting fourth-round projected players in the top two picks, even if you think they will lead the league in points. So it’s quite obvious we need to implement game theory into our draft strategies, even if it is in the form of a quick comparison to consensus rankings. The beliefs of the competition must affect our decisions. The extent to which we can utilize game theory and the methodology we employ to do so will be the subject of later analysis.
You’re So Predictable
There’s a final component of complex fantasy football draft strategy, and it is two-pronged. Predictability in fantasy football is absolutely vital to draft success, and it is one of the most overlooked aspects of the process. By predictability, I mean the ability to correctly assess the consistency inherent to particular positions, as well as each player within a position.
Let’s take an example. Suppose the top defense in fantasy football scores an obscene amount of points in a given year (we’ll again say 1,000), and all other defenses are very far behind. Securing the No. 1 defense in this hypothetical scenario obviously holds tremendous value. Do so, and you basically win the league.
Also note a draft strategy that combines VORP and game theory would lead to the hypothetical selection of your top-rated defense at every draft spot. If you project the Steelers to score 1,000 points and every other team defense to tally around 200, for example, clearly the Steelers are extremely “scarce.” Further, if every other owner has similar projections, game theory would advocate selecting Pittsburgh’s D with the No. 1 overall pick.
So why might this be obviously boneheaded move? Because if there is very little predictability within team defenses from year to year, the early selection of one is illogical. As long as there is a limit to the number of defenses you can draft, the scarcity of the top defense has zero value to an owner if it is impossible to predict which defense will score those 1,000 points. With no predictive ability, you would be just as likely to draft the top defense in the last round as the first.
As fantasy owners, we want to minimize the luck that is inherent to the game. The early selection of a position whose year-to-year rankings are basically as predictable as a roulette wheel increases the luck needed to win.
Interestingly, the lack of predictability among team defense in my hypothetical scenario is not far from reality. Actually, there is just about no predictive ability within the position. Defenses that finished high in the fantasy rankings in a given year are no more likely to do so the following year than the bottom dwellers.
Predictability Among Players
The ability to predict the final rankings within a given position is a matter of consistency; how consistent are top fantasy performers at particular positions? I will of course analyze this topic more, but another form of predictability and consistency with which we need to concern ourselves is that among individual players.
We all know certain players carry more risk than others heading into a football season. Whether it is due to poor character or an injury in the previous season, there are players we label as “high risk/high reward,” and the risk we associate with them has huge implications on our ability to predict their future performance.
In 2011, I had Chris Johnson projected to score the most fantasy points among all running backs, ahead of Adrian Peterson, Ray Rice, and Arian Foster. But I had CJ2K ranked behind all of those players because the risk surrounding Johnson’s contract situation and potential holdout made the ability to predict his 2011 performance quite volatile. The greater the ability to predict a player’s performance, the more weight that can be given to his projections.
Now, there are formulas I use to assess risk/reward (read on, grasshopper), and there are certainly times when gambling on a high-risk player is prudent. The entire fantasy draft process is a collection of calculated gambles, and with the right use of statistics, you can tilt the scales in your favor.
What To Anticipate
As you read on, try to remember the overarching concepts of VORP, game theory, and predictability that lie at the core of draft strategy. With these notions in mind, I will take you through a more in-depth analysis of each idea, hitting on a variety of subcategories in the process. Among the topics I will discuss are:
- Why week-to-week consistency is almost worthless
- Why you should wait until the last two rounds for a defense and kicker
- The myth of the “overworked” running back
- Why tight end is the most predictable position
- Why fantasy football is a stock market
- The value of pairing a receiver with his quarterback
- How to create power rating systems based on projections, then tiered-rankings based on your power ratings
- How regression can be exploited
- Why quarterback rushing yards are valuable
- Why quarterback/tight end might be the best 1-2 punch in 2012
- How to predict yards-per-carry (and a lot of other stats)
- How to incorporate consensus rankings into your board
- Why you should perform mock drafts
- A whole lot more