For fans of American football, with the start of the new season just two weeks away, a rush of frenzied activity is underway by millions to rate, select, and position their 'fantasy' teams for the upcoming year.
American football, and the evaluation of its players, has traditionally been much less focused on statistical measurements and quantitative analysis of performance than say other sports like baseball and basketball. There are many reasons for this historical de-emphasis on statistics. For one, there are many, many roles on a football team that don't register simple, easy to grasp numbers like touchdowns scored or yards gained. Second, the nature of the game itself, eleven players to a side, highly structured and orchestrated roles and actions on most every play, make considering 'team' success more straightforward and easily understood than individual performance. And lastly, for many of the most important positions like Quarterback, past attempts to develop statistical-based measures or performance have been considered lacking, as many experienced football analysts claim that simply doing calculations on yards gained, passing completions, and even passing touchdowns registered can only offer partial insight into what defines and demonstrates superior performance for that critical position.
The primary metric that has been commonly used to assess and compare quarterbacks has been the Quarterback rating, a measure that takes into account the raw data surrounding the player's actions (passing yards, touchdowns, pass completion rates, etc.), applies some weighting factors to to the data, and produces a combined score or rating for the player, usually falling between about 85 and 100. But the main problem with the Quarterback rating (apart from no one really understanding how it is calculated), is that it is a statistical measurement only, i.e. it applies no situational context to performance. A three-yard pass completion in the early stages of the game gets weighted exactly the same as a three-yard pass completion at the end of the game, perhaps by converting the play, the quarterback's team was able to secure possession of the ball at a critical stage, and cement an important victory.
Some clever statisticians at ESPN are attempting to improves on the statistical evaluation of quarterbacks by introducing a new metric they all 'Total Quarterback Rating', or QBR. QBR will factor in many of the contextual indicators that play an important role in assessing player performance. Game situation, personnel on the field, formations used and more will all play a role in the metric. This will, hopefully, shine a more complete light on the evaluation of NFL quarterbacks. But it is much, much harder to create and calculate than simple math applied to the game box score.
In football, and I suppose even in most organizations, the context in which performance is captured is often far more important, and more difficult to account for, than simply tracking the 'raw scores' or activities themselves. Was the quarterback under extreme duress when he passed for the touchdown? Was your sales manager under extreme duress when she successfully navigated through a complex contract negotiation to win that important account? Are you adequately considering the relative experience levels of your key player's support teams in your evaluations? How about the differences in competitive context across markets, lines of business, or geographies?
The first, and necessary step is chronicling performance - i.e. What happened?
The harder part, and even more important part, is understanding the conditions present when it happened, and what that means for the future.
Aside - J-E-T-S - JETS, JETS, JETS!!!!!