The Dual Nature of Reality
Taoist philosophy identifies two aspects of reality: the true essence, and how we try to describe or categorize that essence. The true nature of things is impossible for us to fully grasp with our limited perspective and as soon as we label something we lose a part of it. It is essential that we not lose sight of this distinction. The true nature of things is more important than our categories, yet our bias pulls us in the direction of our definitions, and as a result, away from reality.
All the frameworks we use to measure performance on the football field suffer from this bias. And as bettors, we want to be grounded in so-called “objective” evaluations so we can make more precise predictions. But that very desire can block our access to deeper understanding and lead us to faulty conclusions. There are probably infinite applications of this principle but I want to talk about two in particular.
Rankings
We love to rank things. It’s one of our favorite forms of categorization. Few things stir up engagement online in NFL circles faster than team power rankings and quarterback rankings. These can be useful tools for discussion and have value. But these static forms of characterization always pull us away from reality.
Quarterback performance is dynamic and does not exist outside of context. Last season analysts disagreed about whether Jalen Hurts was “actually” an elite quarterback or just a product of his circumstances. Some more production-focused rankers had him as the second-best quarterback in the NFL. Others, like the Ringer, put him ninth.
I don’t take issue with either of these rankings. They are both valuable in their own way. The high ranking describes the level Hurts was able to achieve in his specific circumstances. Given the quality of his weapons, offensive line, and coaching staff, how the scheme fit his strengths, and the relative ease of schedule, he was an absolute force.
But the lower ranking tries to judge his performance in a predictive way, recognizing that flaws in his game will likely prevent him from sustaining that level of success in more trying circumstances. Hurts is also a young player who might move up and down within these types of evaluations based on how he progresses.
The problem arises when we become attached to these static descriptors. We proclaim a certain ranking and then it becomes our job to defend that ranking, even subconsciously. Evidence is filtered through that lens and prevents us from responding to dynamic change as a player progresses and circumstances evolve.
As an alternative to ranking quarterbacks, we should identify specific things we observe and incorporate them into our perception of that player. In my view, Hurts is currently a patient quarterback who can capitalize on defensive mistakes and deliver the ball to anyone who gets open. He also trusts his receivers to win downfield so he throws into coverage to give them a chance. But he struggles relative to his peers when asked to create in the passing game under pressure and/or throwing into tight windows, which requires timing and anticipation. He also keeps defenses honest with the threat of elite rushing ability, which helps feed into his strengths by creating more open or one-on-one opportunities.
It’s much harder to turn this into a snappy Tweet, and my descriptions do not holistically capture the essence of Jalen Hurts the quarterback. They don’t even capture my perspective fully, but this isn’t a blog about Jalen Hurts. My description is a point-in-time summary of my current perspective on something I will never fully grasp. But as an NFL handicapper, I will continue to seek a better understanding of the reality of Hurts and shift my opinion accordingly, so as not to get trapped defending static descriptors.
As always, I’m writing to myself as much as to anyone else. I have engaged in these (pointless) ranking debates in the past and probably will be pulled in by my ego in the future. But I don’t want to. I want to focus on productive evaluations instead.
Statistics
Statistics is the ultimate language of sports analysis. Most statistics are useful in some way and the recent developments in analytics have really propelled us forward in using statistics effectively. It’s essentially impossible to model effectively without using statistics. But every statistic is a measurement, and when we measure things, we lose something about the essence of what we are measuring.
We know this intuitively. We know, for example, a quarterback’s completion percentage is not a precise measurement of a quarterback’s accuracy. Even the more advanced measurements of Completion Percentage over Expectation (CPOE) and Accuracy Rate Over Expectation (AROE) fail to account for the circumstances that create them. They measure very specific things and provide value, but do not capture the true essence of quarterback play. Rather, they describe it. And in doing so, they depart from it.
There’s nothing wrong with using statistics. They are like words of any language. They approximate reality in a way that facilitates communication and understanding. They create shared reference points and measure things objectively (for the most part). But they carry flaws.
On a small scale, statistics can be massively skewed. Take the example of an interception Joe Burrow threw against the Chargers in 2021. He threw a perfectly placed ball down the sideline for an easy 71-yard touchdown to Ja’marr Chase, except Chase bobbled it right into the hands of the defender he beat on the route for an interception instead. His statistics for that game were massively skewed by something with little predictive value and completely outside of his control.
But even on a larger scale, where these kinds of fluky errors smooth out with large sample sizes, statistics are a language; not reality itself. They reward certain tendencies and contain biases. Quarterbacks who take the check down short of the sticks on third down will have a higher completion percentage and yards per attempt on those plays than a more aggressive quarterback who insists on throwing past the first down marker, even though the more aggressive quarterback should convert more first downs.
More advanced statistics contain their own biases too. It’s inevitable. As an originator, a person seeking to win must be able to differentiate their view of a game from the consensus market opinion. And the market speaks the language of statistics. The best modelers speak it better than anyone else, but it generally still guides the market.
Therefore, understanding the ways that the language of statistics mischaracterizes reality can be a key that unlocks a winning strategy. This does not mean ignoring statistics. It means recognizing the construct and the biases the statistics contain. Where the biases create gaps between definitions and reality, there can be value.
Now, I’ll emphasize that this way of thinking is not for everyone. And bettors can (and do) win without thinking in this way. It can also make it difficult to engage in conversations that use these necessary constructs when you don’t view reality through them exclusively.
But you have to play the game that you want to win. Mass approval is not my aim. Handicapping effectively is.