Sports Analytics 101: Productivity vs. Style

Sports Analytics 101 is a series of blog posts outlining the core concepts behind sports analytics in non-technical terms. You can find all available installments in the seriehere.

In an earlier post, I introduced a framework for thinking about an individual sports analytics metric. This framework is essentially mental “paperwork” to fill out whenever you use a new metric to ensure you understand what the metric is and what it isn’t.

In using the framework, we first establish the name of the metric and what it’s being used to quantify. Next, we establish whether the metric is a fact or a proxy and whether the metric is descriptive or predictive. That brings us to the question of whether the metric is a productivity metric or a style metric.

Some metrics are intended to provide insight into how productive a particular player or team is, either holistically or within an individual area of the game. These metrics are often on a scale where higher values mean the player or team is better and lower values indicate the player or team is worse (though sometimes the scale is reversed, and lower values indicate that a player or team is better). In short, these metrics quantify whether a player or team is good or bad at something and how good or bad they are at it.

Take, for example, Free Throw % as used to quantify how good of a free throw shooter a basketball player is. In using Free Throw % in this way, we are measuring how productive or efficient a player is from the free throw line. A higher value means the player is better at shooting free throws. A lower value means the player is worse at shooting free throws. Simple.

On the other hand, some metrics are constructed not to illustrate how good a player or a team is at something, but rather the style by which the player or team does something. Consider the soccer metric % of Shots Headed, calculated as:

% of Shots Headed=Headed Shots / Total Shots

This metric can be illustrative of the way a player tends to shoot, but it doesn’t quantify how good a player is at shooting or even how good they are at shooting with their head. It only quantifies the style or method of shooting the player tends towards. Drew might have a much higher % of Shots Headed than Joe, but that doesn’t mean that Drew is a better or worse shooter than Joe, it just means that Drew and Joe tend to play differently.

But just because style metrics can’t tell us how good a particular player or team is doesn’t mean they can’t be incredibly useful. Let’s say a soccer team, the Dragons, is preparing for its upcoming match against the Knights. The manager of the Dragons has asked the team’s analytics staff to prepare some insights on the Knights to inform the Dragons’ tactical strategy for the match.

The Dragons’ analytics staff uses several style metrics to quantify how the Knights tend to play. One of these metrics measures the frequency with which different Knights players pass to one another. Using this metric, the Dragons’ analytics staff is able to identify a certain passing channel that the Knights disproportionately use in possession build-up. With this information, the Dragons’ manager implements a defensive shape intended to disrupt this particular passing channel, thus throwing a wrench in the Knights’ most common build-up pattern.

In this case, the style metric that measured the usage of each passing channel did not make any claims about how good of a team the Knights are or how effective this particular passing channel is, only about the tendency of the Knights to use the passing channel.

Frequently a style metric can be analyzed alongside a productivity metric to provide even more insight. For example, the Dragons might have an additional metric that illustrates how frequently the use of a certain passing channel by the Knights leads to a goalscoring opportunity. This would be a productivity metric because it measures the offensive productivity of a passing channel. The Dragons’ analytics staff could look at both the metric that illustrates the passing channels that the Knights rely on and the metric that illustrates which passing channels tend to be the most productive. If the analytics staff found that the Knights relied on a generally unproductive passing channel, they might be happy to let the Knights use it. If they found that the Knights relied on a rather productive passing channel, the Dragons might be more inclined to tactically counter it.

Another application of style metrics is to identify players that fit a particular team’s system. Remember the example from the second Team Use Cases post where a soccer team, Green Lake FC, is using player data to narrow down the scouting pool to players that fit a profile? That situation is perfect for the use of style metrics alongside productivity metrics. Obviously, Green Lake FC will only want to scout players that are good enough for the role they’re looking to fill, but the team will also want to focus on potential signings that complement the players already on the team.

Say, for example, that Green Lake already has two excellent wingers who specialize in delivering high, accurate crosses into the box. If Green Lake is presented with two strikers of equivalent productivity but one of the two players has a significantly higher % of Shots Headed, the team might prefer the player with the higher % of Shots Headed because that metric indicates that the player is probably more of an aerial striker and would pair well with the wingers that deliver high crosses.

Ultimately, productivity and style metrics tend to work together in a piece of analysis. However, in all cases it’s important to keep in mind whether a metric is indicating that a player or team is more productive (or “better”) at something or whether the metric is simply indicating that a player or team goes about their business in a particular style.