Yesterday, I read this tweet from Amanda Rykoff:
Lakers and Ohio State both with statement wins this afternoon. Very impressive.
This was the second time in about 30 minutes that I heard someone use that phrase "statement win" or, it's cousin, "statement game".
These terms get thrown around constantly, but are they meaningful? What determines if a game is a statement game? And what is the practical effect of a statement win?
Generally speaking, the term seems to mean games between two elite teams where the outcome either a) proves one team's quality over the other (informational), and/or b) makes it more likely that the winning team will beat the other team again later in the seasons--possibly in the playoffs (psychological).
I personally agree more with the first and less with the second.
For the psychological version to be meaningful we have to assume that losing to a team during the regular season can effect a team's psyche to the extent that it will cause them to play worse against the winning team in the future. To me, this is an empirical question that we should be able to answer with data. My guess is that there is no relationship between regular season records and playoff records head-to-head. Now, that might vary by sport, but off the top of my head I would guess any relationship is likely due to team quality and not the effect of a single "statement win". More often than not, the better team will win.
Until we actually run the numbers (and if someone has, please let me know in the comments section) I think we should stop using the phrase in this way.
What about the informational version?
This one, to me, has more merit, but it is also a bit problematic. You can't learn much from a single data point, especially in the sports world and especially in baseball, since we know how randomness can create counterintuitive outcomes during single games. For this reason, "statement game" is problematic, but "statement games" is workable.
Here's my logic:
1) All games are statement games in that the outcome on the field/court reveals additional information to us about the team. The more teams play, the more data we have on their performance, and the more data we have the picture of what that team's true talent level is becomes clearer. Single games can be misleading due to randomness and other factors. Therefore, quality information is only revealed over a collection of games. This is why it's hard to say the best team won in the playoffs and why Billy Beane famously quipped that "my sh*t doesn't work in the playoffs". Sometimes they do, but the regular season is a better gauge of overall team quality because we have more observations.
2) Moreover, games against other elite teams theoretically provide better information about team quality. These games are considered "hard cases" and produce "costly signals" about a team's talent level. Over enough of these games, the winning team sends a reliable signal about their talent level--since only a high quality team could routinely beat other high quality teams over and over again.
Given this, we need to think less about single games as statements and more about sets or collections of games. Single games can be terribly misleading in isolation, for all the reasons noted above. However, we can (and often do) segment by types of games and types of opponents and use those to draw better conclusions about the quality of a teams.