Big Mo: Do Blowouts Impart Momentum?

Does momentum exist?

No, I’m not casting doubt on a fundamental tenet of physics: David Kagan can breathe easier. I’m casting doubt on a well-trod talking point of baseball (and really of all sports). A team that’s been doing notably well recently is said to have momentum, and we expect it to keep doing well, at least for a while. The opposite effect is perceived for struggling teams: Fans expect them to keep stinking up the joint.

Unsurprisingly in our analytical age, this notion has not gone unchallenged. Tom Tango, Mitchel Lichtman, and Andrew Dolphin looked at individual players’ momentum in The Book, and found it to have just slight predictive value. Only a hot pitcher could expect to keep it up the next time he went to the mound.

One can take a step back and look at the matter on a team level instead. Momentum might show up better for a combination of 25 players than for one. Here, though, even the old school evinces skepticism. “Momentum,” any old timer who looks like he misses his chaw might tell you, “is tomorrow’s starting pitcher.” Granted, that’s probably said more when the old timer’s team has been getting knocked around lately.

From our experience examining clutch performance, one suspects momentum will vanish once we look closely at it. Psychological effects seem to do that. Still, to get a result, even an expected one, we need to do the looking, so I’m going to look at team momentum in baseball. I’ll do it through the crudest, bluntest event I can: the blowout.

It seems self-evident that a rout of an opponent should imbue a team with confidence that at least carries through into their next game. Likewise, suffering a trouncing should depress a team and weaken its prospects for success in the immediate future.

The operative word so far has been “should.” Is there a “does” beneath it?

Setting Boundaries

What constitutes a blowout? There’s no obvious answer, so I’m fairly free to choose my own. I settled on a 10-run margin as the criterion. I was tempted by nine, one run per inning, but worried that I might accidentally include forfeits (which officially end 9-0) in the group and went one run higher instead.

To see the full immediate effect of the blowouts, I had a requirement that the two teams involved would play each other again the next day. Waiting a day, or switching opponents, would diffuse the effect. I did expand this requirement to include same-day rematches, as in the nightcaps of doubleheaders. One would think the psychological effects of the blowouts even stronger in such circumstances.

My data set stretched 20 years, from 1998 to 2017, the era of the 30-team majors (not counting the 2018 season, which was in progress as I did my research). This decision made my reasoning for the minimum margin of blowouts look a touch silly. I had specifically excluded the chance of a forfeit being included, but the last forfeited major-league game was in 1995. It’s all right, though: I still got plenty of games to work with. To be precise, 1,183 of them.

Sidebar

Admittedly, 10 runs does not mean the same thing all the time. The run environment of the league changes, and the run environments at specific ballparks are different. Before I make conclusions based on the convenient assumption that all blowouts are created equal, I should give a little information based on the fact that they aren’t, quite.

When the league is in a higher-offense era, blowouts should be more common. You can judge this by other sports: How much more common is it to see a 10-plus point win in football, or basketball, than baseball? (On the flip side, how vanishingly rare is it to see a 10-goal blowout in soccer, or hockey?) The blowout data I gathered shows this, if imperfectly. Do remember that I didn’t collect all blowouts, just all blowouts with a game between the same two teams the next day or later that same day.

The chart shows blowouts (the ones qualifying by my criteria) tracked against runs per game in MLB that particular year. The trendline, which really ought to be a flattening curve, shows blowouts rising by about 30 per year for each run added. The R-Squared of 0.2904 means close to 30 percent of that rise comes from the rising run environment rather than other factors. One more run per game would thus mean about nine more (qualifying) blowouts per season, if we look at it through that prism.

As for the offensive environments of ballparks, there is one venue we should expect to host a profusion of blowouts: Coors Field in Denver. The most offense-friendly field in the game seems a lock to be the most blowout-conducive field also. I did the count of where the qualifying blowouts were played, and found it wasn’t quite the case.

The Rockies were tied for the second-most qualifying blowouts being played in their home ballpark, with their 52 matching the Detroit Tigers and Boston Red Sox. The Red Sox aren’t really a surprise, with Fenway Park well-known for favoring hitters, but Detroit and its only slightly offense-leaning Comerica Park is. Topping all of them, with 56, are the Texas Rangers at Globe Life Park, a venue at least as offense-friendly as Fenway, if not in Coors’ class.

To complete the glimpse at park effects, the team with the fewest (qualifying) blowouts in the last 20 years is a definite surprise. Miller Park is one of baseball’s better offensive locales, yet the Milwaukee Brewers have played in just 14 blowouts there the last two decades. (Second fewest, at 21, goes to San Diego and its much more pitching-dominated Petco Park.) This perhaps can be explained partially by the weird variances possible in taking a sub-set of an already esoteric set. Still, there’s something odd happening there, and I wish I could figure out what it is.

Back on Topic

When a team defeated an opponent by at least 10 runs, then played them again the next game, the winning team had a 651-532 record in those rematches. That’s a .5503 winning percentage, which looks like it argues for a meaningful momentum effect. The question, however, is more complicated than those simple wins and losses.

If one team blows out another team, it’s a reasonable guess that the first team is better than the second. It doesn’t give away too much to say this turns out to be true for my data set. Also, since home teams have an advantage over visitors, one could expect the majority of blowout winners to be home teams, an advantage that would carry over to the next game by the rules I laid down. This also turns out to be true.

This means I need to figure out how much of the post-blowout advantage comes from the blowout winners being generally better teams, playing more often than not at home.

I tallied up the season won-lost records of all the trouncers and the trounced. (Going by Pythagorean records might have been slightly more accurate but would also have been a great deal more digit-pushing to look at something that might wash out over a thousand team-seasons.) The blowout winners had a mean winning percentage of .51930; the losers an average of .48248.

To determine what results we should have expected from these games, I went to the venerable Log-5 equation. The blowout winner’s expected winning percentage against the loser is (W-WL)/(W+L+2WL), where W is the blowout winner’s overall winning percentage and L the loser’s winning percentage. Plugging in the averages doesn’t give us the exact result that finding the average of all 1,183 games would, but it will be reasonably close*. The result is an expected .53677 winning rate.

* Using the averages probably produces a marginally higher number than going singly. The single games would have many wider gaps of winning percentage, leaning toward the blowout winners. Those wide-gap games would be further along on a curve that flattens as it approaches 100 (or zero) percent, lowering the effect of each increment of widened gap. That means on-paper mismatches would, taken together, slightly suppress the expected winning percentage of the superior team, as opposed to taking the average.

Looking to home-field advantage, 615 of the 1,183 blowout winners won at home, a .5198 percentage. For the 20 years covered by this study, home teams in MLB won 53.89 percent of their games. Apply this to the blowouts*, and we find the blowout winners would be expected to win 50.154 percent of those games by home-field advantage, or disadvantage, alone. This tiny edge may feel wrong, but it’s a matter of a mere four percent plurality of home-field games gaining a four percent advantage in winning percentage for playing at home.

* I could have broken this down by year, using the annual home-field advantage numbers. It’s not clear, though, whether this would have been tracking actual yearly changes in the advantage, or Brownian-motion wobbles around a mostly steady mean. I confess I was relieved to take the short cut.

Putting the two influences together in a simple additive way, 3.677 percent plus 0.154 percent, we find we should have expected the blowout winners to have a .53831 winning percentage the next game. That explains more than three-quarters of the .5503 percentage that actually happened, almost all of it due to the generally superior records of the blowout winners.

What about the remaining 1.2 percent or so? Is a residual .512 winning percentage proof of some kind of momentum effect? Perhaps, but there might also be an effect my imagined grizzled lifer would recognize. That is tomorrow’s starting pitcher, which depends on today’s starting pitcher.

It’s highly plausible that blowouts happen more frequently, not only when good teams are pitted against poor teams, but when better starting pitchers are pitted against lesser ones. Some of this will follow from the pitchers being on better or worse teams, but certainly not all. Rotations have their number ones, and their number fives, or sixes, or sevens.

If it’s likelier that an ace will win a blowout against a back-ender than vice-versa, that has implications for the teams’ next game. The team coming off the blowout win will be pitching someone other than the ace; the previous loser will start someone other than the rotation-filler. This should give the blowout losers an improved chance for a bounce-back win the next time out, thus lowering the expected winning rate of the blowout winners. If we don’t see that, it suggests some countervailing force…such as, perhaps, momentum?

There’s also an effect coming from the bullpens. The blown-out team probably will have made heavier use of relievers than its opponent. Even if that comes from low-leverage hurlers, it limits the team’s options in the following game. (They could counteract that by calling up a fresh arm from the minors, but there’s a reason that fellow was in the minors to begin with.) That should dampen the loser’s prospects for the rematch, at least partly balancing out the advantage I posited from the starters.

Conclusion

With the effects of starter and reliever usage working against each other, and very hard to quantify in any case, we are left with the previous results. The differences in overall records between the blowout participants explains over seven-tenths of the winner’s success in the following game and home-field advantage a tiny bit more. That’s a large majority of the effect, but not all. For those believing psychological momentum has a tangible impact, there is still room for it to operate.

That room could be closed off in other ways, though. Perhaps in-season variations in team strength explain the remaining advantage. A couple injuries to key players, a breakout or two that are happening or haven’t yet, and a matchup could be more lopsided at that time than the rest of the season would indicate. If a deep, deep dive into the numbers was done, we might find that effect, but for now it’s speculative.

For now, there is still a narrow indication that momentum may exist on a team level. It’s not a strong indication, adding up to about .012 of winning percentage in something close to an ideal short-term situation. It could dwindle or vanish altogether under closer scrutiny, as similar psychological notions in baseball have done before.

So if you like the idea of momentum in baseball—and I guess I’m such a person, thinking baseball players will behave more like human beings than random number generators—enjoy the minor effect of blowouts while you can. The next hard look might send it the way of clutch performance.

Postscript

While I was writing this piece, my occasional collaborator Paul Golba scored free tickets to a Yankees game. They had walloped Toronto 11-0 the day before, and I humorously instructed Paul to take careful notes on whether New York kept up the momentum. He did: The Yankees lost 8-7.

Some time later, the day before wrapping up my writing, the Boston Red Sox opened a doubleheader against Baltimore with a 19-3 shellacking. Given the records of the two teams involved, this seems almost the expected result. The nightcap, a dominating 10-3 Baltimore victory, was not nearly as expected, except perhaps as a salutary reminder that in baseball, on any day any team can beat any other team.

I’m not saying these two incidents prove anything. For me, they were more a way to take a subject that was becoming just a mass of statistics and anchor it again in the game played on the field. That’s always worthwhile.

As an ironic counterpoint to the hypothesis I was examining, they were worth a chuckle as well. That’s worthwhile, too.

References and Resources

• Baseball-Reference
• Tom Tango, Mitchel E. Lichtman, and Andrew E. Dolphin, The Book
• The Bill James Handbook 2018 for ballpark scoring indices