The Different Ways of Defying Team Projections by Jeff Sullivan March 10, 2015 This is something I’ve already shown you before. I mean, this, specifically, is not, but this is a slightly sharper version of the same graph. Team wins vs. projected team wins from the past decade: Right, so: the Angels exceeded their preseason projections by the most. The Cubs undershot their preseason projections by the most. By the projections, over the decade, the Angels should’ve won 48 more games than the Cubs. What actually happened was that the Angels won 150 more games than the Cubs. That’s pretty wild. We can also take this a little deeper. I requested a sheet of historical BaseRuns standings. BaseRuns aren’t projections — BaseRuns are founded upon what happened. Team win totals do fluctuate around estimated BaseRuns win totals, but you can’t really expect a projection system to be able to see that coming. So I wanted to break the first graph down. I wanted to look at BaseRuns vs. projections, and at actual wins vs. BaseRuns. Here’s the former. This might be a better measure of which teams have most outplayed or underplayed their preseason projections. By looking at BaseRuns instead of actual wins, you strip away a lot of noise, and relatively little signal. Anyhow: We still see the Angels in positive territory. We still see the Cubs in negative territory. But in this graph, we have a new leader and a new laggard. Here, we see the Blue Jays in front, outplaying their projections by 39 wins, or just about four per season. Right behind them, the White Sox, and then there’s a substantial gap between second place and third. At the other end, now we have the Mariners, separated significantly from the next-worst Cubs. The Mariners were projected to win 778 games over the decade — by BaseRuns, they should’ve won 730 games. That’s a good deal of underperformance. This factors in a lot of stuff that borders on unprojectable. Differences between BaseRuns and preseason projections can come from a lot of places, but mainly you’ve got surprising breakouts and collapses. You’ve also got surprising health and injuries. Take the White Sox, for example — in the past, probably thanks in large part to Don Cooper, the White Sox have enjoyed tremendous pitcher health, relative to the rest of the league. The Cubs, on the other hand, have had worse health. The Mariners experienced both injuries and collapses. All projection systems try to account for injury risk, but no one nails it, because the regular season gives us an n-value of 1. I should also note, before concluding this paragraph, that projections never foresee midseason transactions. Sellers sell, and buyers buy, and though one shouldn’t exaggerate the impact these moves have, they count for something. The next graph looks at actual wins vs. BaseRuns wins. So now this has nothing to do with projections. This is about wins vs. estimated wins after the fact. This is where the Angels have really stood out. As shown earlier, the Angels did overperform projections, overall, but they also just overperformed their own statistics. The Rockies, meanwhile, have done the very opposite, and here you notice the Blue Jays at second-worst. While the Blue Jays have significantly overperformed projections, they’ve also ultimately underperformed in terms of wins, so a lot of it has canceled out. And there’s a very strong relationship here between the numbers in the graph above and team clutch performance. As you probably know, we keep track of a Clutch statistic for both hitters and pitchers. I examined the past decade and combined those numbers to yield one overall number for each team. The Angels have been by far baseball’s most clutch team over the whole of the past decade. The Rockies, Blue Jays, and Cubs have been the three least-clutch teams over the 10 years. It makes sense why this would lead to deviations from BaseRuns — good or bad timing when the leverage is high disproportionately swings the win expectancy. Comparing Clutch score to (Actual – BaseRuns) wins yields an r-squared value of 0.67 So the following shouldn’t surprise you. I’ve struggled before to find evidence of Clutch sustainability. I looked at the difference between actual wins and BaseRuns wins in Year X, and then I looked at the same difference in Year X+1, for each team. The resulting plot: It’s nothing. I mean, it’s a scattered cloud of neat little army-green circles, but all it tells you is this doesn’t appear to be a sustainable team trait. Teams tend to play to their BaseRuns. Teams that deviate from their BaseRuns generally don’t do it again the next year. While we’ve observed that some teams have deviated over the past decade, that’s to be expected over any fairly small sample. That which has happened is quite different from that which will happen. This is fairly well-trod territory. There are ways for teams to defy their projections, happily or sadly. You can have someone break out or collapse, although you generally don’t see those things coming. You can keep players healthy or watch them all take turns on the disabled list, and here there does anecdotally seem to be a hint of team control. Certain teams have been better about preventing injuries in the past, and when you don’t have to so often lean on depth, that means you get to lean on the players you wanted to get the most playing time. And there’s the clutch aspect. As far as I can tell, clutch performance is essentially random on the team level. But it can make a huge difference when you get it, or when you don’t. I don’t think there’s such thing as routinely and consistently separating yourself from your BaseRuns, since that’s what the evidence says, yet while timing isn’t everything, timing is a lot of things. Good timing might be the equal of good talent. Good talent’s more dependable, but life doesn’t always balance out.