Ben Lindbergh (75:50)
Here's today be stopl. Okay, so I idly mused. Well, it turned out not to be idle musing, I suppose, because I I made something of it. But last week when we were talking about the Mets kind of completing their extreme makeover, their off season remodel, and putting some of the finishing touches on the roster, I was wondering aloud whether they were historically anomalous when it came to being a pretty good team, being a decent team. Even though their season ended ignominiously last year. They were a winning team. They had a better than.500 record. And so I wondered what kind of precedence there might be for a team coming off of a winning record to turn over its roster to the extent that the Mets have, and also to seemingly be in a position to contend again and be a winning team again and maybe be better than they were last year. Because I figured it would be tough to do all of those things. And so I did some stat Blasting about this. Earlier in the off season I did a Stat Blast with the assistance of Michael Mountain and we talked about the Mets exodus. This was before the McNeil trade, but we shared some information about how rare it is for a team to lose a lot of long tenured players in a single off season. But this time I called upon another member of the fellowship of Stat Blast correspondence, in this case frequent Stat Blast correspondent Ryan Nelson, who actually has been off for a bit. He has been on a bit of a Stat Blast sabbatical, a statbaticle, so I'm happy to have him back in Blast. He is back to blasting and up to his old excellent tricks here. And he supplied me with some information which I have also turned into a piece published@theringer.com so there is a written version of this and links to various spreadsheets and data sources. As always. But here's what I have learned. Slash Confirmed. First of all, I guess the big takeaway is turnover. Just in the abstract, in isolation, all else being equal. Turning over your roster is not good. That tends to be bad. It tends to bode ill for you. Or it just tends to be associated with bad teams and that it makes all the sense in the world, right? Because if you've got a great team already, then you don't have to turn over very much of your roster. You can just bring it back and be pretty confident that you will be good again. That was sort of what prompted my musing in the first place, because it seemed like the Mets weren't really in that situation where they were bad and thus they needed to drastically reconstruct themselves. But if you graph this and chart it as I did in the piece, it is just a very linear sort of trend. If you graph roster turnover rate by your winning percentage or do it the other way around, a very clear trend and pattern here that the lower your winning percentage, the higher your roster turnover rate tends to be. Which again is probably not really blowing anyone's mind here or breaking any news, because, yeah, bad teams, they have more incentive to do more to get good again. If you're bad, then you don't want to just run it back with your bad team. You want to change things up. And in that case, change could be good, maybe. And if you're good, then why mess with success? So teams with losing records in year one have collectively had roster turnover rates of about 32%, whereas teams with winning records in year one have collective turnover rates of about 22%. Now, as I have also written and documented previously, you can take don't mess with success too far. You can become complacent. And there does appear to be a trend where World Series winners sometimes are overly fixated on just bringing back their World Series winning team. And sometimes that costs them. And at least the last time I looked at this, World Series winners tend to do worse in the year after the World Series than the World Series losers do, which I, and also Sam Miller in his own work, have attributed to a little resting on the laurels by the World Series winners or just a commitment to keeping the gang together as opposed to prioritizing upgrades or maybe just feeling like you can sort of sit on your hands because, hey, you just went all the way, so why wouldn't we just try to do that again? But sometimes that can come back to bite you. On the whole, though, the worse the team is in year one, the higher its turnover rate in year two. And the other way, if you look at year one winning percentage by roster turnover rate. So I broke it down into different buckets of turnover rate. If you know, know less than 10% of your roster is turned over if 10 to 19, if 40 to 49, if 60% plus is turned over. There's a. A clear change in your winning percentage too, where the lower the turnover rate, the higher the winning percentage. It is true that the higher a team's turnover rate, the better its winning percentage tends to be in year two compared to year one. So, for example, teams with turnover rates above 60% tend to improve the next season by about 20 points of winning percentage, whereas team with Turnover rates below 10% tend to decline by about the same amount. But that's just regression to the mean at work, because those high turnover teams tend to be terrible in the first year. And you take any group of terrible teams, they're more likely to be a little less terrible the next year. You take any group of great teams, they're likely to be a little less great the next year. Okay, so this is not really rocket science. And I should note that Ryan developed. That's a kind of clever method, I think, for calculating turnover rate, because there are all sorts of ways that you could calculate that, and I've done it differently in the past. But the way that he did it was to look at the total playing time across both seasons, just consecutive seasons, by the same team, and look at the plate appearances and the batters faced, and then just look essentially at how much of the playing time in those two seasons is produced by players who were present in both seasons. So that's the way that he went about it. As he described, instead of measuring what percentage of plate appearances batters face in season two were players from season one or vice versa, he did a dual season waiting. So of all playing time opportunities across both seasons, what share came from players who were present in both? And as Ryan wrote, this is symmetric and automatically penalizes, penalizes both losing big contributors from year one and adding big contributors in year two, and also considers players changing roles. Okay, so we've established the basics here, but what I wanted to know is if you do hold winning percentage constant and you just account for that, then in isolation on its own, is turnover still bad? If you can sort of disassociate it from the correlations with how you did before the turnover, does it still portend doom for you? And it turns out that even if you do adjust for all of that, it's still actually not a good sign. So I grouped all the teams into buckets of winning percentage, like 25 point spans, basically 300 to 325, 325 to 350, all the way up to 625 to 650. And within each of Those groups. I then compared the higher turnover half of the teams that qualified for that group to the lower turnover half of teams. So for instance, limit the sample to teams with winning percentages between.300 and325 in year one and then sort those teams by their turnover rate and compare the least stable half of those teams to the most stable half of those teams. And the then do that for every one of these winning percentage groups and in every category, every bucket. The high turnover team tends to do worse in year two than the low turnover team relative to their records in year one. So essentially, if all you know is a team's record in year one and its turnover rate and you don't really know anything else, you don't have projections or anything else, but if the, the records are equivalent between two teams, but one has a higher turnover rate than the other, then that's actually bad, then that suggests that that team will do worse relative to its performance in year one than the team that is coming off the same sort of season but had a lower turnover rate. So even if you account for that all else being equal turnover, it's still bad, I guess. And you know, again, it's because teams, they're skilled self evaluators, the teams that know they don't need reinforcements are less likely to go get them. Plus, team building tends to be a pretty painstaking process. It's just, even if you set out to do it, it's difficult to dramatically reconfigure a roster in a hurry over the course of a single off season and have those plans pay off and, and it all just, just comes together as it maybe seems to have for the Mets. So that's what I've learned. Turnover, not good. Not a good sign. However, I also developed lists of the highest turnover teams, the lowest turnover teams, et cetera. So the highest turnover team, to probably no one's surprise, it's the Cleveland spiders. The infamous 1898-1899 Cleveland spiders where basically the entire roster got, got transferred to another team that had the same owner and it was just an entirely different team. And as Ryan calculates it, that was an 89.6% turnover rate. So that's as high as it gets. The next highest, it's the 1917-18 Philadelphia Athletics, 82.6%. It's Connie Mack teams. That's like post initial Philadelphia A's fire sale. Then, then there's some like 19th century teams, the 1898-9 Cardinals, the 1905-6 Boston Braves, the 2014-15. Atlanta Braves 70% turnover rate 1945-6 Red Sox. You do get a bunch of like 1945, end of World War II. Of course, you know, they're often sort of extenuating circumstances with the extremes. And also you do see say the 1970s to 6 to 77A is another Charlie Finley famous infamous fire sale. So that's the high range basically almost 90% for the spiders, that's the extreme. But more of the leaders are more in the 70% or so. That's kind of the more conceivable scenario. The lowest turnover teams are in the low single digits. So 1.7% turnover. The White Sox 1904 to 1905, essentially the same team. The Orioles 1969 to 71.8%, the Pirates 2.3%, 71 to 72. So these are all ancient history teams. Not ancient. Some of those are barely divisional era, but a lot of early, early 20th century. And nothing really more recent than, I don't know, I guess the top 10 includes 1980-81 Reds at 4.7%. That was a strike shortened season in there. So this doesn't really happen anymore that teams are that stable. And that's because there is also an extenuating circumstance for the Mets here there's an era effect because there is more roster turnover these days than there used to be. And this is another thing I graphed. And again, probably something that won't shock you really, and there's been some previous research that has shown this, but the roster turnover rate rate from year to year on a team level by decade has ticked up significantly in the 21st century. So if you graph it, in the 20th century, there's a little spike. In the 1910s there was the Federal League, you had people jumping among leagues in the 40s, of course you had World War II, there was a spike. But generally up until like the late 80s, it was in the low 20% turnover mean or median. And now it has really jumped up. So it wasn't, you know, the second they flipped the switch on free agency, it took about a decade for, for that to kind of the full effects to be felt. But as of the 90s, the roster turnover rate jumped up over 30%. And in this decade, this half decade that we have seen so far, it's up to about 35%, almost 35%, which would be a new high. And maybe part of that is pandemic and lockout and who knows what else. But still it's a long term trend. So the Mets turnover seems a little less anomalous given that it's in this era where we have become accustomed to two more roster turnover. But even by the standards of today, it's still a ton of turnover for them. Now here's kind of the the money maker of of tables, the highest turnover teams with winning records in year one. So that's the Mets are maybe aspiring to join this group because they are coming off of a winning record. And well, it's still the Cleveland Spiders because they did have a winning record in 1898 before that team got destroyed. And then it's the 76A's before Charlie Finley broke it up. And you can see the whole tables and spreadsheets and it's all linked. There was actually the 2001 Mets are on this list too and I mentioned them early in the article because that's kind of a negative precedent because 2000 they made the World Series. 2001 they missed the playoffs. And then 2002 they were very busy and Steve Phillips tried to get them back there by bringing in Roberto Al Mar and move on and Jeremy Bernitz and Roger Sedano and a whole bunch of pitchers and it just did not work. They finished last in the NL east and that's sort of the precedent for winning the winter, not necessarily equating to winning. But here's the the takeaway. Only 19 teams with a winning record in year one have turned over at least half of their rosters at least a 50% or more than a 50% turnover rate. Of those 19. So it's already a small group. Only two, the 2003-4 Yankees and the 2023-24 Padres improved their winning percentages the following year. And even that there's kind of a caveat because the the 2003-2004 Yankees. That's an interesting one. I mentioned this in the earlier stack last this off season with Michael Mountain and the Exoduses. But they lost arguably the most single season WAR of any team ever. And yet they won the same number of games the next year. So they lost a ton of war from 2003 and then in 2000, 2004 they had also 101 wins didn't phase them whatsoever. And that was the era when, you know, the Onion had headlines about the Yankees signing every player and they were vastly outspending everyone, which helps and it's helping the Mets these days too. They only qualified cause baseball reference counts a tie for the 2003 Yankees because there was this weird suspended game that was official and it was five innings and then they replayed it another time. But but it it kind of counts in baseball references figures. Anyway, the 2023-24 Padres are a good recent example of doing this, and in their case, of course, they were extremely unlucky in 2023 and we were all wondering why aren't they better? And they have the underlying numbers of a better team and then finally that did show up in 2024. So it has happened, but it is very, very rare. Now, one minor drawback to this method that Ryan concocted is that we can't actually tell what the Mets turnover rate is until they play this season, and we see how the playing time shakes out and everything. But based on their projected playing time via the fan graph step charts, I calculated their projected turnover rate and that was 43.2%, which is significantly lower than turning over half your roster. But I would imagine that the final figure is likely to be higher than that because as great as Jason Martinez and co are at prognosticating playing time, they can't anticipate future acquisitions and injuries and that's going to lead to extra and anticipated turnover. So I would guess that the Mets turnover rate will end up higher than that. But even if they have a 43.2% turnover rate, of the 1361 qualifying teams with winning records in year one one, only 57 of them, 4.2%, had a turnover rate that high, and of those 57, only 13. So 22.8% of that tiny sample had a higher winning percentage in year two. So it's a tall order, is what I'm saying, that it is actually pretty darn uncommon, as I was musing out loud on that episode, for a team to turn over a lot of its roster coming off of a a superficially successful season. You certainly wouldn't call the Mets 2025 successful. But it looked good for a while until those last couple months, until the final standings even even the last day of the season, you know, they were in playoff position except for that pesky tiebreaker. So yeah, it's rare for a team in that position coming off that kind of season to decide to do an overhaul like the Mets have done and for it to work well enough that they are good again and actually improve their record, which the Mets are certainly setting out to do here. Because if they don't improve their record from last year, then this will be a big failure, right? And, and they will almost certainly miss the playoffs again. So they need to be better and make the playoffs. And historically speaking, going by, I don't know, a century and a quarter or so of hot stove history, there's not a ton of precedent for a team doing what they're doing. And as I noted on that episode, the. The Yankees just across town, quite a contrast. Cause their projected turnover rate is less than half of the Mets, 17.1%. And yet the funny thing is that their projected team more totals are almost identical. They're separated by less than one win above replacement. And I looked up what the projected team WAR totals were one year ago. So just last January. And both the Mets and the Yankees of this year, their projected totals currently are within a win of where they were last year. So the Yankees are basically bringing back last year's roster. Maybe that makes some sense. But the Mets, they have done this whole dance and they have changed everything. And yet the projected WAR total is almost identical to what it was one year ago, which.
