Now that the deadline has passed, what did we notice?
First, let’s check how our model did (note that we’re counting every trade in the true deadline period, starting with Aroldis Chapman to the Rangers):
- Number of trades: 57
- Trades accepted: 52
- Acceptance rate: 91.2%
- Variance: 2.6
Our model nailed the two biggest trades of the season, Scherzer and Verlander.
It also got most of the medium-sized and smaller deals right: Civale, Montgomery, Flaherty, Candelario, Barlow, Chapman, Robertson, Santana, Urias, Yarbrough, DeJong, Hand, Velazquez, Pollock, Canha, Sewald, Hicks, Cron/Grichuk, Rosario/Syndergaard, Patino, Hernandez, Floro/Lopez, Johnson, Cabrera, Gott, Fujinami, Lopez/Hearn, Pham, Hedges, Middleton, Cooper/Weathers, Leone, Sampson, and a bunch of smaller ones.
It was a bit stretched on a few, which were technically accepted as overpays but debatable: Falter/Castro, Chafin, Lorenzen, Hill/Choi, Peterson, Graveman
And it missed on these: Giolito, Lynn, Moll, Burger/Eder, Bell/Segura
As with past deadlines, this pattern is consistent. It’s a bell curve: the 46 wins fit pretty clearly in the middle; the next 6 fit in a tier outside of the middle, and the remaining 5 are outliers.
What that tells us is that our model is doing its job. It’s never been perfect, but then, no model in history ever has been (otherwise it wouldn’t be a model). But as always, we think we can learn from each trading period and tweak it a bit to be even more accurate. So what can we learn from the trades that weren’t accepted?
Giolito and Lynn: These two were cases where the acquiring teams jumped the market, trading for them about a week ahead of the deadline. Because of the pronounced sellers’ market this year, they would have commanded overpays if their team had held them until the deadline, but less so. Here, the buyers paid a major premium by grabbing them early, as the sellers rightly demanded it. In Giolito’s case, that meant the Angels gave up top prospect Edgar Quero; in Lynn’s case, the Dodgers gave up solid pitching prospect Nick Nastrini. We don’t see a reason to adjust the model for these two outlier cases.
Moll is another instance among many where the A’s model differs from ours. They consistently seem to get poor returns from their trade assets, and not only at the time of the trade, but after – the returns from previous deals have not borne out what they’d hoped, suggesting that our model may be more accurate than theirs. The prospect they received in this deal, Joe Boyle, is lauded by some experts for his raw stuff, but he has severe control issues and is clearly a project, whereas Moll is a consistently solid lefty reliever with several years of control. Again, we think this one is just an outlier.
Burger/Eder: we’ll eat this one. Burger was probably too high, as his profile is driven purely by home runs – we may need to tweak our model to even out this sort of player; and we know Eder may rise up prospect lists once he gets a few more starts under his belt after being out with TJS. This one may look more fair in time, but we have work to do on modeling the Burger types.
Bell/Segura: This was a case of trading bad contracts. We know infielders who can’t hit have little value to the market, so we probably should have discounted for that even more in Segura’s case – he’s never been much of a defender, so once his bat deteriorated, there was nothing left (the Guardians released him afterwards). We presume Miami valued Bell’s bat more than most teams, and also wanted to rid themselves of prospect Khalil Watson, who has had discipline issues. So not much to learn here.
You can buy prospects!
The Mets, using Steve Cohen’s money, did just that by sending boatloads of cash to both the Rangers and Astros, to cover not only the underwater portions of the respective contracts of Scherzer and Verlander, but even more to effectively buy Luisangel Acuna, Drew Gilbert, and Ryan Clifford. If you didn’t believe in the concept of surplus value before this, it’s proof-positive that it works. These two trades came up almost perfect in our model, which also validates our valuations of the prospects.
Why don’t other teams do this more? I think it’s because Cohen is a seasoned trader in financial markets – he’s used to cutting his losses and not thinking twice about it, and understands the concept of sunk cost. He used the money to reshape his team’s portfolio, in a way, from older, expensive declining players to younger, cheaper improving ones.
We often say the money matters when evaluating player values, and some people prefer to ignore it and disagree with us. After all, it’s not their money. But if you want to follow the trade markets the way front offices do, it’s important to respect the power of it.
Starting pitchers required premiums
Duh. This was obvious to everyone paying attention. I even wrote an article about it a month ago, anticipating it. Still, we heard from lots of folks telling us there was a supply/demand imbalance – too many contenders needing starters, not enough on the market.
So why didn’t we adjust more for that? Two reasons: 1) The market tends to even out closer to the deadline, as more teams decide on their path, and as both parties feel the pressure of the clock. The former happened a little, as the Mets jumped in as big sellers, but not enough. 2) We’ve learned not to put our thumb on the scale – our model is pretty good, and if we introduce too much subjectivity it can sometimes make things worse. Trust the numbers, in other words. Still, given the obvious imbalance, I think we should have adjusted a touch more for this market.
The one exception was back-end arms, like Ryan Yarbrough and Bailey Falter. Our model was fine on Yarbrough, but a bit high on Falter, who should probably be modeled more as a long reliever going forward.
Notice who didn’t get traded, after all the reports died down? Dylan Carlson. Teoscar Hernandez. Adam Ottavino. Tim Anderson. Jose Cisnero. Chasen Shreve. There are more.
What do they have in common? Little expected impact. At the deadline, contending teams want upgrades – players who can improve the roster, not just add to it. Carlson was reportedly shopped heavily, but his below-average 89 WRC+ probably scared away most buyers. Teoscar is striking out over 32% of the time, and has produced at replacement-level this year. Tim Anderson, despite a good recent stretch, has a 60 WRC+ and -0.5 fWAR. No one wanted these guys.
Instead, the hitters who were moved were the ones having better years – Candelario, Canha, Santana, et al. There weren’t many of them, though.
Relievers are still tricky
Our reliever model did better than in years past. It may not ever be perfect, but it’s closer. The closeness of the Lopez/Floro trade kind of made that point on its own. In the past, our inputs were a little too volatile, causing higher-variance misses despite a few wins. So we changed a few things, and now the inputs are steadier, and the results are too.
Interestingly, most of the relievers traded did not require overpays. This was true of the middling ones – Barlow, Hand, Middleton, et al (and, oddly, Sewald). Hicks and Robertson were minor overpays, as befits their market perception of being higher-impact.
And again, notice who didn’t get moved – the controllable ones. David Bednar’s price is very high, as we have it, and teams may be reluctant to pay that much for a reliever (even one as steady as him). It’s curious that Kyle Finnegan stayed put, as the Nationals wouldn’t seem to be competitive anytime soon, but again, the price may have been prohibitive.
What happened with Eduardo Rodriguez is beyond our scope of expertise here. Lots of talking heads are speculating about why he veto’d the deal to send him to the Dodgers. From our perspective, it’s yet another point that underlines how tricky his case is – will he opt out or won’t he? If he doesn’t, will the Tigers get stuck with a risky contract? There are clearly non-baseball issues here. But even so, one would have hoped the Tigers’ front office had had a backup plan. They wasted a prime opportunity to cash in on a tradeable asset.
It was a great deadline for us. And one that, at first, didn’t seem all that exciting – what we thought might be forgotten years from now as “the Giolito deadline” turned into “the Scherzer/Verlander deadline.” Thanks, Steve.