Trade deadline analysis: how did our model do?



Overall, we were pleased with how our model did during this trade deadline period. Every trade except one was accepted by our system. On two very prominent trades (Clevinger and Marte), the numbers matched up very well, as they did on almost all the others. A few cases were borderline, and there was only one miss. That’s a healthy distribution.

Our scorecard since August 2019:

  • Total trades: 98
  • Total accepted by our system: 93
  • Acceptance rate: 94.9%
  • Average margin of error: 1.5

  • Key takeaways

*As we expected, it was a seller’s market. With so few teams out of the hunt, those sellers were better able to dictate the terms and prices for their trade chips. Still, most of the deals seemed reasonable from our model’s perspective.

*The media called the Clevinger trade a haul. That’s because it was a nine-player deal, with six going to the Indians. But those six pieces were all mid-tier players, whose value added up to one premium player, so we see it as a fair deal.

*The Diamondbacks threw in the towel and became sellers. Reports are that they need to cut budget given pandemic losses, which, along with their disappointing season, drove the need to unload guys like Starling Marte and Robbie Ray. With Marte, they got a fair deal — Caleb Smith’s number matched Marte’s exactly; Humberto Mejia and the PTBNL (reportedly Julio Frias) were throw-ins.

*Padres GM A.J. Preller is under pressure to win now. His boss, owner Ron Fowler, gave him an ultimatum to win, or he’ll lose his job. So all his trades were motivated more by that then a sense of value. He was willing to take a big gamble (and big value hit) on Austin Nola in the hopes that Nola’s late-career breakout is real, which will help Preller and team accomplish that goal. Notably, Preller is constrained by his owner by budget pressure as well. He can’t just trade for an expensive (in salary terms) superstar, because the team can’t afford to pay that. So he had to target low-budget impact players like Nola. Further, he was also trading from excess: he can’t fit all those prospects on his team; many were blocked, and some needed to be added to the 40-man roster to protect them from the Rule 5 draft. They were burning a hole in his pocket. So spending a big chunk of that capital made sense in this circumstance.

*To a lesser extent, it seemed to us that Mets GM Brodie van Wagenen is also desperate to save his job, given that the team is about to be sold. He made some late trades that, from a value standpoint, don’t look great. Robinson Chirinos, Todd Frazier, and Miguel Castro might or might not help much (the first two were salary dumps on the Rangers’ part). And although he didn’t give away much to get them (two PTBNLs and low-mid-level prospect Kevin Smith), it felt to us like he was grasping at straws.

Smaller trades: fungibility

We consider players in the low ranges (say, 0-3) to be quite fungible. At this level, any value difference is relatively meaningless. This is because, in the grand scheme of things, all teams have large asset bases, and these are rounding errors. If your total prospect capital was $207M (just making this up) and you traded for a guy whose value was $1M less on paper, your total capital is now $206M. You probably don’t care about that, because the difference is couch cushion change. What matters more is that you filled a need. In the Angels/A’s trade of Tommy La Stella ($2.5M estimated median value) for Franklin Barreto ($1M), the A’s filled a short-term need and the Angels took a flyer on a potential long-term asset. The fact that La Stella is worth more at this moment is just a function of time. He’ll be at zero at the end of the season, while Barreto will probably stay at his level. So you could say the A’s got the better deal today, but the Angels will in a month. It’s fungible.

Relievers are volatile

We’ve noticed that, when it comes to relievers, teams tend to take a shorter-term focus. They all know that bullpen arms are volatile — often wildly so. So they go with a hot hand. For that reason, we’ve skewed our reliever model with a bit more bias to recent results. But that creates some volatility in the numbers; and sometimes even that isn’t recent enough. When we updated our numbers just last week, Miguel Castro’s numbers were pedestrian. In the interim timeframe, he pitched better, and perhaps that was enough to interest the Mets to take a flyer on him. He’s always had good stuff, but has been maddeningly inconsistent. But they paid a higher price for him than we would have thought, given his spotty track record and lack of options.

More on Nola

In our process, we lean conservative, as does most of the baseball industry. Small sample sizes are generally considered landmines — you have to treat them very carefully. The cutting room floor is littered with SSS breakouts, who turned into pumpkins once they regressed to their means.

So all the projection systems — PECOTA, ZiPS, Steamer, et al — were skeptical of a 30-year-old journeyman who’d been signed as a minor-league free agent by a rebuilding team. Rightly so, because 99% of the time, those guys never pan out. They’re quad-A org filler. But every once in a while, you get a guy who changed his approach and turns into Max Muncy, or more recently, Mike Yastrzemski. The Padres are thinking a step ahead, and want to believe Nola is one of those as well. And since he’s making the league minimum in salary, it’s a good combination: impact performer who doesn’t cost anything (in salary, at least).

What he did cost is prospect capital. Credit Seattle GM Jerry Dipoto for taking advantage of Preller’s situation. I’m sure Dipoto is giggling gleefully at the haul he got for a guy he signed as a minor-league free agent last year. (Dipoto seems to have a nose for these situations, as he also took Brodie van Wagenen to the cleaners in the Cano/Kelenic deal.)

One could argue that Preller didn’t give up that much. It’s true that Taylor Trammell’s stock has fallen a bit, but we’ve already taken that into account — and respected prospect evaluators such as Baseball America and Fangraphs’ Eric Longenhagen still believe he’s a 50 FV prospect — a 2-ish WAR regular. Andres Munoz has a lightning arm that, once he returns to health, could turn him into the Mariners’ future closer. Ty France is the kind of guy who gets lost in the shuffle, but all he’s done is hit at every level — he’s been on a path to being at least a 1-WAR utility guy, with a fair amount of certainty.

But even though these guys have value in the conceptual sense, Preller could afford to move them because he’s trading from excess. He has a glut of prospects and post-prospects on his roster, so it’s not terribly painful to him to part with them in order to save his job, please his boss, and win now. That’s what this trade was.

More on Clevinger

Why the focus on quantity instead of quality? The Padres were moving excess capital; the Indians were dangling a guy who had violated their trust and culture. Other teams knew that, so perhaps they didn’t have that much leverage. Add to that the fact that Clevinger had not pitched particularly well this year, on top of a knee injury before that, and you have something of a damaged-goods case. Granted, he still has a strong track record, which carried most of his value. And from Cleveland’s standpoint, most of these players are already at the MLB level, which means there is less uncertainty in their valuations.

And finally: Minor and Ray

This is a tale of two former top pitchers (and trade targets) who were not pitching all that well in their walk years, reducing their value as time went on. Minor got a bit more — in a deal we thought was quite fair — but Ray, even after we downgraded him multiple times this year, still went for virtually nothing but salary relief for Arizona. One of the flip sides of the deadline is that if you have a less attractive asset, you’re sometimes forced to take whatever you can get before the clock expires.


About the Author

John Bitzer

John Bitzer

Founder and editor of
  1. the statbook

    Wait, you got called out last year for fudging the numbers on players to force a trade through your system (i.e. Marcus Stroman trade), you’re now taking credit for being right a bunch of times.

    Proof: Read the comments.

    Your response also didn’t pass muster. Instead of saying, “we think it’s a bad deal and therefore, we recognize it won’t pass”, you changed the numbers to force it to fit.

    This leads me to a couple this belief:

    You don’t have confidence in your own numbers. You acknowledge “information gaps”, but the truth is, sometimes it’s not an information gap to not know how a team like New York values Stroman, or how Toronto valued Anthony Kay. It’s okay to disagree. I’d imagine some GMs thought New York underpaid, some might feel Toronto got a steal. If you’re confident in your values, you can stand them up and disagree and be fine.

    Why didn’t you just chalk it up to disagreeing on value? Why didn’t you just say, “here’s what we believe about Marcus Stroman; here’s what we believe about Anthony Kay”? Instead, you lied by changing the numbers after the fact. How many other trades did you do this on?

    I’m all for changing the numbers based on new information. But when the new information is, “well, the trade didn’t work in accordance with the numbers we gave”, that’s not new info. And when you just change the numbers hoping no one will notice, it makes articles bragging about your success a little insulting.

    For what it’s worth, Stroman’s value was at ~29M the morning of that trade, it dropped to 18M, and Woods-Richardson’s value also rose. John just thought no one would notice.

    • John Bitzer

      Thanks for your comments. Your points are fair. That situation was a mistake, and we learned from it. We weren’t trying to hide anything, as that would be impossible, since our history is readily available for anyone who wants to look. The thinking at the time was to incorporate market data, but yes, it was too soon, and came across as dishonest. No model is perfect, and each time we’re off, we try to ask hard questions and learn from it. That has made our model stronger and the values more accurate. And we own it one way or the other.

      • the statbook

        Perhaps a way to prevent this going forward is to just balance out a trade using the “add cash” feature. It might not reflect the actual trade as recorded, but it allows readers to interpret the trade through the values you’ve given.

        I think it would have been better to see the Stroman trade as the Mets getting 29M in value and only giving up around 5-6M, based roughly on the numbers (if I recall them correctly). It allows us to understand how you value players and what things you observe about them.

        Then, AFTER THE FACT, you can acknowledge where the numbers may have been wrong; what new information was gleaned; and how that’ll affect numbers going forward.

        I’d appreciate more to see the trade reflect an overpay or underpay rather than a force-fit.

        • John Bitzer

          Agreed, and that’s the way we’ve been doing it for the past year+. The Nola trade is an example of that. Thanks again for your comments.

Submit a Comment