Building a Consensus Top Hitting Prospect List by Peak Projected OPS+
As we’ve seen with the FanGraphs Depth Charts and ATC, averaging projections from multiple systems is a common approach for improved accuracy when forecasting performance for the current season. This article applies the same “wisdom of the experts” aggregation logic to combine peak projections from various systems — in this case, ZiPS, Clay Davenport’s projections, and OOPSY — to build a consensus top hitting prospects list.
Overview
Earlier this offseason, I published OOPSY’s top hitting prospects by peak projected major league wRC+. You can check out the article for a detailed explanation of the methodology, but the short version is that peak projected wRC+ is essentially a 2025 projection, except with extra aging added to it in order to forecast how good each prospect will be at their (late-20s) peak.
Dan Szymborski’s ZiPS and Clay Davenport’s work also provide projections that make use of aging curves, major league equivalencies, and regression to project future performance for prospects. Projecting baseball players involves many choices about what to include and exclude, and how to weigh or account for different things. For instance, every system uses its own flavor of aging curve and major league equivalencies. ZiPS uses aging curves specific to each type of player, while Davenport’s system and OOPSY apply a more general aging curve across players. Unlike OOPSY, ZiPS and Davenport’s projections also capture college performance, so I tend to trust those systems more for the 2024 draft class given that those players have little, if any, professional data. OOPSY also offers its own unique methodological quirks. Given the methodological diversity across projection systems, averaging them tends to be an accuracy-enhancing approach.
You may have seen the annual ZiPS Top 100 Prospects list that was published yesterday. That list focuses on overall prospect value, including defensive value, whereas this article focuses exclusively on offense. Dan and Clay were both kind enough to provide me with peak hitting projections for this piece. To rank the prospects, I took an average of their peak OPS+ projections from the three systems, as this was the easiest metric to compare across systems, even though FanGraphs is generally a pro-wRC+ publication. OPS+ is a slightly less refined version of wRC+, and is on a similar scale — a 100 OPS+ is league average, with each point above or below 100 representing one percentage point above or below the league average. My peak wRC+ and peak OPS+ projections are all within four points of each other. I had to make some minor tweaks to harmonize the Davenport data so that it was on the same scale — rescaling Clay’s EQA metric to OPS+ — so you should consider me solely responsible for any poor decisions made en route to the final output. Players without stateside professional experience were excluded.
Without further ado, here are the consensus Top 50 hitting prospects by peak projected major league OPS+:
A few other players merit honorable mention, as they all have at least a 110 consensus peak project OPS+:
- Josue Briceño
- Sebastian Walcott
- Demetrio Crisantes
- Mike Boeve
- Agustin Ramirez
- Dylan Crews
- Deyvison De Los Santos
Some Stray Observations
Looking at the 300 players with a projection across all three systems, the correlation between ZiPS and Davenport is .68, versus .71 between ZiPS and OOPSY. The correlation between OOPSY and Davenport is .84. The systems assume a similar average player and agree that the current crop of baseball’s top prospects project for around a 130 OPS+ at peak.
Emmanuel Rodriguez, Nick Kurtz, and Samuel Basallo ranked first for ZiPS, Davenport, and OOPSY, respectively. The consensus top four — Emmanuel Rodriguez, Basallo, Lazaro Montes, and Roman Anthony — are the only prospects to rank among the top 10 for each of the three systems.
OOPSY’s top five prospects, in order, are Basallo, Jasson Domínguez, Montes, Kristian Campbell, and Anthony. Relative to the other two systems, OOPSY is most bullish on Kevin McGonigle, Domínguez, Moises Ballesteros, and Matt Shaw. Conversely, it is least bullish on Travis Bazzana, Christian Moore, and Kurtz, but as I don’t account for college performance, I’d easily side with ZiPS and Davenport over OOPSY on that trio. OOPSY projecting Domínguez so well can be partly explained by the system’s inclusion of his lightning-quick bat speed, but ZiPS also had him in its top five.
The ZiPS top five prospects by peak OPS+, in order, are Emmanuel Rodriguez, Bryce Eldridge, Eric Bitonti, Montes, and Domínguez. Comparatively, ZiPS is the most bullish on Bitonti, Arjun Nimmala, Moore, Yeremi Cabrera, and Emmanuel Rodriguez; it is the least bullish on McGonigle, Mike Boeve, and Colt Emerson. Emerson and McGonigle still rated highly on yesterday’s ZiPS Top 100 thanks to their defensive chops, while ZiPS’ enthusiasm for Robert Calaz is more tempered after accounting for his play in the outfield, per my conversations with Dan. ZiPS has three-year forecasts available on FanGraphs, which are worth checking out for comparison, but three years typically isn’t far enough out to cover the peak for most prospects.
Davenport’s top five prospects, in order, are Kurtz, Anthony, Xavier Isaac, Luke Adams, and Emmanuel Rodriguez. Davenport’s system is, relatively speaking, the most optimistic about Bazzana, Kurtz, Sal Stewart, and Adams, while it’s the least optimistic about Domínguez, Cam Smith, and Coby Mayo. Davenport also offers six-year forecasts for each player, which can be found here.
Wrapping Up
Compared to looking at a single projection in isolation, combining peak projections from various systems produces a consensus list that inspires greater confidence and (typically) better forecasts. As the creator of one of the systems featured in this piece, I certainly feel more confident in my projection for a player after seeing it align with those of other (smarter) prognosticators — or with the evaluations of scouts.
The consensus list offers a snapshot of the current state of public-facing peak forecasts for hitting prospects. When considering any differences between it and an excellent scouting-based list like Eric Longenhagen’s, it’s interesting to consider what may have been left out of the peak OPS+ projections (other than defense, naturally): swing speed and other Statcast data for minor leaguers, scouting grades, amateur statistics, etcetera. Giving more weight to Statcast data probably explains part of why scouts tend to be higher on Sebastian Walcott, for example. While projection systems excel at accounting for the precise relative impact of different data points in a systematic way, they could never take into account the same breadth of data a scout can. In any case, improvements in prospect projections might reasonably come from the inclusion of some of these missing but quantifiable data points.
This feels pretty fantasy geared, and is an interesting addition to prospect week, if not entirely real-life centered. One way that I think this could be more helpful and applicable is if the defensive projection in ZIPS is included, with normalized def WAR by position attributed to the OOPSY and Davenport’s projections (i.e. have those projection systems norm the defensive data to everyone being average at their projected position.)
From this you would get the benefit of composite rankings based on multiple projection systems, while not forgoing the defense/position dependent nature of player evaluation. I would be interested in how this type of composite projection would compare in predicting team controlled WAR in the way that we are able to evaluate both current projection systems and prospect evaluation rankings.
Though I guess you’d also have to have some sort of BSR to actually pull WAR figures from this composite.
yeah, a focus on hitting, even real-life hitting, is naturally more closely related to fantasy…i’d also be interested to see how the consensus projections look with defense added–whether that means incorporating defense in the sensible way you suggested, with everyone assumed average at their position, or in a more a rigorous way…a good idea for a future OOPSY article!
I imagine defense is extremely difficult to predict via algorithm using only minor league data.
Fantastic article, thank you. This will be invaluable for my minor league fantasy draft in the spring.
Someone needs to explain to me what the park factors are like for the Mariners’ affiliates, because Lazaro Montes and Michael Arroyo’s numbers are cartoonish and Eric has been very skeptical this can continue. But you would think the projection systems would address this, and they do not seem that it is a problem.
Yeah we do account for park factors in the minors too…those guys project well even after adjusting for that!
Thanks Jordan, great information.
I think the question is whether the park effects are enough for this task. In the majors, we know that park effects mostly take care of what they are supposed to, except we have no idea how to construct a park effect for Coors. This is probably more of a problem for the most extreme parks, such as the D-Backs’ and Athletics’ upper level affiliates. But when you combine the projection systems’ seeming disregard for strikeouts as a predictor with some park effect questions, I have a lot of questions about the projections for Montes and Arroyo. And Calaz too.
I’ll trust someone like Jordan to handle the normalization of performance over someone who thinks “we” still have no idea how to adjust for Coors Field.
Eric can answer for himself but he seems more guarded than skeptical. The parks and leagues Montes has played in and whether or not Montes can field a position have him on the brakes a bit.
ZiPS Top 100 is also a bit less bullish on Montes for defense reasons
I don’t know if this is still the case today but at least as of a few years ago, wRC+ was not park adjusted in the minors. BP’s DRC+ is and has Montes/Arroyo’s marks as 124 in A/93 in A+ for Montes and 121 in A/111 in A+ for Arroyo (vs. 148/125 wRC+ for Montes and 141/155 wRC+ for Arroyo)
It’s not a perfect comparison since DRC+ includes some process stuff in the metric compared to wRC+, but yeah pretty clear the parks are very offense boosting, which these projection systems can also account for
The California League (A) is notoriously offense-forward, FWIW. Though both guys were pretty great in A and A+. And both parks (Modesto and Everett) have been around for a long time.
But yeah it is kind of odd to have two dudes who had more or less identical huge years, at the same age (literally, they were born 12 days apart), with almost the exact amount of time as the other in each park.
As someone who is not a prospect hound, adding the team whose system the prospect is in to the table above would make it about 100 times more useful to me.
As it is, it is a list of players, a few of whom I recognize.
My apologies, you can find each player’s player page by clicking the hyperlinks in the list. and fwiw, only two prospects lack unique names (emmanuel rodriguez and jacob wilson, you want the one on the twins and on the athletics, respectively!)
Excellent article, thank you! I would second the idea of having each player’ organization listed. Looking each player’s name up, if you are most interested in a particular team’s prospects, is not a solution. An added column listing organization would fix this problem.
Thanks! Too late now but i’ll add team next time
It’s too late? Can you not, like, just add it now?
I can’t edit the piece after it’s published. Lemme see if I can throw the teams in a Google doc and share it here
Update: I got permission to edit…team and position are in there now 🙂
Thanks!
Co-sign
Position (and team!) have now been added. I will make sure to remember to include them the next time around. Thanks for the good feedback and for reading 🙂
If Basallo can catch at all and stays healthy… my oh my