clock menu more-arrow no yes

Filed under:

Andrew McCutchen, Dave Winfield, and Goose Goslin: Rolling out a new projection system

After introducing the comparison and likeness prospect system a couple months ago, CAL has taken the next step: turning it into a projection system.

If you buy something from an SB Nation link, Vox Media may earn a commission. See our ethics statement.

Matt Kartozian-USA TODAY Sports

A few months ago, I introduced the first iteration of CAL, the comparison and likeness prospect classification system I derived based off Bill James’ Similarity Score. In a nutshell, the algorithms try to provide a better context to evaluate minor leaguers by determining similarly performing players (based on age, position, level of competition, specific points of production, etc.). Since then I’ve tweaked the formulas, made a few bold claims – such as Maikel Franco’s High Bust Potential — and rolled out the updated results in my book, The 2015 Prospect Digest Handbook.

With the handbook firmly out of the way, and the subsequent writer’s burnout squarely in my rearview mirror (I think), I’ve spent the last couple of months developing not only a MLB-version of CAL, but a projection system based squarely off of a player’s top comparables. And the initial results — and I say initial because I’m sure I’ll continue to tweak the formulas in an effort to improve the results — appear to be encouraging.

First, though, the process, some notes, and a little extra commentary. And, of course, I’d love to hear any feedback — positive or negative. If you’d like to skip through the verbage just jump to the last paragraph and click on the hyperlink to view all the projections.

• The CAL database was constructed from FanGraphs MLB statistics, beginning with the start of the Live Ball Era (1920).
• The system uses a player’s previous three seasons. Or in the case of younger players, whatever the available sample size (i.e. if a player is entering his third season, it will use the previous two years)
• After CAL determines a player’s top comparables, I developed a system of projection algorithms of varying weights.
• And, finally, to check the accuracy of the projection system I used statistics from 2011-to-2013 to project what they would have done in 2014. One additional note: the only 2014 statistic I did not project was plate appearance totals in order to correctly gauge the system’s accuracy.

Now for some of CAL’s biggest hits and, of course, digging into some of the bigger misses as well.

The Hits:

Andrew McCutchen

Heading into the 2014 season, the All Star center fielder’s top comparables, according to CAL, were a pair of Hall of Famers (Dave Winfield and Goose Goslin), another pair of high-performing All Stars (Roy White and Bobby Abreu), and former Pittsburgh outfielder Jim Russell. And here are McCutchen’s 2014 projections based off those five players:

PA AB AVG OBP SLG OPS ISO H 1B 2B 3B HR BB% K% SB CS wOBA
CAL 648 562 0.314 0.405 0.546 0.950 0.232 177 109 34 5 28 12.51% 17.61% 20 7 0.413
2014 648 548 0.314 0.410 0.542 0.952 0.228 172 103 38 6 25 13.00% 17.70% 18 3 0.419

Matt Carpenter
Heading into last season Carpenter was coming off back-to-back seasons in which he hit a combined .310/.383/.475 through more than 1,000 plate appearances. And yet, his production took a rather noticeable downturn in 2014 — he batted .272/.375/.375. CAL was able to sniff out his decline:

PA AB AVG OBP SLG OPS ISO H 1B 2B 3B HR BB% K% SB CS wOBA
CAL 709 604 0.276 0.383 0.390 0.773 0.114 167 117 41 0 9 13.29% 16.95% 2 2 0.347
2014 709 595 0.272 0.375 0.375 0.750 0.103 162 119 33 2 8 13.40% 15.70% 5 3 0.343

Alex Gordon
CAL linked the former can’t-miss-superstar-turned-bust-turned-one-of-the-better-players-in-baseball to four All Stars (Al Smith, Hunter Pence, Sam West, and Ken Griffey Sr.) and Derek Bell, who hit .334/.384/.442 in 1995 but failed to make the Midsummer Classic. For his part, Gordon was coming off of a bit of down season after two stellar campaigns; he slugged .298/.372/.478 in 2011-2012 and followed that up by hitting .265/.327/.422. CAL, however, wasn’t too fond of Gordon’s bounce-back potential:

PA AB AVG OBP SLG OPS ISO H 1B 2B 3B HR BB% K% SB CS wOBA
CAL 643 586 0.269 0.334 0.432 0.766 0.163 158 100 36 6 16 8.26% 23.33% 7 5 0.338
2014 643 563 0.266 0.351 0.432 0.783 0.166 150 96 34 1 19 10.10% 19.60% 12 3 0.351

And here are two of CAL’s biggest misses:

Chris Davis
The Orioles slugger set the world ablaze by bashing 95 extra-base knocks en route to slugging .286/.370/.634 in 2013. He promptly followed that up with a paltry .196/.300/.404 triple-slash line. His top comparables heading into his disappointing season: Richie Sexson, Danny Tartabull, Craig Wilson, Mo Vaughn, and Nate Colbert.

PA AB AVG OBP SLG OPS ISO H 1B 2B 3B HR BB% K% SB CS wOBA
CAL 525 462 0.288 0.374 0.575 0.949 0.286 133 70 28 1 34 10.64% 30.01% 4 1 0.409
Actual 525 450 0.196 0.300 0.404 0.704 0.209 88 46 16 0 26 11.40% 33.00% 2 1 0.308

Ouch. CAL didn’t think Davis would be able come close to even matching his career year, but it definitely didn’t think he’d crash as far as he did.

Carlos Gomez
It took a long time for the former Johan Santana-trade-centerpiece to develop into the perennial All Star candidate that we recognize today. Gomez batted just .247/.294/.379 through his first 2,100+ plate appearances, including his .260/.305/.463 triple-slash line in 2012. His top comparables heading into last season: Claudell Washington, Andy Van Slyke, B.J. Upton, Sammy Sosa, and Mike Davis. And here’s his projection:

PA AB AVG OBP SLG OPS ISO H 1B 2B 3B HR BB% K% SB CS wOBA
CAL 644 602 0.238 0.288 0.365 0.653 0.127 143 101 25 17 5.63% 28.46% 38 6 0.290
Actual 644 574 0.284 0.356 0.477 0.833 0.193 163 102 34 4 23 7.30% 21.90% 34 12 0.368

Obviously, the CALs range from very, very good (Sammy Sosa) to very, very poor (B.J. Upton). Gomez was, in CAL’s opinion, pretty volatile.

For the rest of the projections click HERE. I’d love to hear some feedback — good, bad, or indifferent — so let me know what you think.

****

All statistics courtesy of FanGraphs.

For more analysis check out Joe Werner's site: ProspectDigest.com. You can follow him on Twitter at @JoltinJoey.

Check out Joe's book, The 2015 Prospect Digest Handbook, click HERE.