Seattle Sports Insider

HQ MLEs and What they Mean

Candid photo of the inventor MARCEL hard at work

Posted by

3/16/17

I was reticent to post on this topic, if only because Doc's shtick is far...far more fun than math formulas and how they're obtained.

Doc has, however, been fishing for alternative content - I think we exhaust him by making him do most of the talking!

Ultimately, I decided to post on HQ's MLEs because the way the system works is actually interesting (to an egghead...OK...I'm being liberal with the word interesting, but cut a math geek some slack :D )

An MLE (Major League Equivalency) is an attempt to project a player's likely MLB statistics would look like if you transitioned him to the big leagues without any changes in his current skill set or established level of performance.

Back in the early 2000s when MLEs were being popularized by Baseball Prospectus, the state of the art was to take the player's batting line, as a whole, and correlate it with what batting lines for similar players or for players who made the transition mid-season, looked like in the big leagues. So you'd take your AAA stat-line (back then it was as simple as BA/OPS with a park adjustment, usually), and you'd run a simple one-way correlation or get a composite pair study together or calculate player similarity scores and project based on what similar players did and that was your answer.

It was innovative work at the time, and I am not taking anything away from the pioneers, but we know a lot more about how players tend to perform when they change leagues and about which types of statistical profiles are the most likely to translate. For example, minor leaguers with poor K/BB ratios, very high K rates (>23% of plate appearances), or low batting averages on balls in play tend to struggle. Now that we know these things, it's easier for us to spot Dallas McPherson and Brandon Wood Mike Olt coming (and failing) before they arrive and ruin our day.

But that knowledge tends to come at you in pieces...we have thumb rules about things like minor league BB rates vs. minor league power (hint: if they walk a ton in the minors and have an ISO < .120 or so...they won't walk in the big leagues because they will be pitched to very aggressively), or GB/FB or usable speed from SB% and 3B/(2B+3B). Each rule is handy, but it's hard to see how they all come together.

As I understand what HQ does - they blend all of that together by running multiple regression analysis on a set of component skills. You hear Doc quote many of them when he reads from his HQ guide each spring. PX, SX, CT%, pull%, Eye, BABIP (hard hit rate now, since we have that data). They account for which minor the available statistics that influence each component skill and the covariance between components.

So...for each component stat, we get an expected performance (in that component), and from there, they make a projection to theoretical MLB statistics (by correlating the components to the statistical outcome they produce on average).

It ain't your daddy's MLE.

I posted a few MLEs for Mariner prospects of interest yesterday - a short summary:

Maniger: .273/.345/.484 (this does assume a neutral home ballpark)
Gamel: .286/.341/.390
Vogelbach: .246/.349/.411 (more on this in a moment)
Zunino: .236/.300/.414
Heredia: .258/.334/.331
Motter: .187/.241/.318 (YIKES!)
Freeman: .238/.293/.311
Peterson: .226/.274/.383
Smith: .220/..249/.291 (um...ew)
Law: .270/.338/.353
Landry: ,197/.248/.293
Littlewood: .245/.335/.332 (little wood indeed...no pop here)
Lopes: .278/.331/.336
Miller: .232/.292/.274
O'Neill: .284/.359/.487 (wow...color me stunned, given his atrocious K rate)
Pizzano: .203/.253/.290
Powell: .232/.281/.301
Romero: .235/.278/.401
Shank: .249/.297/.319

So..........

Three things jump out at me.

1) Maniger :)

2) O'Neill is somehow getting credit for enough things that he does well to offset the horrendous K rate...wonder what went into those formulas that benefited him...hard-hit rate, flyball rate?

3) Vogelbach may not be the juicy find we hope.

I can tell you, based on the actual batting line he posted last year, why Vogelbach has such a low BA and ISO. Way...way too many grounders. Doc is right that his swing produces too many fliners and grounders...that was statistically the case in 2016. It is possible to run a .33x BABIP in the minors as Vogelbach has consistently done while hitting a blizzard of grounders (they don't shift in the minors very much), but it is not possible to even hit .290 on balls in play if they're mostly grounders in the big leagues.

Something to keep an eye on.

Also...our vaunted amazing depth doesn't look very amazing...just FWIW.

Tags:

sabermetrics

Comments

Darn-nation

An OF of Gamel-Haniger-O'Neill wouldn't look too bad, would it?

I love the projected .236 for Zunino. Over .230 he gets really good for a + defense catcher.

Vogs at .246 just seems way too low to me. But not if he's hitting billions of grounders, I suppose. Will be interesting to see how he's defensed. I'm more bullish than .246.

And I still am wondering, given that he's bashed in the desert but not out of it (his explosion may be unrelated to his PCL desert mashing streak, but it came at the same time), if Haniger is a 20+ HR guy for us. Of course, he doesn't have to be to be pretty juicy. At the numbers projected, Haniger comes across (in '16 performance) very Piscotty-like (as I've said before). That's a sweet player. Ozuna had similar numbers, too. Springer was a bit better. Those guys had OPS+'s of 112, 109, 124. (right in line with Matt's 110 "floor"). Ozuna hit 23 HR's and Slugged .186. For Piscotty it was 22 and .184. Springer was at 29 and .196.

The MLE has Haniger Slugging at .211. That's 30 HR territory, or 40 doubles and 24 HR's. Those are lofty numbers.

.200 Slg% fits for O'Neill more, IMHO, than Haniger.

I don't think it's far off at all.

Haniger's ISO. I'll be astounded if he doesn't settle in as a .200+ ISO hitter. Too much natural loft on his fly balls and too much opposite field power for him not to slug. The system gave him 30 2B and 19 HR in about 480 ABs, BTW...so...Seager-ish but with more doubles.

But I tend to agree on Vogelbach hitting for a better average than this shows...eventually...he might struggle for a while before Edgar fixes his swing path.

If Boomstick has Haniger's MLE this year I'll take it

Will admit

that the difference between Piscotty/Ozuna and Slg% of .200 is pretty much splitting hairs.

Vogelbach's MLE

Wouldn't get too depressed that it amounted to 'only' about a 100 OPS+ in the American League.

(1) Most these kids are young and growing. If Vogelbach was able to hit average-solid in the AL at age 23, his peak would be pretty nice.

(2) MLE's are going to miss a lot of good players by their very nature. Vogelbach's OPS was #3 in the PCL last year; are we going to conclude only one PCL player is going to make it in the bigs? Well, y'know

...

The grounder and fliner stuff is cool Matt. Thanks for pointing that out.

Adding to that ... in Vogelbach's case he's got a unique Ichiro, Boggs-like ability to flick the ball out of the catcher's mitt and roll it into LF. That defeats the major league shift, and that's important.

...

'Bach is still evolving, little question in my mind. He's got the ability to select how much Boggsy stuff he's going to go for, how much he wants to turn on the ball, etc. His MLE's are nice to know but not a decisive factor in my mind.

Got to address Heredia

That MLE is in his first stateside PAs, during a predicted "rusty" period after time away from the game to defect, get gov't & MLB approval, sign and get started. He only signed 13 months ago. He was 22 his last full season in Cuba. I've said before that I do not think his ceiling is established due to these factors.

He's visibly looking like a better base stealer this spring than he was in Cuba statistically. Looks better than I saw last year, for that matter. Learning from the newly acquired pro thieves? Still developing in general? I think so.

MLE for him doesn't look completely horrible and yet he's the last player on that list that I'd even look for an MLE for because of his stateside newness.

...O'Neill...putting up lesser numbers at a lower level than Maniger.

Has an MLE...

.017 higher? And this with Maniger's being neutral park? .014 of that is OBP for O'Neill over Haniger. Tanks AA numbers were ever so slightly better but it was all SLG, significantly lower OBP. And that's ignoring the AAA stats that...Vogue was 3rd best PCL OPS. Mitch Haniger needed about 70 more PA (at the same rates) to qualify and lead the PCL by .099.

The only thing I can see that make O'Neill's year look clearly better to me is age. Does that affect MLE? Should it?

I'm comparing them partially because they're comparable, partially because Man's AAA performance was basically top in the league, the MLE is not seeming so. Not trying to knock O'Neill or kill the messenger for that matter. It's all interesting to me.

Two things...

Maniger played in the hot desert at high elevation in a known hitter's league to get the OPS he did. The figures project how he would hit in a neutral park.

Maniger is also older, and age vs. league is one of many factors considered.

As for Heredia, I absolutely agree. His MLE is far from the last word...he looks much better now than he did last September in Seattle.

If MLE is saying

"This is what that equates to in MLB terms", Should age be a factor at all? I can not figure reasoning for it to. I understand factoring age when projecting future performance. That's not the purpose of MLE as I understand it. It is difficult to avoid thoughts of projection when viewing them, but that's not what the tool is supposed to be. Why should age factor in at all?

BB% and K% are both supposed to translate and those are the numbers Maniger has the clearest separation from Tank in.

Age is considered a factor because....

...when you compare two hitters who have the same stat line at the same level in the same park environment that get promoted to the big leagues, the one who is younger when the promotion occurs is statistically considerably more likely to hit better immediately (and through time). That's been proven empirically. This is because:

1) Old players who have more reps in a level learn what it takes to beat that level, but not necessarily what they need to do to transfer that to the big leagues...
2) Players who perform well at a younger age for level are usually the players with more natural talent
3) Players who perform well at a younger age preserve more of their "young man" skills...speed, bat speed, resilience...but neither the young guy nor the old guy will have old man skills when they first try MLB (at least...neither will be more likely than the other)
4) The younger guy would have to have improved more in the last year than the older guy would to get the same results, so this usually is an indication that the younger guy can learn faster.

Then personally I would consider these projections, not MLE's

More like PECOTA or KATOH than James' old MLE's.

I understand 1000% what you are saying about using age *only* as an adjustment factor for same-season performance. But my own opinion still is different from Shandler's.

When I'm talking about MLE's, I just mean "what would Seager hit in the PCL if you put him there right now." That age factor changes everything on the M's 24-year-old ML-ready players. Their non-age-adjusted MLE's would presumably be much higher.

Thanks again pardner.

And seriously could covet all of the brainiacs on here!!! Average-guy-syndrome here!!!