Predicting the Power Tool Using Metrics
Former Braves and current Yankees “prospect” Cody Johnson blasted the longest home run I’ve ever seen. In 2008, it came against St. Louis Cardinals pitcher Maikel Cleto (A Mets prospect at the time). Johnson turned on a 95 MPH fastball at the letters, and in turn, seared an “80 power” grade into my mind for eternity. At the time, I wasn’t really concerned about the metrics behind actually producing to the home run equivalent “80 power” (38+ home runs at the Major League level), but neither were scouting contacts. In retrospect, I should have paid more attention as his K%, 34.4%, made it impossible to project elite power from any prospect.
Why? Fangraphs veteran Jeff Zimmerman looked at true “80 power” seasons over the past 38 years and found a sweet spot from a metrics standpoint amongst players accumulating 38+ home run seasons. These include:
FB% = 43%
HR/FB% = 23%
K% = 22%
In Johnson’s case, his 34.4% K% should have raised enormous red flags, as hitters generally strikeout more as they move up levels in an organization. This line of thinking held true as Johnson went on to post a slightly higher strikeout total in high-A, followed by a 43%-plus rate in double-A which has essentially ended the discussion of Johnson as a true prospect and is responsible for his being shipped to the Yankees as a potential reclamation project.
Entering the season, Baseball America’s top-100 included specific tools and their associated grades. For example, the number one prospect in baseball was Bryce Harper, whose power tool was graded as a true-80 on the 20/80 scale. For reference, the full scale is below.
**Thanks to Dave Gershman of pennleaguereport.net for the offense/hitting grades chart**
Look further down the list and Oakland Athletics first baseman Chris Carter ranked as the 91st best prospect in the game with a power tool good enough to earn a “70 grade”. Having scouted Carter in person with Kannapolis, his in-game power made it easy to hang an identical grade on the young slugger. However, the peripherals behind Chris Carter producing 30+ home run seasons at the big league level are murky at best. And if one is unable to envision a navigable path to such lofty home run totals, then is the grade even useful in the first place?
In triple-A, Chris Carter has now accumulated well over 900 plate appearances at the level with a K% of 24-25%. A small sample at the Major League level indicates a strikeout rate above 33% which should somewhat regress to the mean, but may never fall below 27-28% based on this piece on MLE which surmises players strikeout 2.5% to 3.6% more at baseball’s highest level.
Additionally, it’s important to pay passing attention to walk rates. Even though it is generally accepted that higher walk totals lead to better pitch selection and more consistent production as a batter, it does cut into the total number of batted balls to work from. In Carter’s case, I’ll just average his walk rates between what he accumulated in Sacramento this season and his career walk rate in Oakland. That average is an even 10%.
Based on 600 plate appearances, a 10% walk rate and 28% strikeout rate means Chris Carter is unlikely to put a ball in play 228 times. This leaves 372 at bats for which he will actually make contact. For those remaining at bats, let’s see what Jeff Zimmerman’s “Sweet Spot” rates produce in terms of home run production.
With a 43% fly ball percentage, Chris Carter would project to hit 160 balls in the air over the course of a season. If 23% of those balls left the park, his home run total would be 36-37 allowing Carter to exceed his “70 power” projection.
Unfortunately, only 26 qualified players in Major League Baseball this season actually reached a 43% fly ball rate in 2011. Even more damning is the fact only one player (Mike Stanton) has crossed the 23% HR/FB threshold. This brings up a simple, common sense question. Is Chris Carter a player capable of posting FB% and HR/FB% amongst the best in baseball? Whatever sample size we do have points to a resounding no.
And while this line of thinking creates a bit of a sliding scale in terms of a path to power totals based on FB% and HR/FB% both being able to affect a player’s home run totals, Carter achieves his “70 grade” with percentages in the 40% FB and 20% HR/FB range — a far cry from the 50% FB and 8% HR/FB he has posted which would produce just less than 15 home runs over 600 plate appearances (40 grade)
A nearly perfect batted -ball comparison this season would be Cardinals Lance Berkman who has accumulated 31 home runs based on a FB% of 40% and HR/FB of 20%. However, his walk and strikeout totals only count for 31.7% of hit total plate appearances, a full 6.3% less than we can expect from Chris Carter adding nearly 38 additional opportunities to make contact and about three more home runs — a difference of nearly half a grade.
Having scouted Chris Carter early in his career in the South Atlantic League, I find myself pulling for the young first baseman to win the first base job in Oakland outright. However, the more I analyze the numbers behind Carter’s projected power totals, as well as other players with similar skills, the less confident I am in their ever reaching those lofty grade expectations.
Mike Newman is the Owner/Managing Editor ofROTOscouting, a subscription site focused on baseball scouting, baseball prospects and fantasy baseball. Follow me onTwitter. Likeus on Facebook.Subscribeto my YouTube Channel.
Does the power scale change? In other words, would 80 power have been higher than 38 home runs in, for example, 1998?
I have the same question.
Isn’t a “50” supposed to be MLB average? Cause the average player doesn’t hit .277 with 18-19 homers. MLB average is actually .255 with 12 HR’s per 500 PA’s.
My understanding is that a 50 grade is a MLB regular, not a MLB average. The average includes every MLB player. 40/45 players are generally regarded as the up-and-down, AAAA bench types which .255 with 12 HR would represent.
Predicting… The… Power… Tool… Using… Metrics….
Oh, you mean: Predicting Power Using Metrics. Great headline
Predicting the Telo tool is a much easier exercise.
Telo is a tool? You don’t say!
That isn’t what he meant, which is why that isn’t what he put. Regardless of your reading incomprehension skills, the headline accurately reflects the article.
What about development? They are not going to improve any of their metrics as they get older?
Yes, but the closer a player gets to his prime in general, the tougher that task becomes.
Bryce Harper, 38+ HRs and .320+ BA?! Dream on!!! He hit .256 with 3 HRs in almost 40 games in AA in 2011!!
To be fair to Baseball America, the ONLY tool they graded for Harper was the power tool. In fairness to Harper, he should have just graduated high school and he finished the season in double-A. He’s a special, SPECIAL player without a doubt.
No way, Bob obviously is a genius and knows something that everyone else doesn’t.
As a teenager. Players his age don’t get to that level, he did.
This year gave no reason to doubt his potential.
You do realize that Bryce Harper is just 18 years old, right?
Fantastic stuff, Mike. If Bill Simmons has not called you yet to take Grantland’s BB coverage to another level, he’s a damn fool.
Thank you Paul! Personally, writing a metrics-centered piece for the Fangraphs crowd was a bit daunting considering I’m 70% scouting, 30% numbers at most. This is an idea I’ve been kicking around for quite awhile in my head and Jeff really helped tighten up my line of thinking. The “sweet spot” he found was really projection gold.
Yeah, the piece is obviously great in large part to Jeff’s outstanding work.
Regarding the scouting/stats composition of your work, seems to me the timing is perfect. With the whole Moneyball debate going on these days, I have seen stats-based writers falling all over themselves to say the anti-scouting narrative is overdone. I’m glad their views have supposedly become more nuanced, but frankly I haven’t seen any evidence of it in their writing. You bring that nuance to the site, and your ability to put stats in context is very welcome in general from my point of view.
An excellent chance to toss in some descriptive stats – namely standard deviation on the average FB%, HR/FB%, and K%. Helps describe how uniform these 80+ players are, and what an outlier (+2 s.d.) might look like.
Here is list of the players since 2001 (since FG has FB data):
https://docs.google.com/spreadsheet/ccc?key=0ApDc5PGsBzgVdHpZM0JNMkg4X01ZakdEOGgyb1ZzeUE&hl=en_US
Probably the most confusing article I’ve seen here, starting with the title.
Seems pretty straight forward to me. What part of it are you having trouble with?
I think in some ways it’s meant to be confusing because Chris Carter has received a “70 grade” on his power from a major publication and I just can’t connect the dots to see how it can realistically happen using metrics. Should this line of thinking even enter the conversation when it comes to scouting?
Scouting grades are presented in a pretty neat-and-tidy fashion, but should they be? For a scout to submit a power grade to his respective organization, should he be able to prove it’s within the realm of possibility?
Take Cody Johnson again. During that game, I witnessed a moon shot, but saw him swing through fastball-after-fastball. Should I grade based on the blast or the fact I had huge doubts on whether he would ever figure it out? Going by the blast, his power was an 80. If one goes on the fact he probably will never hit MLB pitching, then a true grade would be a 20. I’m 99.9% sure scouts have submitted both which makes it seem pretty arbitrary. I wanted to put some basic numbers to that and open the conversation up to the FG crowd.
Mike: Along these lines, I’d love to see a treatment of the ability by hitters to foul off good pitches and stay in an at-bat. We of course know that slap hitters do a good job of this, but Michael Morse, who it seems is widely panned as a fluke, looks like the real deal to me. Also, I haven’t seen a lot of him, but I have seen some marathon ABs by Lonnie Chisenhall. Two guys who are either impact hitters now or are projected to be, with two different stats profiles at different periods in their careers. Okay, so I know your job is not to take requests, just throwing this out as one of those interesting subjects that I think neither traditional scouting nor stats treatment gets all the way correct.
To me, a 70 power, 20 hit says that a player has the ability to hit the ball a ton, but he probably won’t make enough contact to do much with his raw power. If the power tool’s also low, it suggests that he doesn’t have any real offensive talents.
I think that tells us more when we’re looking at potential. With a 20 hit tool, we know his contact rates could keep him from showing much in-game power, but even a modest increase in contact skill – say, to a 30 – and you could see a leap in game power as the player develops into a Mark Reynolds type of hitter.
Excellent point Wilson. That’s definitely a way to look at it. It just still brings me back to the idea of a player putting enough fly balls in play to project those home run totals. For what Mark Reynolds does, I’d consider him to be an outlier in general.
One more thing….
My pieces are written from a scouting and player development perspective. I will use numbers to support the scouting I do, but the scale I work off of in my mind is the scouting scale and not metrics.
That’s not saying I won’t use metrics, or am not interested in learning more, it just means pretty much any piece I write will be through a bit of a different perspective than most of the work here at FG.
Yeah, maybe fangraphs readers just aren’t familiar with scouting scales.
This is a great article, one of the best I’ve read on FanGraphs all season. Would be interesting to compare these “sweet spot” metrics to minor league players for 2011 to see if it helps to identify any under the radar players.
Will be interesting to see how Hosmer looking on this scale in the next few years…was he a “70 power” prospect?
Thanks Lucky! You know the metrics and his current power totals point to a guy who should hit that ceiling. When I see a player so young doing what he’s doing with a 30% FB rate, it makes me think he just hasn’t completely mastered the ability to generate lift and that there’s more power to come.
I watch a lot of his ABs. He is actually really a hacker right now and swings at way too many pitches down out of the strike zone. He actually creates lift really easily for a guy his age, oppo field too. The Royals not having the slightest idea how to develop power hitters are not going to mess with him at all. It’s scary to think what he would be doing if he was in a org. that taught having pro ABs. But within a year of arbitration, Mr. Boras will pay him a visit and show him some nice color charts that make it clear he needs to start taking a lot more pitches. And then the light show begins.
Interesting stuff on Hosmer Paul. A more mature batting eye often comes later in a player’s development. As that tightens up, his walk rates should rise and power totals improve. The indicators are obviously there for him to be one of the better offensive first basemen in the game.
I’m not much into scouting, so I could be completely off-base here, but . . .
It seems to me that the power tool should have nothing to do with predicting how many homeruns a player will hit but rather with how many homeruns he’ll hit per batted ball. Essentially, I’d think that the power tool should be used to project isoP . . . it seems wrong to me to include strikeouts and walks in a batter’s power tool.
I guess I agree with Telo above — this seems more to me like predicting a player’s actual power, not what I would consider his “power tool”
I should say —
I liked the article and I think the reasoning is completely sound, I just don’t understand why a player’s “power tool” would be based on anything besides how far he hits the ball when he hits it
jessef,
The answer is pretty simple. The tool is judged in terms of a grade and that grade is associated with a specific home run total. In all honesty, most scouts I speak to do look at the tool in terms of how far a player can hit a ball. My piece tries to question that notion and asks whether metrics should be used as a better way to grade. It’s all about finding the best predictor of future power. I’m just not sure judging how far a guy can hit a baseball is the best way to do that.
Thanks for replying!
If the tool is judged to predict specific homerun totals then the way you’re suggesting makes perfect sense.
I guess what I don’t understand is, isn’t the whole point of separating out skills in detailed scouting reports (I’m not a scout, so I could easily be wrong here (and, please, let me know if I am)!) that players are already being graded for strikeout- and walk-rates through batting eye- and contact- tool ratings?
Wanted to add another comment lauding this article. This is a really fascinating way to break down this issue, and it points to how stats and scouting could fit together.
The idea that HR totals are a combination of the power and bat tool and thus should not be a product of simply the power grade is a really clearly illustrated one. It seems to me that tools as scouted and projected should be used as proxies for HR/FB%, contact%, LD%, FB%, and things like that. I.e. a scout will be much better at projecting a A or AA ballplayer than the stats will, and if he or she focuses on what type of player from a power and ability to barrel up the ball, the type of stats translations that Mike does above are how we paint the picture of the major league ballplayer.
Very cool.