Not so technically speaking and strictly within the context of this post…
Strasburg is to Starting Pitching what Bard is to Relief Pitching. Oh how I wish this could be true for Bard.
Comment by OtherSideoftheCoin — February 19, 2013 @ 3:20 pm
I think you’re mostly right, but there may also be a position factor. BA rated Bogaertes as a 55 defender; I would think he would end up being below average at shortstop. But if Sky were right, some of those catcher defense grades would be pretty awful.
Comment by Izzy Hechkoff — February 19, 2013 @ 3:22 pm
How would you place Total WARP on a 20-80 Scale. I was thinking about this when reading Baseball America’s Prospect Guide
Here is what I came up with
80- 8 WARP- A Hall of Fame/Perennial MVP Type Talent- A Top 5 Player in the Game on a yearly basis. Mike Trout/Justin Verlander
75- 7 WARP-Perennial All-Star- A top 10 Player in the game.
70- 6 WARP- All-Star- A top 20 Player in the game
65- 5 WARP-A top 40 player in the game
60- 4 WARP- A top 75 player in the game
55- 3 WARP- A above average starter. A Top 125 player
50- 2 WARP- A league average starter- A Top 200 Player
45- 1 WARP- A below average starter/plus back-up or back-end starter.
40- O WARP- A fringe MVP Player (Back-Up Catcher/Long-Reliever/Pinch-Runner)
35- -0.5- WARP- A 4AAAA Type Talent (Might or Might not be worth having on a 40 Man Roster
30/25/20 Organizational Filler- Used primarily to fill out Minor League Rosters.
I’m not claiming this to be a definitive guide. I’m just wondering if these numbers seem to be in the right ball-park.
This is my favorite article I’ve read in a long long time. I think this really bridges the gap between the perceived SABR/Scout war that never really existed in the first place while also giving a statistical measure to a scale that, really, lets be honest, we all love to use and observe.
I remember back when Vlad Guerrero came up and scouts would joke that his arm required another category unto itself? Because of how strong it was? No because you never know where it was going to end up; either an on the fly dart from the right field foul pole, or somewhere in the bleachers in center.
Thanks for this Mark. Fantastic work.
Comment by Tony Gwynn's Knee — February 19, 2013 @ 3:31 pm
A++ for effort. Love the idea.
Comment by Dan Rozenson — February 19, 2013 @ 3:39 pm
What do you make of Mike Trout’s 12 BsR score last year? Since a 9 BsR average = 80 Speed, does Trout’s score imply a rating-system-busting 90 speed, or 4 standard deviations above the mean? I mean, assuming he can maintain that level, of course.
Comment by Steve Staude. — February 19, 2013 @ 3:42 pm
Fine article, Mark. I’m not sure about your “hit tool” comment, though. It was my understanding that hit tool translates not into “ability to make contact” (i.e. simply avoiding strikeouts), but rather the ability to hit for average — which incorporates but isn’t limited to contact-making.
I absolutely could be mistaken, though.
Comment by Don Mossi's ears — February 19, 2013 @ 3:53 pm
FWIW, I use this scale to introduce people to WAR:
10 All-time great season
8 MVP finalist
6 MVP candidate
4 All-Star caliber
2 league average
btw, I don’t think a normal distribution applies to steals. It’s extremely skewed towards low steal numbers, and obviously there’s no such thing as negative steals. Actually, you could say something similar about a lot of these categories.
Comment by Steve Staude. — February 19, 2013 @ 4:17 pm
I’m a little confused, on Teixeira’s fangraphs page it says his babip is .250, here it says .254, but if I did everything correctly, it’s actually .258.
This article is interesting and certainly highlights some of the major disagreements between the scouting community and the SABR community with regard to individual players (e.g., I don’t think any scout would call Drew Stubbs a 50 grade CFer).
Also, I really don’t think BABIP is a good way of measuring the hit tool. Dexter Fowler’s 70 grade is a good example of why.
Agreed. One of the best of the early season so far. Nice work. Interestingly, Baseball American slapped an “80” grade on G. Cole’s fastball today in their Top 100 reveal. Would that mean he’d likely average >95+ mph?
Comment by LuckyStrikes — February 19, 2013 @ 5:35 pm
I’m guessing you didn’t mean fringe MVP player for 40 and probably just meant fringy player for any roster. That’s a fair assessment though. I’m not sure, however, that judging a player’s production as a whole (No 80 tools, but a bunch of 60s and 70s equals an 80) is a fair assessment, but it’s a solid guess.
Silly me- Fringe MLB Player. I realize it’s not perfect since the 20/80 model is used to evaluate tools. I just found the distinction interesting when reading Baseball America’s Prospect Guide. I figured I’d throw numbers around.
Terrific piece, Mark. I really enjoyed this. Enjoyed the discussions in the comments it provoked as well.
Comment by William Tasker — February 19, 2013 @ 6:55 pm
From a scouting perspective, a lot of scouts will look at them from each individual position. It would have been more interesting to look at from each position, but space and sample (though I could have stretched it out to 5 years or something) made things a bit more difficult.
Comment by Mark Smith — February 19, 2013 @ 7:06 pm
Depends on how one looks at it. Usually, a scout will break it up into velocity, movement, and command with another grade for the pitch overall. Here, I was just looking at velocity.
Comment by Mark Smith — February 19, 2013 @ 7:07 pm
I’m considering doing another post similar to this rating other metrics (OBP, wOBA, WAR, etc.). Would people be interested?
Comment by Mark Smith — February 19, 2013 @ 7:08 pm
Thanks for the kind words, everyone. As for Cole, an 80 is put on his fastball because he can hit fairly regularly 97 and above. Whether or not it “averages” that is another story. Would be interesting to have minor-league PITCH f/x data. Pitchers also start to lose velocity after the age of 21-22, so by the time some guys get to the majors, the higher velocities have diminished a little due to age/use.
Comment by Mark Smith — February 19, 2013 @ 7:11 pm
I think it’s possible to be better than an 80, but it’s so rare anyway that it’s almost not worth doing anything else than just admiring it. Trout did a lot of things to impress last season, and his success on the bases was one of those things.
As for the steals, it didn’t really work out for some of the reasons you mention, but I included it to give a different look. Guys who are really slow won’t steal at all, and if a two players rate as a 35 and a 20, they’ll have 0 stolen bases because there’s no way to be negative. BsR is better because it starts taking into account other baserunning events that even slow runners can’t avoid.
Comment by Mark Smith — February 19, 2013 @ 7:15 pm
Yep. What Dan said. It was in consideration, but the needed sample for batted ball statistics to stabilize is something like 9 years.
Comment by Mark Smith — February 19, 2013 @ 7:16 pm
Max, what wilt said. His .254 BABiP is based on the last 3 seasons.
Comment by Mark Smith — February 19, 2013 @ 7:17 pm
You’re probably right. I’ll look into it.
Comment by Mark Smith — February 19, 2013 @ 7:20 pm
Yes, very cool stuff, Mark. I love the beer and tacos analogy from Dayn Perry at BP. Seeing articles that combine the two, especially with familiar names as real life examples is pretty rad. More, please!
Comment by bcp33bosox — February 19, 2013 @ 8:12 pm
This is great. We used the scouting scale this year in out draft kit to scale risk / reward for fantasy players. I use it all the time, because once you become confident with it it holds all sorts of utility. Thanks for running those numbers for us.
That’s not how you regress stats that have a small sample size.
Mike Newman has stated multiple times that his scouting contacts all claim Andrus as the best defensive SS in MLB. And it is rumored that the Rangers tried to get Simmons just to flip him to Arizona. Atlanta fans need to temper their expectations.
Comment by El Vigilante — February 19, 2013 @ 8:35 pm
It should probably be HR per batted ball, as you would in theory want to evaluate HR power completely independent of strikeouts.
I’ve done similar to this with minor leaguers. If you add HBP% onto the BB% you push the results away from 0% and you start to see more pitchers who are 2 to almost 3 standard deviations better than the mean. In essence it lessens the “right skewness” of the BB% distribution a little making it a bit more “normal”.
Comment by reillocity — February 19, 2013 @ 9:31 pm
Haha. Probably not. But when they give grades, they usually have an idea of what stats they’ll produce with those tools (or may produce, given development). This just gives us a more precise and current look.
Comment by Mark Smith — February 19, 2013 @ 9:41 pm
Peter Bourjos is an 80 defender, imho. You didn’t see him on the field last year but in 2011 he was spectacular on defense.
Mark, this article is beautiful. My only suggestion is to do a bit of research on non-normal distributions (chi squared makes sense with steals). Many recommended this on the steals, but i would also say to take a look at the rest (map real results against normal distributions first, this will get you most of the way there).
Funny how the scouts eyes and the statistical distributions are almost perfectly in line.
Putting this kind of work together with analysis over time could provide significantly improved forecasting methods. Very cool stuff.
This may have already been addressed, but I would strongly suggest you use sample quantiles rather than normal approximations so you don’t get ridiculous results like negative HR/PA.
Comment by guesswork — February 20, 2013 @ 1:08 am
Not trying to denigrate Atlanta fans specifically, but last off-season the consensus groupthink amongst Atlanta fans was that Jair Jurrjens was an Ace level pitcher and the fanbase in general couldn’t understand the rest of the baseball world’s lukewarm reception of his abilities. (This came about in discussing a possible Adam Jones to Atlanta trade scenario)I suppose it is why the term fan comes from fanatical.
I feel like part of the point Mark was making is that 80 grades are meant to be rare. Look at the hit tool. It is Miguel Cabrera. That is it. The power tool. Jose Bautista and probably Giancarlo Stanton. Really, for someone to get an 80 grade they are the best player with that tool of all the players in baseball. In general 80 grades are tossed around a little too liberally. 60 and 70 grades should be seen as very high praise as well.
Jerry Dipoto developed this years ago. He translated the 20-80 scale into performance, and even used OPS+ and ERA+ extensively in his metric, (as well as all the traditional).
It was a terrific tool for the scouts to really get a handle on what they were actually projecting.
It always amused me how many people on this very site criticized the heck out of Dipoto for some press conference comments he made about Joe Saunders and “winz” when he made the Haren for Suanders/Skaggs/Corbin trade.
Anyway, great article, and glad to see the concept advanced. But I couldn’t pass up the opportunity to point that things are not always as they seem to appear in this fabulous, mystical game we love.
Comment by shoewizard — February 20, 2013 @ 7:09 am
Very useful. I was surprised, though, to not see Aroldis Chapman in the relief pitchers for fastball velocity. At least by the popular conception, if he isn’t three standard deviations above average, I don’t know who would be… Is it possible that he doesn’t really use that express-speed fastball very much? The two games I saw he certainly did, though.
What Jason described was a shorthand way of describing the power tool. A lot of scouts don’t really think in terms of ISO, etc. HR will give you the basic idea of a guy’s power, but when scouts grade out power, they’ll take everything into consideration. It’s just a point of direction, essentially.
Comment by Mark Smith — February 20, 2013 @ 10:31 am
Again, these are based on averages. There are many nights where Chapman has more than enough fastball to grade out as an 80.
Comment by Mark Smith — February 20, 2013 @ 10:32 am
Home runs make it easier. Take a speedy hitter for example. If he is batting leadoff, he is probably going to be able to rack up some extra doubles and maybe triples that if he were batting say 7th, he wouldn’t be able to because runners would be on base in front of him that he couldn’t out run to gain those extra bases. That isn’t power at play, that is his speed at play. Home runs are home runs (assuming neutral field environments of course). While “doubles power” certainly can boost overall numbers, home run power exists all over the place, no matter the situation. Doubles power doesn’t.
I take it that Chapman (to name one) averages a lot less than 100+mph. On the other hand, it also surprises me not to see Kimbrel – I know his average is way up there, definitely not “only” 95mph. Is there a resource for finding the averages?
Comment by fishbait — February 20, 2013 @ 11:13 am
The Stubbs grade stood out to me as well. His MO around the league is plus range in CF but terrible at the plate where he K’s too much and doesn’t walk nearly enough. If he were an average defensive CFer I don’t think he would get the ABs that he gets.
This idea could make an already great article even better. Could you show a graph of the distribution of data points around the mean? And when you hover over a point, it tell you what player the data point refers. That would be cool. I don’t know if it’s practical to implement though.
Comment by indyralph — February 20, 2013 @ 3:17 pm
Would 2009/10 Ubaldo Jimenez have gotten an 80 fastball grade?
If you’re talking about the people who post on the AJC blogs then sure. Anyone who was even mildly sabr inclined knew that Jurrjens was riding the luckiest comet in the galaxy. As for Simmons, only time will tell my friend :)
This is the real divide between traditional and sabrmetrics – the skill vs tool divide. Kubel may be slow, but he could be smart about base running which improves his “speed” grade. His BA speed would be 20 or 25, but his fSpeed grade would be 40. Stanton is a 80 BA power, but not quite there for fPower. Bautista’s fPower is higher than his BA Power.
The hitting mappings work well, the pitching, not so much. If only the run values on types of pitches weren’t so unusable.
I never said it was. But some guys can add doubles/triples that others can’t. But they aren’t showing more power, just more speed. Maybe it is only 5-7 a year, but it is showing up and it isn’t actually showing more power, but more speed.
Has anyone at fangraphs tried to do an ‘all-time tools player’ write-up? I’m thinking Mays has to be at or near 80 in every tool, both from a ‘tools’ perspective and SABR. Bonds (with an average arm) and Griffey probably aren’t far behind. Young A-Rod comes to mind as well. It would be fun to see an ‘all-time’ ranking, based on the ‘sum’ of one’s tools (with a max of 400 – 5 x 80).
Hmm interesting. Maybe it would be better to take a 3-4 year peak for this kind of study? I think you’d want to measure a player when his ‘tools’ are in their prime, rather than having his final years drag down how good of a player he actually was.