(This post expands on Hitter Plate Discipline from this post.)
I emailed someone the following research outline for my idea on measuring Hitter Plate Discipline with granularity, instead of the traditional binary way (swing at ball bad, swing at strike good). Thought I’d share if anyone’s interested in tackling it. Go for it on your own, or include me if you want. I just want to read the results.
—- —- —-
Might be worth collecting research done on any of these questions, and reading through it before starting.
1. We’ll need a way to measure the value of a pitch without a swing (should we use the actual ball-strike call, or a probabilistic model of what tends to be called a strike? I lean the latter, although maybe we start with the former)
a. location is most important
b. ball-strike count is probably very important (the strike zone changes size, and the value of a ball/strike changes)
c. pitch type likely matters for what gets called a strike/ball; medium importance?
d. pitcher/catcher framing ability probably matters; small importance? (and difficult)
e. base-out state probably matters, not sure to what extent
2. we’ll need one of these two things, maybe both:
a. a way to measure the expected value of a pitch with a swing
— easiest would be to lump all hitters together, adjusting for location/count/pitch type/framing/base-out state
— harder, but likely important, would be to adjust by hitter, and maybe pitcher; my guess is we’re starting to cut the data really fine at this point, so maybe we start with a league-average profile, and modify a bit given whatever we can find out about specific hitters/pitchers. Things like inside/outside hitter, fastball hitter, contact rate, power, etc. Maybe we take exact data for a hitter and/or pitcher and “regress” it using league-average rates, with X pitches per micro-bin.
b. a way to measure the actual value of a swing; easier (although insert hit f/x pipe dreams in here)
— I know this has been done, and is “just” pitch-level linear weights
3. Once we have 1 and 2, we’ll need a way to combine them to judge discipline. Probably something like the difference in expected value between what they did (swing/take) vs what (likely) would have happened if they did the other one. Either report a mean score on a per-pitch basis, or maybe show the distribution to see who makes some really stupid decisions vs who consistently makes mildly poor ones? Can slice and dice this in tons of ways — who has good/bad discipline early in the count, on sliders, on high fastballs, etc.
4. Can we then combine this plate discipline metric with other hitter skills, like power, contact rate, hit rate, maybe even distribution of pitches seen? Some of this might already be worked into step 2, dunno.
Below is a list of research and/or article ideas I’ve had. I don’t have the time, motivation, or ability to tackle them, bud I’d love to see them happen. Feel free to tackle them yourself. If you want to brainstorm with me, I’m willing to collaborate — try email (skyking162,gmail) or twitter (Sky_Kalkman). Even if not, please ping me when you post something using one of these ideas, as I probably want to read it!
This post exists just to anchor some scenarios for Mike Trout’s value over the next six years, given the rumor of a possible 6/$150 deal. It’s very important to note that as a third-year player, he’ll earn basically league-minimum this year, is at the mercy of the arbitration process the next three years, and the final two years would be his first two of free agency.
Using @leokitty's Hall of Fame ballot tracker from the 2013 vote (public ballots only), I mapped out common voting patterns. Essentially, if two players tended to appear on ballots together, they appear closer together below, and if ballots tended not to include one player when another was included, those players appear farther apart. (Only players with more than 1 vote were included in the analysis.)
[Click here for the full size version.]
Some obvious patterns emerge. Guys to the left are PED suspects, guys to the right aren’t. The vertical axis is less obvious — any ideas? Players who got more votes tend to be more towards the middle (which is obvious when you think about it.)
We could use this map to identify different types of voters. For example, “Those who don’t overreact to PED allegations” or “Those who love the 80s so much they still wear Zubaz.” In fact, we can see who falls into each of these segments by adding the voters to the map:
[Click here for full size version. Seriously.]
You can interpret this map similarly. Two voters appear closer together if their ballots tended to include the same players and farther apart if they voted for widely different players.
(Interpreting which voters voted for which players isn’t as obvious, however. For example, if someone voted for Rafael Palmeiro and Kenny Lofton, they’d appear in the middle — an average of the two locations — not near either guy. Other quirky stuff can happen as well, thanks to the two-dimensionality of this analysis.)
Do you notice anything interesting going on, perhaps patterns of which types of writers tend to vote for which types of players? Maybe older newspaper guys tend to vote one way, while younger national guys go another?
—- —- —-
(If someone reminds me, I’ll run these again after the 2014 vote.)
General consensus is that Tom Glavine in a sure-thing Hall of Famer, while Mike Mussina falls a bit short. Apparently Mussina is seen as good-but-not-great. This is absurd, likely due to a combination of the following: he pitched during the highest scoring era ever, in the American League, in hitters parks, and failed to attain that oh-so-important 20 win season until his final year of his career. (That final year, by the way, is the best final year of any pitcher’s career since 1902 besides Sandy Koufax, using bWAR.)
In order to defend Mussina’s honor, let’s do a little comparison exercise by running through Tom Glavine’s best seasons. He won two Cy Youngs. The first was in 1991 with a 153 ERA+ in 247 innings. That’s pretty similar to Mussina’s 1992, when he had a 157 ERA+ in 241 innings. Those seasons cancel out. Glavine’s second Cy Young came in 1998 with a 168 ERA+ over 229 innings. Mussina had a 164 ERA+ in 1994, but can’t match the innings (only 176 — yay, strike), so put that in the Glavine-only column.
Tom Glavine finished second in the Cy Young twice, in 1992 and 2000, and third twice, in 1993 and 1995. Mussina can match those seasons individually (1997, 2001, 2000, 1999) and in aggregate (averaging 223 IP and a 134 ERA+ to Glavine’s 226 IP and 133 ERA+).
Now, Tom Glavine had three seasons better than any of those second and third-place Cy Young seasons, in 1996, 1997, and 2002, where he averaged 233 innings and a 143 ERA+. Mussina has a similar set of seasons in 1994, 1995, and 2003, although his innings average a bit less: 204 IP with a 144 ERA+.
So what’s left? Glavine has 1240 more IP with an ERA+ above 100, averaging a 115 ERA+, while Mussina has 1231 averaging a 117 ERA+. Again, very similar.
As for the below-average seasons, Glavine has 907 innings averaging a 90 ERA+, while Mussina has 496 averaging a 94 ERA+.
Stepping back, these two pitchers had very similar careers. Tom Glavine has one additional excellent season, 90 more innings of very good pitching, and 400 more innings of below-average performance. Everyone draws their line somewhere different, but there’s not much gray area in between these two guys where that line can fall, and if Tom Glavine is such a sure thing, that line probably falls well below Mike Mussina.
—- —- —-
If you’re more of a visual person, try this graph of each pitcher’s seasons, sorted from best to worst. I used bWAR here, which is a handy shortcut for combining IP and ERA+ (plus some other minor adjustments.) Glavine has the single best season, but the rest of Mussina’s top seasons outpace Glavine’s. For what it’s worth, Mussina has more career bWAR, 83 to 74.
[Click here to view full size image.]
—- —- —-
* Yes, I averaged the ERA+’s the right way.
I don’t know. I’m not actually going to answer that question. But here’s what I am going to do: estimate to what degree the Rays ballpark, defense, and pitch framing abilities might have helped Price’s ERA. There are many reasons my final estimate is probably wrong, which I’ll discuss later, but the biggest one is that I’m going to push the extremes of how much these things matter.
Put it all together, you get an ERA bump of .80 pts of ERA aggressively, and .40 pts more realistically. In other words, David Price might be more of a 3.43 ERA pitcher and less of a 3.03 pitcher.
Ok, so where is this analysis going astray? I can think of a few things:
Random note: Baseball-Reference neutralizes David Price’s 2013 ERA of 3.33 to 3.93 when pitching in a neutral 2013 AL environment. That’s a huge shift, and I’m not sure how they get that number, especially given a non-extreme 95 park factor.
* Ben Jedlovek of BIS noted that the Rays defenders didn’t rate amazingly well in 2013, but the team as a whole was near the top in shift runs.
This article is a proof-of-concept. I’m going to outline a methodology, then use it to write about a few interesting storylines. I’m hoping you will improve on the methodology, and use it to tell better stories.
Everything below follows from the question, “How much of Elvis Andrus’ offensive production is from walks?” His batting average isn’t spectacular and he hits for no power, so why walk him? And yet, he still walks a fair amount. Weird question, perhaps.
My approach is to add up the linear weights values of basic, positive hitting events (BB, HBP, 1B, 2B, 3B, HR), each multiplied by the frequency they occur. Then find the percentage of the total for each event. And there’s one twist — I subtract the value of an out, a la wOBA, except I subtract the out value according to total runs created, not the out value relative to average.
Here’s an example. In his career, Elvis Andrus has 232 walks. Each walk is worth .33 runs above the average offensive event (including outs). The out is worth -.1 runs, so each walk nets out at +.43 runs. 232 times .43 = 100 runs created via walk. Repeating the process for other major events yields 9 runs by HBP, 329 runs by single, 86 by double, 30 by triple, and 21 by HR. That’s 576 total runs, so the 100 runs from walks is 17% of the total package.
Is that high? I don’t know, we need some context. In the 2012 American League, walks were 17% of offensive production. So, Andrus is typical. Where he isn’t typical is in his lack of power. Whereas the 2012 AL has a full 20% of production from home runs and 41% from singles, Andrus is at 4% and 57%, respectively.
How about some other examples? Barry Bonds, in 2004, hit .362/.609/.812. Those 232 walks (Andrus’ career total, by the way), account for 42% of his offensive production. Singles, 19%. And home runs, 29%. While Bonds certainly had a good eye, his walking skill wasn’t necessarily greater than his home run skill. It’s just that pitchers chose to get beaten by his walks rather than mess with his power. (You know this, I just wanted to point out that I’m measuring what happened, not trying to describe skills. Accounting, not prescribing)
Speaking of power, we can split offensive production between hitting for power, hitting for average, and walks/HBPs. For power, I like to subtract a hitter’s production pretending all hits were singles from his total production across all hits. Sort of like isolated power, but with linear weights. For Miguel Cabrera in 2013, his production is 23% due to power. To measure the hit tool, I pretend all hits are singles. Cabrera 2013 is at 60%. And walks/HBPs make up the other 16%. As a comparison, Joey Votto 2013 is at 14% power, 58% hit tool, and 26% BBs.
Finally, because no good saber article is complete without a list, here’s a breakdown of some other 2013 hitting profiles, picking out the highest and lowest rates in each of the three categories. I really enjoy comparisons like Sal Perez and Alex Gordon. They have somewhat similar relative profiles, but Gordon has been way better across each category.
Player Hit Pow Walks Jeff Keppinger 87% 9% 3% Salvador Perez 81% 14% 5% Alex Gordon 72% 18% 10% Adam Dunn 42% 33% 25% Josh Willingham 42% 23% 35% J.P. Arencibia 62% 34% 4% Chris Davis 51% 29% 20% Bryce Harper 51% 27% 21% Dustin Ackley 71% 8% 21% Ben Revere 81% 6% 13% Shin-Soo Choo 48% 20% 32% Ike Davis 54% 16% 29% B.J. Upton 52% 19% 29% Manny Machado 71% 22% 7% Adam Jones 69% 24% 7%
Ok, your turn. How could you improve on this?
Account for additional skills, like avoiding strikeouts, BABIP, baserunning, etc…
Do something with IBBs (I suggest ignoring them or else giving them the average runs/PA value of a hitter overall.)
Get the whole “subtracting the value of an out” thing right — maybe find a number between -.1 (total runs created) and -.3 (runs above average) that represents runs above position-neutral replacement level?
Customize the linear weights for year, league, and park.
Do something similar for pitchers.
Produce a file with values for tons of players and seasons and careers.
Create a spreadsheet tool to plug and chug any values.
Create a useful, intuitive visual way to represent this info (personally, I like the trio of hit tool, walks, power.)
To be honest, I didn’t really figure out any of this methodology. I mixed some ideas from Colin Wyers, Patriot, Matt Klaassen, and Lee Panas. They get any credit this idea deserves.
If you don’t care about my fantasy team or discussing Scoresheet strategy, then don’t read this.
When I managed Beyond the Box Score, I weasled my way into a media Scoresheet league. There’s been a bit of turnover, but I know a lot of the guys in the league via twitter and there aren’t any idiots — something that’s not easy to do with 24 teams.
Scoresheet is a “live” simulation fantasy format — kind of like Strat-o-Matic, but using in-season weekly stats. It’s real baseball: OBP matters more than AVG. Defense matters. Saves don’t matter. It’s not perfect, but it’s my favorite format. (I hate the keeper structure, though.*)
In previous years, I kept only 3-4 players. The limit is 10, but for every player below 10 kept, you get an extra draft pick at the front of the draft. This year, however, I kept the full ten. I’m not sure that was smart, but I did:
I love the first three guys, especially Zobrist, who’s going to be my starting shortstop. Span has a sweet D rating. But then I’m less excited— Werth’s platoon rating hurts him against righties, which isn’t a great fit for these leagues. Ellis, as a catcher, isn’t a full-timer, and also hits better against lefties — not a great fit considering I kept Travis D’Arnauld with a minor league slot. And the pitchers, well, I like them more than most, but they certainly aren’t workhorses or aces.
Befire my first draft pick, I made two trades. First, I dealt Span and a late pick for Kevin Youkilis and an end pick. I needed a big bat at 1B or DH, and I love him in Yankee Stadium. Then I swapped McCarthy and a later pick for Carlos Beltran and two end picks. Boom, offense greatly improved. I really like McCarthy’s skills, but don’t trust his ability to rack up innings and don’t love the change in parks (Scoresheet adjusts for league, but not park.) I consider this trade a win. The Span for Youkilis deal was more even, but I pulled the trigger because there were more decent CFs remaining than big bats.
Headed into my first pick, my open positions were 2B/SS, CF, starting pitching, and an entire bullpen. Plus depth — Scoresheet’s replacement level is like a .200 hitter/8.00 ERA pitcher, so having backup PAs is critical.
I quickly nabbed Colby Rasmus, who should be only a small downgrade from Span vs righties (DeAza and Pagan were my two other options and both were picked before I had a chance at them.) Then I made the one pick I really regret: Jeff Niemann. I’ve always had a crush on him, but didn’t realize his velocity was down so much. 12th round (2nd round after keepers) was too early for him, and I didn’t have a pick in the next two rounds from my previous trades. Oh well.
In the 15th I grabbed Travis Hafner for offensive depth and Scott Feldman for the rotation. I really wanted a 2B/SS here, but couldn’t pull the trigger. Daniel Murphy, Jeff Keppinger, Omar Infante, Maicer Izturis, Gordon Beckham, and Chris Nelson were all queued up, but went before my pick. I probably should have taken Zach Cozart, but I still like the Hafner/Feldman picks.
I got my 2B in the 16th round: Kelly Johnson. Don’t love him, but the rest of the options were even worse. He has no value vs lefties, but that makes his production vs righties better, a nice tradeoff. Definitely a position I’ll be looking to upgrade via trade.
The 17th round brought Chris Heisey as CF platoon-mate for Rasmus and general OF depth, plus Alex Rodriguez. ARod should be useful for the playoffs, either on my team or as trade bait.
My two round 18 picks are crapshoots: Faux-sto Carmona and Scott Kazmir. The first is completely motivated by the Rays track record with Peralta/Farnsworth/Rodey and their lack of a track record signing starting pitcher free agents. Kazmir’s velocity is back, although control is also a big issue for him. If one turns into a 3rd/4th starter, I’ll call this round a win.
Rounds 19-21 bought on Mike Aviles (platoon mate for Johnson), Kyle Crick (young pitching prospect who will be trade bait), and Jason Bay (a big bat flier). Oh, and I nabbed my first relief arm in Casey Janssen.
I hate relievers. Well, I hate valuing any specific reliever. A deep bullpen is important in Scoresheet, especially with my iffy rotation. I’d rather have a ton of 3.50 ERA relievers pitch innings 6-9 than 3.75-4.25 ERA starters. I made a lot of trades where I acquired late-round picks (I have 8 of the last 23 picks in the draft). These will be bullpen arms. But it’s nice to have a couple high-end relievers, too, and if healthy, Janssen would have gone higher.
That’s everyone so far. I need some depth and a lot of bullpen arms, but I also have some picks to spare on prospects and breakout candidates. I’m very happy with my lineup — while lacking a stud hitter, it has both OBP and power, with good depth and lots of flexibility. My defense is no worse than average. The starting rotation is meh, but not bad for lacking an ace. I have the pieces to trade for a pitcher mid-season, if I’m in the hunt.
One lesson I’ve learned is that I jump at underrated older players (call me Scoresheet’s Brian Sabean.) While a solid strategy for current-season succss, it helps explain why I lack superstar keepers. Old guys don’t become stars. You have to take fliers on younger guys for that.
Definitely interested in feedback, bring it on.
* Or league punishes the value of young players compared to veterans. You can keep prospects for next to free until they lose rookie status, but then their cost is the same as a veteran: one of your 10 keeper spots. Most youngsters aren’t stars immediately. It’s worth keeping a young Matt Wieters with a $1M pre-free agent contract, but not a $15M one, as an analogy. And since there’s a soft cap, it’s only worth keeping a player if they provide value beyond who you could nab in the draft. Except for my top three guys, I would be ok throwing my keepers back and choosing from the available pool. Therefore, acquiring studs (often through rip-off 3 for 1 deals) is a priority. I’d rather have a contract system, or a hard keeper limit (so that having more mediocre keepers has value). Of course, I don’t dislike the current system enough to leave the league.
[I wrote this for me, but published it. And it’s not about baseball. You’ve been warned.]
I just finished the fourteenth and last book in the Wheel of Time epic fantasy series. I started the first book about 12 years ago, so it’s been a long, sometimes frustrating journey. (When you wait three years for the next book to come out and *nothing happens*, that’s frustrating.) It’s not a series for everyone, and I actually enjoyed listening to the books during my commute more than reading them (there’s too much descriptive prose for my tastes, so I’d often skip ahead when actually reading.)
Anyways, this is the first time I can remember getting emotional when finishing a book series. It’s happened many times for a tv series, but never books.
* Let’s Pozterisk my most emotional tv series endings (not the best endings, but the ones that most hit home that the series was over):
Finality reminds me a lot of graduating high school, breaking up with a long-term girlfriend, moving cities/jobs, etc. While the next step might be exciting, it’s a complete unknown, and you’re forced to give up the status quo. It’s the end. You have to start over. What was won’t be any more. There’s probably a good Shakespeare quote to insert here about fleeting moments.
The Wheel of Time has its issues, but I enjoy epic plots, and the way this epic reduced down to the fates of a few individual characters at the end (the epilogue, to be exactly) really left an impression. One most of all.
In the epilogue, it’s revealed that Rand (the main protagonist) switched bodies with one of the bad guys. Everyone except about five people think he’s dead. He no longer has his magical abilities. He no longer has an obligation to save humanity. He’s starting fresh, but coming off a pretty stressful three (?) year period where he conquered countries, risked his life daily, lost his hand, dealt with two un-heal-able super-wounds, killed really powerful baddies, plotted to kill Evil itself, and eventually came to understand everything about the nature of the universe itself.
I mean, he’s probably relieved, but what a downer — "it’s about the journey, not the destination" and all that. What do you do next? How do you find meaning in anything else? How could anything else seem important?
Back in the real world, I wonder if I’ll ever get to a point in my life where my best work and my most challenging activities are behind me. Can you adjust? Can you pretend your current state isn’t a letdown? Can you learn to enjoy the destination? Of course, to pull from tWoT, maybe it’s not the end but just an end. There are always more beginnings.
On a more right-now time scale, I suppose it’s time to find a new book series.
/meandering thoughts, catharsis achieved