THE BOOK--Playing The Percentages In Baseball

Thursday, June 25, 2009

Linear Weights, by pitch movement

Dave Allen strikes again, this time from Fangraphs:

Friday, June 19, 2009

More spray charts

Dave Allen gives us a bunch of spray charts.  The more I see these, the more I presume that breakdowns by individual batters is where the real stories will come out.  So far, we’re getting the right basis to whet our appetites. This chart shows the best spray angle off the bat to get maximum horizontal speed of the ball off the bat:

For RHB, you best spray angle for speed is zero to -30 degrees (i.e, from up the middle to the SS position).  And similarly for LHB, it’s at zero to +25 degrees.  It very close to being a mirror, but tt could be a small but real difference; most likely it is because the RHH prefers to pull the ball more because of the longer throw, and the LHH might be tempted to not pull as much because of the play to 1B.  I will also guess that the spray charts would be similar for a RHH with a runner on 1B or not, but for a LHH, he probably has a more noticeable difference in spray when a runner is on 1B or not.

Love this stuff!

Thursday, June 18, 2009

The best and worst 82mph fastball in baseball

Dave points out that it could belong to the same pitcher:

That’s his arsenal. An 80-84 MPH cut fastball. That’s it. He throws that pitch, and that pitch only, 99% of the time. He also has thrown this loopy 72 MPH curveball a couple of times this year, but practically every pitch he throws, in every game, is this low-80s fastball with some tailing action away from right-handers. I faced guys in high school who had better stuff than this.

However, whatever DiFelice does to his cut fastball, or however he hides the ball, or whatever voodoo spell he chants before he pitches, it works wonders against right-handed bats. Here’s his career splits since joining Milwaukee:

Vs RHB: .149/.192/.202, 120 PA
Vs LHB: .321/.357/.717, 56 PA

I presume it’s location, location, location?

Tuesday, June 16, 2009

Linear Weights… by batted ball speed

Jason sent me an email:

Hey Tango,

I was going to post this but couldn’t figure out how to put the chart into a post. I figured it would go best here: http://www.insidethebook.com/ee/index.php/site/article/the_hitf_x_database/

This is my first go at using LW’s, but I believe it’s correct. I am attaching the excel file if you want to take a look. I got the LW’s from here, but if there is a more up to date chart that I can use that: http://baseballanalysts.com/archives/2008/02/writing_about_t.php

If these results look good, I can also do it by hit launch and spray. Thanks

Results:
The harder you hit the ball, the better it is. Interesting that it doesn’t cross 0 until at least 90 mph.

hit_initial_speed     Average of LW     Count of Events
<50     -0.188     1358
50-55     -0.240     372
55-60     -0.207     456

60-65     -0.096     643
65-70     -0.062     783
70-75     -0.080     961
75-80     -0.116     1097
80-85     -0.094     1509

85-90     -0.056     1787

90-95     0.047     2115

95-100     0.208     2086

100-105     0.473     1420
>105     0.535     457
Grand Total     0.030     15044.000

Bias in PITCHf/x?

Self-named replacement-level PITCHf/x-ers Harry Pavlidis (*) points out:

Ken Takahashi and Brian Stokes both pitched in relief of Santana on Sunday, and we happen to have another game (May 11) from Citi Field where all three pitched.

Talk about the fantastic luck to find the same three pitchers who pitched in two different games, with the game we are most interested in be one of those games.

(*) Harry three months ago called himself a replacement-level PITCHf/x-er.  I have no idea if it was true then, and if it was, it simply shows what dedication, inspiration and hard work can do to turn such a baseball fan into a quality analyst.

Sunday, June 14, 2009

100 pitches, 5 innings, 5 walks, batter is at 3-0: what do you do?

Batters swing in a typical situation just 8% of the time on 3-0.  So, you should already have a small strike zone to begin with.  The debate is how much to shrink your strike zone, if you should shrink it even tinier in the particular situation Phillips found himself.

Here’s what Brandon Phillips saw in the fifth inning:

I’m thinking he should only swing if it’s right down the middle.  Others may say he should keep his strike zone at his standard 3-0 strike zone.

Here’s the fastball he swung at:

Identical pitch to pitch #2.  His reasoning is even more suspect:

“Honestly, in that situation, why wouldn’t I swing on 3-0?” Phillips said. “We only had two hits at the time. Our offense stinks right now. I respect my teammates and they know it, too. They see what’s going on. We really haven’t faced anyone overmatching us. I can understand if we’re facing No. 1 pitchers but we’re not. Our team is too good to get out by the pitchers we’re facing. We should be undefeated on this road trip.

“We’re struggling. We’re not getting anybody on base. We haven’t been hitting with people in scoring position. I had an opportunity. I swung on 3-0. To tell you the truth, I didn’t even look down there. I was trying to make something happen.”

(Hat tip: Peter)

Friday, June 12, 2009

PITCHf/x: Blogger v Blogger

According to blogger Matt Sisson, Wang’s (horizontal, meaning side-to-side) release point shows a 1 1/2 foot difference:

According to blogger Harry Pavlidis, the poor data quality of this subset of PITCHf/x data is almost certainly the reason:

As Harry said:

Instead, I remind the well meaning people who like to use the PITCHf/x data to study Wang’s release and movement: If you’re going to publish data and “conclusions” take the time to understand what you’re working with. Doing otherwise is irresponsible.

Thursday, June 11, 2009

Launch angle, by contact point in the strike zone

Dave Allen gives it to us (chart from the perspective of the catcher):

Remember that we learned in the other thread that the optimum launch angle is highly dependent on the kind of hitter you are.  Ryan Howard for example would likely have red blobs in the middle of the strike zone, and it would be blood red above the letters.

Wednesday, June 10, 2009

The HITf/x Database

This thread is reserved only for sharing your work, in creating a database, loading data, creating tables or views, creating equations to use, etc, etc, etc.  You can ask direct, specific questions.  Consider this thread to be a workshop.

More general posts can be made at the Summit thread.  Consider that thread to be a welcoming type of thread.

Specific discussion of results can be continued at the great thread that was started from Harry’s article on launch and spray angles.  Consider that thread to be alot of yapping, some results-based posts, etc, like a classroom discussion.

Tuesday, June 09, 2009

When is a line drive a flyball?

Harry gives us a great look into HITf/x:

I have no doubt that if you look at where a line drive intersects with the ground ball, and where the line drive intersects with the flyball, that the outcome of those “line drives” will look alot more like GB and FB, respectively, than line drives. 

Indeed, we don’t even need to classify things as GB, LD, FB, and Pops, if we have the launch angle.  The ONLY reason we bother with the classification is that it’s easier for us to process and “see” the data.  This is similar to trying to figure out if a ball was pulled, up the middle, or opposite hit.  It all depends on where you draw the line.  And once you do, you have to ask why have a “line” as some sort of demarcation point.  UZR is better than ZR for the exact reason that it doesn’t treat all balls in the “zone” as if it’s one monolithic zone.

The horizontal angle that Harry shows is the spray angle.  Indeed, I would recommend that we call this the spray angle, since this tells us exactly what it’s doing.  I had to think about about what a “horizontal” angle is.  I don’t have to think about the spray angle.  The same thing for vertical.  I’d prefer to call it the launch angle.  Horizontal and vertical is more technical and less descriptive.

***

Harry: for your SLG axis, stop it at around 1.000, and show “breaks” for those that go above.  There is a huge difference between a .400 and .500 SLG, but that is almost completely invisible when the scale of the SLG axis goes to 3.0 I understand why you did it this first time.  I DO prefer that you keep the same scale, as you did.  This way, we don’t need to readjust our scale for every picture.

Saturday, June 06, 2009

PITCHf/x Summit: Introducing HITf/x

Register now:

The 2nd PITCHf/x Summit brings together analysts and interested parties to learn from each other, better understand the possibilities, and provide a forum for discussion about PITCHf/x and its contributions to the sport of baseball. Attendees will participate in demonstrations, discussions, and panel presentations…

Hat tip: Mikr

Friday, June 05, 2009

Linear Weights by Strike Zone

Max is at it again.  Here’s one of his many charts, which focuses on the horizontal location of the curve ball, and what its run value is:

The smaller the number (the more toward the left), the better for the pitcher.  So, if the pitcher is looking for a 10-inch vertical zone to throw his curve ball, he should throw it in the center.  But if he does that, his margin of error is very very small at the top end.  If you expand that to a 20-inch vertical zone, then it’s a bit above the center, to below the knees (outside the strike zone).  The “hanging” curve ball that is often mentioned (a strike, but hitting the top of the strike zone) is equivalent to throwing a curve ball halfway between the knees and the ground.  That’s how bad it is to hang one.

By the way, those run values don’t make any sense to me.  The average run value of a called ball is something like +.08 runs (and the called or swing-and-miss strike as well, but negative).  But, a curve ball thrown right in the dirt, according to Max’s chart, is worth +.02 runs.  This means that batters must be swinging at ALOT of curve balls in the dirt, something like 35% of the time.  Doesn’t sound right at all.

Scaling PITCHf/x

Someone asked me how I scale a pitch from PITCHf/x to a common scale.  This is my response:

Read More

Thursday, May 28, 2009

PITCHf/x … Tracers!

One of my favorite parts of the Bill James books was these “tracers”, where Bill’s assistant (like Rob Neyer) would investigate some claim, and see if it was true.  Dave Allen has gone all high-tech on us and looks at Milton Bradley’s claim that his strike zone has been expanded by the umps:

These contours are closed lines such that a pitch inside the line is a strike 50% of the time or more and a pitch outside the line is a ball 50% of the time or more. Here you can see how the outside edge of the strike zone is shifted farther outside to Bradley as a lefty, as is the case to all LHBs. The inside edge of the pre-2009 and 2009 zones are almost exactly the same. Up and outside the pre-2009 zone is larger, but down and outside the 2009 zone is larger. As a whole the two are almost exactly the same size.

Friday, May 22, 2009

Effective fastball-changeup combo

Dave Allen shows us the run value of the changeup, based on the difference in speed to the preceding fastball:

On average, the gap should be about 8% lower, which means 7mph slower.

Monday, May 18, 2009

Dave Allen meets Dave Cameron

In the spirit of previous winners of “my favorite saberists” (Dan Fox, 2006; John Walsh, 2007; Mike Fast, 2008; no offense to the dozens of other super-high quality saberists that I rave about… I should probably give out awards every quarter), Dave Allen is pulling ahead with his relentless attack on multiple sites, using PITCH f/x, and tying Linear Weights run values to pitches.  This time, he talks with Dave regarding the particular platoon splits based on the two-seamer and sinkers:

This analysis includes two-seam fastballs and sinkers, which are hard to differentiate with the pitchf/x data. These fastballs are typically low in the zone, rise less than other fastballs (so the appear to ’sink’ to batters) and tail in to same-handed batters 10 inches on average, more than any other pitch type. I am not sure why they show such a large platoon split, but I think the large horizontal movement probably plays a big role.

No matter the reason, as you have said before righty dominated lineups make mediocre two-seam throwing pitchers look amazing.

This is the kind of useful, actionable analysis that is the future of PITCHf/x.

Friday, May 08, 2009

Wasting pitches

Dan looks at what happens when you waste a pitch once the pitcher is up 0-2:

In plate appearances where the 0-2 pitch was not wasted, the aggregate wOBA was .226. In plate appearances where the 0-2 pitch was wasted, the aggregate wOBA was .245—substantially higher. Just to be clear, these numbers include all plate appearances that pass through 0-2, not just those that ended at 0-2 or the next pitch.

I have the wOBA by count, for 1993-2008.  This is for all PA.  At 0-2, the pass-through wOBA is .220.  At 1-2, the wOBA is .248.  The 1-2 includes: SSB, SBS, BSS.  That is, it includes those PA that have the potential to be part of the wasted-pitch sequence, and the other sequences.  The gap between 0-2 and 1-2, among all PA, is 28 points (.248-.220).  I have over half-a-million PA in my dataset.

Dan looked specifically for the wasted pitch on the 0-2 count, compared to the non-wasted pitch.  (I’m sure his dataset is within the last year or two, so be careful about the comparison of the absolute numbers.) He shows the gap to be 19 points (.245 minus .226).  So, we see that the wasted pitch does have *some* impact, not the least of which is that the batter might chase it.

I would not be surprised that if you look at how a batter got to the 1-2 count (the called ball was the first, second or third pitch), that it’s irrelevant, and that the wOBA is the same in all three cases.  That is, wasting a pitch is a foolish thing to do in that particular at bat.

However, suppose that a batter “knows” that a pitcher will not waste a pitch.  If he knows the pitcher’s waste-rate is much below the league average (for a pitcher of his caliber), then he would approach that pitcher differently, and maybe get a higher wOBA on the 0-2 count.

Furthermore, what if wasting a pitch confers some advantage on the pitcher in future at bats in that game.  That is, the pitcher will waste a pitch inside to show the batter he’s not afraid of him and will take things in his own hands if need be (Pedro’s favorite strategy).  Or, if he wastes it wide, he might show that he might be a bit wild.  In either case, the pitcher might get into the hitter’s head.

So, we can also look to see how the batter hits in next at bat that game, against that pitcher, if he had a wasted-pitch at bat.

Love this game theory stuff…

Fielding positioning, part 1

Max starts to explore positioning fielders based on the pitch location, and the sprayability of the hitter.

Friday, April 24, 2009

Linear Weights, by pitch location

Max now is writing for Hardball Times, and he’s got an easy-to-understand article, with lots of great graphs.  Here’s the one for the slider:

What do we see here?  Never, ever throw a slider down the middle.  It is as bad as throwing a pitch way off the plate. 

Now, here’s the one for the curve ball:

Never ever through a curve ball inside on a batter.  From the middle of the plate to way off outside of the plate, the curve is very effective.  Throw it inside over the plate for a strike, or inside for a ball, and it has the same bad impact.

This is fantastic stuff!

Remember, Linear Weights is like Donuts (there’s nothing they can’t do).  When trying to come up with this types of heat charts, try to think in terms of Linear Weights.

Thursday, April 23, 2009

HR distribution by pitch location

Great work by Jonathan Hale.  I think I would prefer to see the same charts per swing, and not per BIP+HR. 

Also, since HR rates vary greatly by count, a control by count might be in order as well.