THE BOOK--Playing The Percentages In Baseball

So we might expect that for starters (maybe not a full 1.5, but something), but the smaller jump for relievers is a surprise.

Assuming the pitch f/x wasn’t running hot of course…

WRT #1, how do fastball speeds vary for relievers based on days of rest?  Everyone had at least one day in the ASG, and knew they would be limited to one inning and have a day off on Wednesday (and many Thursday as well).

Great data.  I seen to remember another study (also by Mike?) that reported a much smaller velocity difference for pitchers as starters and relievers.  I think that study was problematic, because pitchers often relieve when they are older and/or less effective.  This gives us a much cleaner look at what a starter can do when asked to go just one inning.

Brad/Patriot, I address some of that in my article.  I didn’t find any convincing evidence that PITCHf/x was running hot, but it’s not completely conclusive.

My previous research suggests that starters throw about 1 mph faster in relief, although there are huge error bars around that, and I don’t know how well that applies to guys that can already throw in the mid nineties.

Also, Josh Kalk did some research about fastball speeds based on days of rest for relievers here:
http://www.hardballtimes.com/main/article/do-relief-pitchers-suffer-from-pitching-back-to-back-days/

It might be interesting to see how the pitchers offspeed stuff compared velocity wise to their season average velocity…

If you have a starter making an emergency relief appearance, he’s likely going to throw just as hard as he’s done in teh past.  And if a reliever has an expectation that he might still pitch in a day or two, he’ll again not overthrow.

But in an ASG with all the off days and the guarantee of limited innings?  So, I think this was a pretty good lab experiment the players provided us with.

Also, Mike, can you check on movement?  As you know, I believe in the trading of the three principles: speed, movement, location.  And that you can increase speed at the expense of something else.  If the pitchers had less movement on their fastballs, then this helps my argument.  If they didn’t, well, I still have location to fall back on.

I would think in terms of a pitchers fastball, movement and velocity have a positive correlation for an indivdiual.  Whatever he’s doing to throw harder should also generate extra RPM’s on the ball.  Yes, there is less time for the ball to move, but the last bit of research I saw suggested the extra spin outweighs that factor.  I’d link it if I remembered where to find it…

I agree with B/7. (I have a Bingo!  Or, you sunk my cruiser.)

Tom/6, looking at movement across a population is more difficult than looking at speed because what movement means is so specific to each pitcher, but I’ll see if I can look at a few pitchers from the ASG.

Looking at Price, Verlander, Johnson, and Wainwright--the four pitchers with the biggest speed bumps from the regular season--their spin movement on their fastball seems about the same as the regular season, within around 100 rpm either way.

I suppose there is enough wiggle room there that you could still think whatever you want, but given that I don’t buy the physics behind what Tango proposes, I don’t see enough smoke to go looking closer for the fire.

If you look at all the fastballs thrown in the All-Star game relative to the ASG average for that pitcher, one extra mph cost the pitcher 0.1 inches of movement, with a correlation coefficient of r=.08.

By comparison, our average random measurement error on the spin movement of a single fastball is a little over 2 inches, which for the 10 fastballs thrown by the typical pitcher in the ASG, would mean that we would know their average fastball spin movement for the game within +/- 0.7 inches.  (Similarly, for the speed, we would be within +/- 0.3 mph.)

Last night I happened to check the home/road Pitch F/X velocity splits for the Angels.  For pitchers with at least 50 of a single pitch type (just using Gameday classifications) in both the home and road samples, the average difference for a pitcher was .8 mph slower in Anaheim than on the road for pretty much all pitch types (a couple were a tenth or two higher or lower than that, but the differences were all around .8).  That matches what Mike reported in the article, of course, but I thought it was interesting that I was looking at that exact thing last night.

Part of the gain could also be that pitchers were going all out to light up the guns just because it was an All Star Game and they could showcase their stuff without too much worry about results.  I don’t know how much, if any, effect that might have over them knowing they have just 1 inning sandwiched by rest days.

Given that we find somewhere around a 1 rpg difference between when a pitcher starts and when he relieves, and given that that the times through the order advantage is probably only .28 runs (8 points in wOBA) for the relief appearances, we still have .72 rpg or so to explain.  According to Mike Fast, a 1 mph change is worth between .25 (starters) and .45 (relievers) rpg, so we would expect the difference between starting and relieving to be more than 1 mph.

I think a velocity increase of about 1.5mph could explain the rest of the reliever advantage.  Note that Mike found an effect of .45 rpg among relievers.  If we apply that, a gain of 1.5 mph would account for about .7 rpg, just what we’re looking for.

I think the reliever velocity-to-runs multiplier, which is higher, is probably the relevant one.  I can see two reasons it might be higher:  1) a marginal gain is worth more at higher velocities (and relievers throw harder in general), and 2) high velocity pays off more later in the game, because hitters are used to a lower-velocity starter (essentially, magnifying the 1st-time-thru-order effect).  Since the starters who become short relievers are mostly hard throwers, both of these would apply to the pitchers we’re looking at.

Finally, it seems likely that pitchers who actually transition to short relief are those pitchers who especially benefit from the switch (or at least, excludes those who can’t throw faster in relief).  So the role-changers might gain something like 2 mph or more.

I thought for sure the gun was hot when I saw Wainwright dialing it up about 5 miles per hour faster than normal, and throwing his cutter at his fastball speeds when he struck out Hunter. His curveball was coming in at around 76-78 mph which is 2-4 miles per hour faster than normal. I’m not sure what would cause that to occur.

CJE/14, if that were our only data point for the game, that would probably be the most reasonable conclusion.  However, the pitchers that immediately preceded and followed Wainwright in the game, namely Drew Bailey and Rafael Soriano, did not display the same speed boost.

Bailey threw 95.1 mph, compared to a season average of 94.5 mph.  Soriano threw his four-seamer around 94 mph, compared to a season average around 94.  (Exact numbers for Soriano are complicated by the fact that the slightly slower cutter is his main pitch.)

It’s tough to see how that would indicate a problem with the PITCHf/x speed measurement for Wainwright. 

Sportvision does occasionally make very minor tweaks to the calibration of the system during the game if it gets beyond spec.  In practice I have never been able to notice the effect of these changes, so I think it’s reasonable to assume that whatever system was measuring one pitcher was the same system that was measuring the other pitchers.

Wainwright was *awesome* when he was a relief pitcher in the playoffs in 2006.  (Like David Price was too.)

It would be cool to see his speed numbers from the playoffs back then, and compare it to 2010 ASG (notwithstanding that nearly 4 years elapsed).

***

You put Joba in any other team, and he would develop like Wainwright and Price (and Santana, etc).  Put him in the media circus of NY, and, well, who knows what happens.

In looking at mlb’s Gameday from game 7 of the 2006 NLCS, it looks like he was getting it up to 95-96 then too. After the All-Star game, Wainwright did say he was pretty amped up. I don’t doubt the results based on the stats from the entire game, but Wainwright does appear to be the outlier.

Click on gameday and choose enhanced. The graphics have come a long way in four years.

http://mlb.mlb.com/news/wrap.jsp?ymd=20061019&content_id=1718119&vkey=wrapup2005&fext=.jsp&c_id=mlb

CJE: great stuff!

He had something like 28 or 29 outs in the playoffs, of which 15 or so were strikeouts.  And he was throwing so effortlessly.  I remember thinking that he must have been thinking “This is so easy to be a relief pitcher, why don’t they make me a starter already.”

Yep, good work by CJE/17.

People often forget that we have a smattering of PITCHf/x data from the 2006 playoffs.