Jim Hickey Analysis Part 1

by matthan on Jun 16, 2009 11:46 PM EDT Comment 67 comments

Lately there has been a bit of talk about the performance of our pitching staff and our bullpen. Naturally this leads to Jim Hickey. Is he a good pitching coach? Bad pitching coach? Are pitching coaches irrelevant? I'm going to attempt to shine some light to this dilemma. There just has to be some way to quantify the performance of pitching coaches....right?

I decided on two methods to analyze the performance of Hickey. Both methods involve looking at his time with both the Rays and Astros

1. Comparing the performance of a pitcher under Hickey to the same pitcher's performance under a different pitching coach. In order to do this I found a sample of pitchers (30 IP min) that had 1 year with Hickey and then the next MLB year w/o Hickey (or vice versa).

2. Comparing how a pitcher performs while under the eye of Hickey. Does pitchers improve or get worse over a year under the eye of Hickey?

If pitching coaches are irrelevant the metrics should remain static. Things such as declines due to age or improvements due to maturation should be filtered away due to sample size.

For Part 1 I am going to look at the first method comparing a pitchers performance with and without Hickey. My initial assumption (which is why I wanted to do this to begin with), is that performance suffers under Hickey.

Lets get right to the results

I've found a statistically significant decline in performance for the following metrics:

tRA, FIP, HR/9

I've found no statistically significant decline in performance under Hickey for the following metrics:

ERA, K/9, BB/9, and K/BB

The sample size is 37. In other words we have pairs of data (one portion of the pair is the performance under Hickey and the other is the performance under another pitching coach). Some of the pairs include Andy Pettittes 2006 with Hickey (Astros) and 2007 with Yankees. In fact nearly all of the 2006 Astros as they all had a new pitching coach the following year. Edwin Jackson from 2006 (no Hickey) to 2007 (w Hickey). Also Edwin Jackson from 2008 (w Hickey) to 2009 (no Hickey). James Shields 2006 (no Hickey) to 2007 (w Hickey).

*I did not break down the performance of the 2004 Astro pitchers compared to the 2005 Astros. It would have been too labor intensive for me to get the breakdowns for each metric for each pitcher under the previous PC in 2004 and under Hickey. Also it is questionable how much influence Hickey would have had in 2004 anyways. RJ made a post with some of these details Looking at Jim Hickey

I do wish there existed a bit larger of a sample, but this will have to do (the skewness and kurtosis have some issues). However I did want to post what I have now

Now onto the numbers

FIP Analysis
FIP with Hickey		FIP w/o Hickey

Mean	4.66	Mean	4.24
Standard Error	0.15	Standard Error	0.12
Median	4.79	Median	4.18
Standard Deviation	0.93	Standard Deviation	0.75
Kurtosis	-0.73	Kurtosis	-0.09
Skewness	0.04	Skewness	0.36
Minimum	3.02	Minimum	2.83
Maximum	6.73	Maximum	6.04
Confidence Level(95.0%)	0.31	Confidence Level(95.0%)	0.25
P Value	0.0131

tRA Analysis
tRA with Hickey		tRA w/o Hickey

Mean	4.95	Mean	4.45
Standard Error	0.21	Standard Error	0.18
Median	4.76	Median	4.49
Standard Deviation	1.25	Standard Deviation	1.03
Kurtosis	-1.03	Kurtosis	-0.37
Skewness	0.10	Skewness	-0.27
Minimum	2.79	Minimum	2.28
Maximum	7.52	Maximum	6.29
Confidence Level(95.0%)	0.44	Confidence Level(95.0%)	0.36
P Value	0.0236

ERA Analysis
ERA with Hickey		ERA w/o Hickey

Mean	4.89	Mean	4.49
Standard Error	0.27	Standard Error	0.23
Median	4.69	Median	4.43
Standard Deviation	1.67	Standard Deviation	1.40
Kurtosis	-0.73	Kurtosis	0.71
Skewness	0.29	Skewness	0.41
Minimum	2.04	Minimum	1.80
Maximum	8.13	Maximum	7.98
Confidence Level(95.0%)	0.56	Confidence Level(95.0%)	0.47
P Value	0.1019

K/9 Analysis
K/9 with Hickey		K/9 w/o Hickey

Mean	7.23	Mean	7.33
Standard Error	0.29	Standard Error	0.23
Median	6.78	Median	7.02
Standard Deviation	1.78	Standard Deviation	1.38
Kurtosis	0.83	Kurtosis	1.94
Skewness	0.95	Skewness	1.21
Minimum	4.54	Minimum	5.04
Maximum	12.48	Maximum	11.82
Confidence Level(95.0%)	0.59	Confidence Level(95.0%)	0.46
P Value	0.3238

BB/9 Analysis
BB/9 with Hickey		BB/9 w/o Hickey

Mean	3.41	Mean	3.33
Standard Error	0.17	Standard Error	0.19
Median	3.20	Median	2.88
Standard Deviation	1.02	Standard Deviation	1.18
Kurtosis	-0.59	Kurtosis	0.30
Skewness	0.24	Skewness	0.94
Minimum	1.51	Minimum	1.71
Maximum	5.63	Maximum	6.19
Confidence Level(95.0%)	0.34	Confidence Level(95.0%)	0.39
P Value	0.3644

K/BB Analysis
K/BB w/ Hickey		K/BB w/o Hickey

Mean	2.33	Mean	2.43
Standard Error	0.16	Standard Error	0.13
Median	2.00	Median	2.57
Standard Deviation	1.00	Standard Deviation	0.80
Sample Variance	0.99	Sample Variance	0.64
Kurtosis	0.82	Kurtosis	-1.07
Skewness	1.21	Skewness	-0.03
Minimum	1.22	Minimum	1.08
Maximum	5.11	Maximum	4.00
Confidence Level(95.0%)	0.33	Confidence Level(95.0%)	0.27
P Value	0.3226

HR/9
HR/9 w/ Hickey		HR/9 w/o Hickey

Mean	1.26	Mean	1.01
Standard Error	0.07	Standard Error	0.06
Median	1.24	Median	0.96
Standard Deviation	0.45	Standard Deviation	0.36
Kurtosis	-0.29	Kurtosis	-0.45
Skewness	-0.11	Skewness	0.37
Minimum	0.23	Minimum	0.41
Maximum	2.22	Maximum	1.78
Confidence Level(95.0%)	0.15	Confidence Level(95.0%)	0.12
P Value	0.0054

The data here FURTHER emphasizes how much of a crappy metric ERA is.

There are some outliers within this sample. I wish it was larger to smooth out some of the possible variables. Trever Miller has two entries that really help Hickey. Camp has two that really hurt Hickey. Percival really hurts Hickey. Jackson really hurts Hickey (had lower FIP w/o Hickey in 2006 also). Clemens did great with Hickey in 2006 and flamed in NY in 2007 etc.

Here is the file Google Doc

This is my first fanpost so pardon my crappy tables. I'm too busy watching the Rays dominate to figure that out

I'll take a look at whether players improve or decline under at least 2 years of Hickey next.

5 recs | Comment 67 comments

Story-email Email Printer Print

Read More: Tampa Bay Rays, Houston Astros

Comments

Ugh, my knowledge of college statistics is worse then I thought.

I don’t remember what more then half this shit means.

"Where we all wait in earnest with pudding in hand for the Upton comet to sail through the roofed skies, so that we may meet Him."

by kericr on Jun 16, 2009 11:57 PM EDT reply actions 0 recs

Well the sample isn't the best but its not bad

Main point is the P-Values. Anything under .05 is significant.

Basically the gist is we are really sure that FIP, tRA, and HR/9 were worse under Hickey than they were under a different pitching coach when those two seasons occurred back to back given the 30 IP minimum.

by matthan on Jun 17, 2009 12:01 AM EDT up reply actions 0 recs

Ok, at least I got that much.

This is good work BTW.

"Where we all wait in earnest with pudding in hand for the Upton comet to sail through the roofed skies, so that we may meet Him."

by kericr on Jun 17, 2009 12:11 AM EDT up reply actions 0 recs

This is good stuff.

I want to look it over when it’s not midnight, but I’m glad you went through with it.

by R.J. Anderson on Jun 17, 2009 12:01 AM EDT reply actions 0 recs

I was going to post it tomorrow

But the Rays were winning by such a large amount that posting seemed a better alternative than dishes or job searching. Anyways it is midnight. Time for bed

by matthan on Jun 17, 2009 12:03 AM EDT up reply actions 0 recs

Rec'd. This is fantastic

because it seems to prove what some of us have been arguing the last few weeks: Hickey sucks. I’m also really pleased that your provided a p-value for all of your analyses (something almost no one seems to do in sabermetrics and it throws me off because i’m in the sciences and without a p-value you’ve got nothing). Did you just used a paired t-test to determine significance? Also, you said you went in with the assumption that Hickey was a bad PC, but I hope you still used a 2-tailed test as opposed to a 1-tailed test.

Just today, Sandy Kazmir wrote:

some morons think the Pitching Coach should be fired

Well maybe he shouldn’t be fired mid-season, but certainly there is no reason to believe he shouldn’t be fired after this season. Of course it will never happen if the Rays are successful.

by RaysTheRoof on Jun 17, 2009 1:37 AM EDT reply actions 0 recs

We shouldn't be talking about firing at this point, especially mid-season

Besides this is an incomplete picture. Perhaps Hickey gets more long-term results than short term? Perhaps the Rays are trying to achieve a specific goal out of their pitching coach such as certain results with young pitchers?

Anyways I did used a paired test, but I used 1 tail. I just threw in 2 tails and obviously I’d get the same metrics as being significant . Sorry about that.

tRA .047
FIP .026
HR/9 .0107

All still significant but the p-values are a tad bit higher.

by matthan on Jun 17, 2009 8:12 AM EDT up reply actions 0 recs

Perhaps he simply had better pitchers?

I could be wrong though

by staplemaniac on Jun 17, 2009 10:51 AM EDT up reply actions 0 recs

I'm comparing the same pitchers

A pitcher with one season with Hickey and the same pitcher with another season with another guy.

As FreeZorilla pointed out the difference could be ballpark based.

by matthan on Jun 17, 2009 10:55 AM EDT up reply actions 0 recs

stupid

need to read before opening my vitrual mouth

I could be wrong though

by staplemaniac on Jun 17, 2009 11:25 AM EDT up reply actions 0 recs

Yeah, I figured they were still going to be significant.

Agreed no firing of Hickey during the season.

by RaysTheRoof on Jun 17, 2009 12:30 PM EDT up reply actions 0 recs

Yes I was referring to you.

You must have known that already since I used the word moron.

Rays Win!

by Sandy Kazmir on Jun 17, 2009 9:38 AM EDT up reply actions 0 recs

Never said you were.

Maybe just choose your words a little more carefully in the future. Second time you’ve called someone stupid or a moron in the past week when there was not definitive evidence to support your argument. Condescension doesn’t help anyone.

by RaysTheRoof on Jun 17, 2009 12:23 PM EDT up reply actions 0 recs

FTFY

~~Condescension~~Mob mentality doesn’t help anyone.

Rays Win!

by Sandy Kazmir on Jun 17, 2009 12:33 PM EDT up reply actions 0 recs

Iran

by RaysTheRoof on Jun 17, 2009 12:35 PM EDT up reply actions 0 recs

It would be funny if they used their new found nuclear capability to nuke their own citizens

Would Israel join in? That’s enough politics for me, I’m just highly encouraged by the brave people over there refusing to submit to tradition.

Rays Win!

by Sandy Kazmir on Jun 17, 2009 12:40 PM EDT up reply actions 0 recs

Me too

by RaysTheRoof on Jun 17, 2009 12:46 PM EDT up reply actions 0 recs

Also, can't wait for Part 2.

If it shows the same trend that would be pretty damning.

by RaysTheRoof on Jun 17, 2009 1:43 AM EDT reply actions 0 recs

My guess that it won't

I don’t think any metric will be significant. I’m thinking about just looking at a players 1st year with Hickey compared to his 2nd year with Hickey. This way I’ll be able to include a bunch of Astros as well as Rays. However this will take away the 2008 to 2009 portion (which has been a decline). As we do know our ptiching staff did get better in 2008 as compared to 2007 so I doubt we will see the same result. However perhaps that increase was more due to the presence of new pitchers (or defense) rather than improvement of the old? If I recall correctly the 2005 and 2006 Astros had pretty similar performing pitching staffs. Of course I won’t know until I dig out the numbers.

But then again it would be pretty damning if the numbers don’t show positive improvement. It wouldn’t be a good sign if pitchers show statistical improvement under other pitching coaches and also have no evidence of any change under his eye.

We shall see.

by matthan on Jun 17, 2009 8:25 AM EDT up reply actions 0 recs

I'm curious as to how many pitchers switched league (facing P vs DH) and also stadium factors due to the HR increase which directly correlates into FIP

Follow Me on Twitter @FreeZorilla

by FreeZorilla on Jun 17, 2009 8:59 AM EDT reply actions 0 recs

Other factors:

Al East
Age
Year
A context neutral measure would be nice.

by rglass44 on Jun 17, 2009 9:03 AM EDT up reply actions 0 recs

It would be nice

Although age shouldn’t be much of a factor. It is a pretty good cross sectional sample, plus each pair is only a year apart. With a good mix of year with Hickey at year 1 and without Hickey as year 2 and vice versa. So we should have a good mix of pitching naturally getting better and worse in the mix.

The ALE is kind of funny. Camp is actually one pitcher that really hurt Hickey in this analysis and he remained in the division.

I’ll break down the pitchers when I can. But for the time a lot of the same are from pitchers that remained on the same team. 2006 Astros with Hickey to 2007 Astros without Hickey. 2006 Rays without Hickey to 2007 Rays with Hickey. So that kind of minimizes the league and park variables.

by matthan on Jun 17, 2009 9:10 AM EDT up reply actions 0 recs

sample not same

by matthan on Jun 17, 2009 9:11 AM EDT up reply actions 0 recs

BTW

In the file I do have the team with the year the pitcher was with Hickey and the team with the year the pitcher was not with Hickey coupled with the data for those seasons.

by matthan on Jun 17, 2009 9:29 AM EDT up reply actions 0 recs

Nice starting point...

A few questions I had…

1) How much does luck factor in to the year by year measure? Would a two to three year measure be better? Percy is a great example. He was great for the Cardinals in 2007 but terrible for the Tigers in 2005 (his previous year).
2) Did the 2007 Rays defense significantly hurt some of the pitchers numbers that year?
3) Age and peak seasons. Wandy improved significantly after Hickey, but how much of that had to do with maturity? Same with Percy, Reyes.
4) And context. Edwin was a reliever prior and a starter with. Other guys switched leagues.

by tallyray on Jun 17, 2009 9:34 AM EDT reply actions 0 recs

1.I’m sure there is some luck component. However "luck" is essentially variation around the mean so we should assume that there is equal probability of being lucky and unlucky. Also take a peak at the data. It is spread across multiple years, multiple pitchers, multiple seasons, multiple teams, and multiple ages and points of their career. It would be awfully coincidental that pitchers just so happened to get lucky without Hickey and then unlucky with Hickey to the extreme of getting statistically significant results.
2. Again as with my explanation with point 1 there is a pretty broad sample so the defense of 1 team shouldn’t bias the sample too much. However FIP and tRA should be pretty independent of defense. However this reason could be why ERA is not statistically significant since that would fluctuate with defense.
3. This of course is an issue. The larger the sample size the better. However in this case Hickey did have a good mix of young and old pitchers. I don’t find this to be that big of an issue with this sample
4. This is the biggest potential problem. The only saving grace is there should be close to an equal amount of guys that went from starting to relieving and relieving to starting.

by matthan on Jun 17, 2009 11:22 AM EDT up reply actions 0 recs

n=37 is not a large enough sample

This looks like a lot of hard work went into it, and even better that you got a couple of sigs out of it, but that sample needs to be at least twice that. As you stated there isn’t much you can do, so maybe you could do this exercise for Mike Butcher or one of the other PC’s around the league.

Also, it’s interesting that the mean HR/9 is 1.26. In 2009 the MLB average was 1.0, same in 2008, and the same in 2007. Perhaps he gets his pitchers to pitch to contact, not worrying about the homer. If that is his philosophy, perhaps we should take a look at the relevancy of that. As is, his tRA, FIP, and HR/9 #‘s are all skewed by that incredibly high HR-rate, which I guess you did mention with the St. Dev. that was spit out on the HR/9. As you mentioned though, larger samples would most likely smooth that out, I’m just not sure if a larger sample would also move the mean.

Rays Win!

by Sandy Kazmir on Jun 17, 2009 10:05 AM EDT reply actions 0 recs

I don't think it is really possible to compare league averages to these means

It isn’t really comparing apples to apples. In this data a HR/9 of 1 throwing 200 IP is weighted the same as a HR/9 of 2 throwing 50 innings. The mean would be 1.5 obviously. When looking at league averages I believe it is calculated much differently. As in: (total # of HR/total # of innings)*9 rather than the average of every individuals personal metric. I’d actually love to see a histogram of every MLB players personal HR/9 ratio. I wonder if it is normal. My initial guess would be that better pitchers pitch more innings at a lower HR/9 which causes the league average to be lower than the actual mean of the individual personal ratio.

by matthan on Jun 17, 2009 10:18 AM EDT up reply actions 0 recs

So you just took the average of each guys HR/9?

Why wouldn’t you take the team totals? I guess you’re doing this on an individual basis, but you’re inviting so much variation by doing it that way.

Rays Win!

by Sandy Kazmir on Jun 17, 2009 10:27 AM EDT up reply actions 0 recs

The problem with team totals is each team is made up of different players. Comparing the 2006 entire Astro staff with the 2007 entire Astro staff is worthless due to the composition. For example one had a rotation of Roger Clemens and Andy Pettitte compared to one with Wandy, Mitch Albers, Jason Jennings etc. That is why I broke it down by individual player performance.

by matthan on Jun 17, 2009 10:40 AM EDT up reply actions 0 recs

I realize that, it just seems like that could be weighted somehow.

I don’t think a guy that pitched 44 innings one year at a 2.2 HR/9 and 112 IP at 1.4 the next compares the same as a guy that that racked up a bunch of innings with realistic HR totals.

Rays Win!

by Sandy Kazmir on Jun 17, 2009 11:00 AM EDT up reply actions 0 recs

There are a lot of different angles to take

I’m more concerned about whether he helps or hurts each individual pitcher. If he helps Sonnanstine and hurts Balfour I want it to have the same meaning. My assumption is that it is totally random the role that he helps and hurts. So if we weighted by innings in this Sonny/Balf example it would show that he had a net help. However in the future if who he helps is random then there is no way of knowing if it will the starter or the reliever the next go around? If you weight them equally at least will be able to say more pitchers improved/declined under Hickey than without. Although I could refine the data even more to answer that question far better.

by matthan on Jun 17, 2009 11:26 AM EDT up reply actions 0 recs

Thanks I appreciate you breaking down your thought process

I try to think of everything put up here as a from of peer review.

Rays Win!

by Sandy Kazmir on Jun 17, 2009 11:35 AM EDT up reply actions 0 recs

form

e gads

Rays Win!

by Sandy Kazmir on Jun 17, 2009 11:35 AM EDT up reply actions 0 recs

No problem

More thoughts the better. I was just sick of people crying about Hickey without having anything to back it up.

by matthan on Jun 17, 2009 12:02 PM EDT up reply actions 0 recs

Yeah and if you've truly stumbled onto him being homer prone, which I'm on the fence about,

surely it is for a reason and is compensated for in some other regard. Batted ball types would be a nice indicator.

Rays Win!

by Sandy Kazmir on Jun 17, 2009 12:35 PM EDT up reply actions 0 recs

Interesting tidbit

Everything with 20IP minimum in 2008

I wanted to see what the difference was between the mean of the sample of “personal HR/9” compared to league HR/9

To get league I first sorted by 20 IP min. Then (Total HR/Total IP)*9= 1.0015

To get the mean of the personal metrics I just sorted by the 20 IP min and took the average of the HR/9 ratio………I got 1.0302

So the mean of the personal metrics is in fact higher than the league average.

Of course we’d have to do look at it over a course of many years to see if it is a significant difference…but thats the data for 2008 pretty much what I hypothesized.

by matthan on Jun 17, 2009 10:48 AM EDT up reply actions 0 recs

And I do agree that the sample is a bit too small and not diverse enough

It was never meant to be a fire Hickey type thing anyways. Just a small piece of the decision making puzzle.

by matthan on Jun 17, 2009 10:19 AM EDT up reply actions 0 recs

No I didn't mean to imply that you were biased one way or the other. I think this is really good and as Torts said

good is the enemy of great. There’s enough smart people here that we can make this even better. I’m a brainstormer, I get a lot of ideas and then I chase them down, even though 90% are usually pretty bad. I’ve never seen anything like this, and it gets my question juices flowing.

Rays Win!

by Sandy Kazmir on Jun 17, 2009 10:29 AM EDT up reply actions 0 recs

If you look only at the pitchers who changed pitching coaches but not teams it reduces your sample size to 13

If you take the total of home runs of that group pre or postHickey/sum of innings pitched * 9 you get the HR/9 of the group.

I did this for the pre+ post Hickey seasons and the Hickey seasonsand got strikingly different results.

HR/9 of the Non-Hickey=1.01
HR/9 of the Hickey= .96

Follow Me on Twitter @FreeZorilla

by FreeZorilla on Jun 17, 2009 10:45 AM EDT reply actions 0 recs

That's what I mean, because of a lack of data, you have to open the floodgates for noise to come in and have a picnic of what

otherwise would be a pretty interesting read.

Rays Win!

by Sandy Kazmir on Jun 17, 2009 11:03 AM EDT up reply actions 0 recs

I agree... It's a starting point

But very few conclusions can be gained from the information right now. Too many other variables are present.

Now it’s about finding where to go with the information and how to make it useful.

by tallyray on Jun 17, 2009 11:09 AM EDT up reply actions 0 recs

In hindsight I i wish included FB, GB, LD%

It would be nice to know the change of those ratios when they switch from Hickey to somewhere else and vice versa

by matthan on Jun 17, 2009 11:11 AM EDT up reply actions 0 recs

If you decide to do it, approach it from the sum of the batted ball types as opposed to each pitchers %.

Follow Me on Twitter @FreeZorilla

by FreeZorilla on Jun 17, 2009 11:14 AM EDT up reply actions 0 recs

I'll do it both ways

I’m actually more concerned about Hickeys impact on individual pitchers. So there is no difference between 200 IP and 75 IP as long as the metric is reliable. I don’t want a pitcher that threw a lot of innings to bias the data one way or the other. I don’t want to help or hurt Hickey by overweighting the guys that threw a lot of innings.

The cumulative number would be nice but I just think that would be worthless. Unless there was some relationship between Hickey having better success with starters than relievers. If his success is random then running the risk of having a pitcher a change with large innings really can bias the sample.

Of course it depends on what you are looking for.

by matthan on Jun 17, 2009 11:20 AM EDT up reply actions 0 recs

I think its more accurate to assess the overall change than individual pitchers

While 30 IP gives it a low measure of reliability, the sum of the group’s innings pitched gives it much greater accuracy. With such a small sample size you can get a degree of accuracy with the sum. When looking at individuals there are just too many other variables in play.

Follow Me on Twitter @FreeZorilla

by FreeZorilla on Jun 17, 2009 11:30 AM EDT up reply actions 0 recs

It really depends on what you want to look at

I’m most curious to see how Hickey does with each pitcher. I feel biasing towards starters isn’t the best way to go. If anything that reduces your sample size as the variation is caused by even fewer individuals.

by matthan on Jun 17, 2009 11:48 AM EDT up reply actions 0 recs

Either way I'm going to dig a lot deeper into indivdual metrics for part 2

It won’t be nearly as tough to find the data since I won’t have to individually search for players.

by matthan on Jun 17, 2009 11:50 AM EDT up reply actions 0 recs

Basically I plan on adding batted ball type and pitch type

Although both of those would have been more useful for Part 1. It would have been nice to see if Hickey favors a certain type of pitch compared to other pitching coaches. Maybe I’ll make that part 3.

by matthan on Jun 17, 2009 11:53 AM EDT up reply actions 0 recs

Now this I want to see, fastball percentages

Follow Me on Twitter @FreeZorilla

by FreeZorilla on Jun 17, 2009 12:04 PM EDT up reply actions 0 recs

Owe an Apology

this info above was incorrect, I missed the 2 in Shield’s 28 07 HRs. i discovered this when trying to break out the people w/ 100 innings in both years. I’ll post that info below

Here are the accurate #s

W Hickey: 1.08 HR/9
W/out Hickey: 1.02 HR/9

Follow Me on Twitter @FreeZorilla

by FreeZorilla on Jun 17, 2009 1:06 PM EDT up reply actions 0 recs

The fact that this doesn't have 5 recs yet is stupid.

"Where we all wait in earnest with pudding in hand for the Upton comet to sail through the roofed skies, so that we may meet Him."

by kericr on Jun 17, 2009 12:42 PM EDT reply actions 0 recs

fix'd

I could be wrong though

by staplemaniac on Jun 17, 2009 12:49 PM EDT up reply actions 0 recs

Okay I nearly finished with part 2

My methodology is quite a bit different. I decided to measure the change between a year with hickey and the next year with hickey for many metrics (part 1 was about hickey v nonhickey pc). I assume the expected change to be zero. I’m attempting to find if the change is significantly different than zero.

The findings are pretty neat.

A sneak peak:

HR/9 insignificant. In fact the mean change comes out to be 0.00
One of K’s or BB’s have been significantly getting worse under a year of Hickey…
Under Hickey the usage of two pitch types have significantly changed…

Of course there is sample size issues again. We only have 26 pitchers and not every one throws every pitch so the pitch type samples are even smaller for some pitches

by matthan on Jun 17, 2009 1:09 PM EDT reply actions 0 recs

Honestly I am really fascinated by the results

I was playing around with one particular player out of the sample (he had two entries) to see what type of impact it had on these metrics and the signifigances. Nothing that was significant became insignificant, but there was definitely an impact. That further emphasizes we may be onto something if I can take the outlier out and we still see significance.

by matthan on Jun 17, 2009 1:16 PM EDT up reply actions 0 recs

You've piqued my interested with the pitch type talk.

by R.J. Anderson on Jun 17, 2009 1:16 PM EDT up reply actions 0 recs

yes pitch type data would be nice

by RaysTheRoof on Jun 17, 2009 1:19 PM EDT up reply actions 0 recs

I'm analyzing it as we speak

Only for periods under Hickey though. I’m going to have to backtrack to look if there is a significant pitch type change from Hickey to another pitching coach.

However I do see some signifigance in the “only Hickey” data. It definitely looks like he is pulling pitchers away from a certain pitch and towards another pitch….

If the data shows this for a period of time with just Hickey I bet it is even more pronounced when Hickey get a pitcher from a different pitching coach. As in if the change is significant from Year 1 Hickey to Year 2 Hickey then I’d be shocked if it wasn’t for Year 1 NonHick to Year 2 Hick.

by matthan on Jun 17, 2009 1:36 PM EDT up reply actions 0 recs

STOP GIVING US BLUEBALLS.

by R.J. Anderson on Jun 17, 2009 1:42 PM EDT up reply actions 0 recs

this

Rays Win!

by Sandy Kazmir on Jun 17, 2009 1:42 PM EDT up reply actions 0 recs

Starters vs Relievers/Both

I considered a starter as anyone with min 100 IP for both seasons. Below you will find the deltas for each pitcher. A negative infers the pticher gave up less HR/9 under Hickey.

Roy Oswalt 0.14
Wandy Rodriguez 0.04
James Shields -0.13
Scott Kazmir -0.15

For the group summation the HR/9 was .94 with and without Hickey. Releivers/both to come later. Have to get to a meeting,

Follow Me on Twitter @FreeZorilla

by FreeZorilla on Jun 17, 2009 1:10 PM EDT reply actions 0 recs

Nice work

I’m in the middle of looking at how pitchers have fared under two consecutive years of Hickey. As in a pitcher for the Rays in 2007 and the same guy in 2008. I also found the mean change in HR/9 is exactly 0.00

by matthan on Jun 17, 2009 1:12 PM EDT up reply actions 0 recs

Although the standard deviations of those changes is pretty high

.43 in fact

by matthan on Jun 17, 2009 1:13 PM EDT up reply actions 0 recs

Relievers

Player Delta
Brad Lidge -0.01
Chad Qualls -0.07
Dave Borkowski 0.01
Trever Miller 0.08
Casey Fossum 0.53
Edwin Jackson 0.57
J.P. Howell 0.56
Jason Hammel -0.16
Shawn Camp 0.50

The summation of these pitchers:
Hr/9 w Hickey: 1.24
HR/9 Wout: 1.10

Obviously with the smaller innings pitched Hr/9 turns out to be a less reliable measure when looking solely at individuals.

Follow Me on Twitter @FreeZorilla

by FreeZorilla on Jun 17, 2009 1:17 PM EDT reply actions 0 recs

About how many IP do you think you need for these ratios to become reliable?

by matthan on Jun 17, 2009 1:37 PM EDT up reply actions 0 recs

Its tough b/c one meaty pitch, or one meaty pitch missed by the batter totally changes the rate

When the average is one event per 9 innings it seems like you would need more innings for HR than for BB or K. BBs and Ks are also the result of several pitches where a HR is just one pitch and swing. Maybe its in the Book, I’ll check tonight.

Follow Me on Twitter @FreeZorilla

by FreeZorilla on Jun 17, 2009 3:04 PM EDT up reply actions 0 recs

So after this was so good

I would love to see something similar to this but focusing on stuff and PitchFx. I know that this is a shit ton of work as it is, and going through a ton of Fx data is even harder, but I would love to see results with Hickey according to a pitchers command/stuff/break. Maybe Hickey is a little below average in certain places, but maybe he is teaching guys new grips/pitches or he is really good at making a pitchers stuff just plain nastier. I know we would probably see an indication of this in things like K/9 but Id love to see the actual data here. Great stuff though man.

by BJ the Bossman on Jun 17, 2009 1:56 PM EDT reply actions 0 recs