Projecting Kevin Youkilis’ Average
In 2008, Kevin Youkilis had a season completely out of line with what he did previously. Everyone focused on the home run power, as his 29 homers matched his combined total from 2007 and 2008. But Youkilis also enjoyed a career-best average last season, as he posted a .312 mark. His previous high was the .288 he posted in 2007.
Because of his big 2008, Youkilis was one of the more difficult players to project. The Bill James forecast was the most optimistic, and it had him with a .289 average and 23 home runs. The CHONE system had him with a .286 average and 18 home runs.
Of course, Youkilis is off to a great start here in 2009. One of the biggest grey areas in fantasy is determining when we can project that it is more than a hot start and more likely to be something to last the whole season.
I think we may have reached that point with Youkilis and his average.
I took all of the players in 2008 who qualified for the batting title and who batted plus or minus five points from the .286 average that CHONE predicted. I came up with a group of 26 players. Then I took all of the players from this group and counted how many hits they had through 75 at-bats last season.
Our 26 players ranged from 16 to 27 hits in their first 75 at-bats last season. It is not a perfect bell curve but the right side of it does display a standard downward-sloping tail as you can see by the crude chart below:
x x x x x x x x x x x x x x x x x x x x x x x x x x 16 17 18 19 20 21 22 23 24 25 26 27
I came up with an average of 20.7 hits and a standard deviation of 3.1 for this group. So, two standard deviations gives us a high and low of 15 and 27 hits. So, for this definition of a .286 hitter, 95 percent of the time he should have between 15 and 27 hits in 75 ABs.
Youkilis ended play last night with 74 at-bats and 30 hits, meaning he falls outside of our 95 percent range. Our most likely conclusion is that he is a better hitter than the CHONE projection prior to the start of the season.
This is just a first step and I have no projection on what Youkilis will end up hitting in 2009. ZiPS projected Youkilis to bat .288 this season. Dan Szymborski just came out with an update to ZiPS based on 2009 stats. He now projects Youkilis to bat .298 for the rest of the season and finish with a .313 average.
My chart got cut off but one player had 27 hits.
After I posted this I see where David Cameron had a similar-type article on the site with Victor Martinez.
Good analysis. I simulated 74 at bats at .286 and found under 2% with 30 or more hits. He’s clearly better than .286 right now.
What’s sort of interesting to me is that his increase in average last year can be entirely explained by the increase in HR — turn 13 of those 29 HR into fly outs (as in 2007) and you wind up with an average of .288 (as in 2007). I don’t know if that’s applicable this year — he’s hit a couple of monsters — but if 2 of his 5 HR were close calls, he’d drop back into that 95% interval.
I forget — does the re-projection take into account park effect, with 12 of the 20 games so far in Fenway?
Altering projections after roughly 1/8 of a season is never a good idea. Just ask those who tweaked Johnny Gomes projected average back in 2006. Everybody goes through a cold spell at some point, and Youk is bound to take a dive in the near future.
Youk’s worst BA in a month last year (excluding the 2 March games) came in September when he batted .275. He was remarkably consistent last year, but the law of averages say it’s impossible to keep this pace up. He didn’t have a cringe worthy strikeout rate last year, but it was a tad higher than league average. More complete hitters than Youk tail off over a month from time to time. Pujols hit .250 in April 2007.
I always treat batting averages like gravity. What goes up, must go down. Count on Youk’s batting average going down significantly during a month this season. I’m not budging from by original .290 projection.
I’m not sure how useful an exercise is, but if you adjust his BABIP from .455 down to his career .339, he’d be at a .320 BA. His HR/FB is almost 2x his career, and about 30% above last year, so if you take away 1 of his 5 HR and turn it into an out, he’d still be at .306.
Why did you only look at the first 75 at-bats for each of the 26 players? You should look at ALL of their 75 at-bat samples, since you’re trying to determine how likely it is that a .286 player tallies 30 hits in 75 at-bats period, not 30 hits in their first 75 at-bats.
In an ideal world that might have been what I did. If I lived in a world with no time constraints or had software to do all the calculations for me then this piece would have been different. Feel free to do all 75 AB samples (figuring out by hand was hard enough for one sample for each, much less 500 or more for each) for 26 players, calculate and graph the results and write up the article. I’ll enjoy reading that.
Jeez, touchy much?
I’m curious – when you post on other free sites suggesting that someone spend thousands of hours on a project – what kind of response do you normally get?
“In an ideal world that might have been what I did. If I lived in a world with no time constraints or had software to do all the calculations for me then this piece would have been different.”
**********
Something like this.
Which would have sufficed.
And that would have been my response had the individual simply asked:
“Why did you only look at the first 75 at-bats for each of the 26 players?”
He raised the ante and then cried when I responded in kind.
Brian…you’re writing for a website. People will read and criticize…get used to it.
Just be glad they are reading!
The point Eric brings up is valid…as is your defense about time and data constraints…no need to get testy.
Hi Brian – thanks for the comment.
As has anyone who has posted things for public consumption, I’ve been corrected, proven wrong, misquoted and ripped a new one for things that have turned out to be proven true. It goes with the territory and I accept that going in.
But this was the first time that someone acted condescending towards me because I did not spend thousands, and I mean literally thousands of hours with the way I prepared this article, in order to do it the way he thought it should be done.
And so I invited him to do that research instead, said I’d like to read the article (which I would, to see how my sample of 26 compares to the 13,000+ one) , and I get accused of being “touchy” and “testy”
All things considered, I think I have been fairly restrained.
I’m sure most of you understand this, but:
If you projected KY at say .286 before the season, then unless you think that for the rest of the year he’s going to be worse than you projected, you must now increase his projection for the entire season. You might think he’ll still hit .286 for the rest of the season, but because he’s already done better than that so far, his *whole season* projection necessarily goes up.
(Meaningless guess: he finishes at .307.)
Assuming 550 at bats per player, you would be looking at something like 190 75 at-bat samples just for 2008. That would be interesting to do, but would be pretty time consuming. Eric has a valid point though….can a conclusion be made from that specific of a sample? 2008 could be drastically different from other years, and the other 75 at bat samples from 2008 could also tell a much different story….very interesting idea though. I’d like to see how this compares to something done including all the samples and multiple years….
And 190 is only if you cut up a season into 75 at bat chunks….to cover all 75 at bat intervals would be some ungodly number….
I almost forgot that Youk has done this before. He had a .402 BA in May 2007. I had him marked down for .285 that year. I never wavered from my original number. After batting .354 the first two months, he finished with a .288 average.
Dont be suckered in by the magical .400 number. Youk will slump soon, and slump hard. Bank on it.