Wrap Up, Part Two
Yesterday, we talked about the offensive contribution of the Mariners hitters and even worked in an adjustment for the position they played. However, we stopped there, as I wanted to give the defensive valuation its own post. So, picking up from where we left off with part one of the wrap-up, we have yet another post with me talking about defensive value.
The Mariners, as a team, allowed 813 runs this year. That’s 32 runs more than the league average despite playing half their games in the lowest run scoring environment in the American League. That’s not good, obviously, and it should come as no surprise to anyone who watched the team play that their weakness was in run prevention, not run scoring. But what parts of run prevention did the team really fail at?
Here are their performances, by outcome type, compared to league average:
Single: +65
Double: +23
Triple: +3
Home Run: -15
Walk: +10
Strikeout: -52
Hit Batter: +3
Well, we see one obvious strength of the pitching staff – they gave up 15 less home runs than league average. Of course, that’s almost all Safeco Field, but for the purposes of figuring out why the team gave up more runs than average despite playing half their games in a cavern, we can eliminate home runs from the equation. The team wasn’t particularly homer prone. And while they issued a few more walks than average, the difference isn’t huge – that’s not the culprit either.
The glaring weakness? Hit prevention. The team gave up 65 more singles, 23 more doubles, and 3 more triples than league average. That’s a lot of hits, and in turn, a lot of runs, going on the board for the other team.
Now, thanks to some inroads in statistical analysis, we’ve come to understand that hit prevention is not solely the domain of the pitcher, despite what mainstream analysis may still tell you. When a hitter makes contact and the ball doesn’t leave the yard, the outcome of that ball in play is, in part, determined by the quality of the defenders on the field. The Mariners, as a team, were terrible at turning balls in play into outs. On the season, they converted just 68.1% of their opportunities to make an out defensively, compared to a 69.6% league average. You may look at the percentage difference and say “ehh, big deal”, but each team deals with approximately 4,500 balls in play every year, and over the course of the season, that 1.5% difference adds up to about 68 plays not made.
Now, thanks to batted ball data, we can get even better accuracy. The Hardball Times, using data from Baseball Info Solutions that classifies every play as a groundball, flyball, or line drive and gives expected outs based on hit type, has the Mariners at 64 plays below average. Pretty close to our rough metric, but shows that the pitching staff did indeed give up a few more extra hard to catch balls than we’d expect. That’s not a big surprise to anyone who watched Jeff Weaver or Horacio Ramirez get torched on a regular basis this year.
Okay, so, using that 64 plays below average number, and thanks to Tango (as always), we can understand that those 64 plays the defense failed to make cost the Mariners about 50 runs. Fifty runs.
Wait a second – the team, as a whole, gave up 32 more runs than average, but we’re blaming the defense for being 50 runs worse than average? That means that we’re saying that the team had an above average pitching staff?
Well, almost. You have to remember to factor in Safeco Field. An average pitching staff that plays half their games in Seattle would give up less runs than average just thanks to the environment they play half their games in. But, yes, I am saying that the main factor in the team’s struggles at keeping runs off the board was not the pitchers, but instead, the fielders.
And, you know what, the numbers are on our side. The Mariners had a team Fielding Independant ERA of 4.48 compared to a league average of 4.51. Based on walk rate, strikeout rate, and home run rate, the Mariners had an average pitching staff, but their defense was worse than every other American League team besides Tampa Bay (whose attrocious defense also masked a pretty decent group of pitchers).
I know some of you will still want to blame the pitchers for the high ERA, because after all, it’s what everyone else in baseball does. But think of it this way? Do you really think the reality is that Felix Hernandez, Jarrod Washburn, Jeff Weaver, Horacio Ramirez, Cha Baek, Sean Green, and Ryan Feierabend all mysteriously lost about the same amount of ability to prevent hits on balls on play over the winter? All of them posted significantly worse rates of outs on balls in play than in ’07 than they did in ’06. Or, is it more likely that the common denominator behind them, their defense, wasted a lot of opportunities to create outs?
The evidence all points to the same fact – the Mariners defense was terrible this year. One of the worst in baseball. There’s no real way to argue differently – this is basically the same thing as stating that the Mariners hit for a high average or won more games than they lost.
The real question, for the purposes of player valuation, though, is which players were responsible for those 50 runs that the M’s forfeited with their defensive problems.
This is where subjectivity comes into play a bit. On a macro team level, we’ve just about got defensive evaluations nailed down. Splitting that pie up becomes a bit tougher, and the level of confidance goes down somewhat. Thanks to a lot of good work by a lot of people, we know that a great defensive player is about 20 runs better than the average for his position and a terrible defensive player is about 20 runs worse than the average for his position. Beyond those limits, teams realize that the defender is playing the wrong position and make an adjustment. It’s very rare to see a +20 or -20 player stick at one position, and everyone else falls somewhere in that range at pretty much each position.
However, as you’ve seen me talk about before, I’m still more comfortable using a +/- 5 run margin of error on both sides of the run values that the advanced defensive metrics come up with on an indvidual player basis – there are variations within teams that could affect the ratings, such as positioning of the fielders, handedness of the pitching staff, or a player “ballhogging” and always calling off all his teammates on plays that multiple players could make. This won’t affect the team’s totals, but it will affect the way we dole out credit/blame to the individual fielders, and so I’d rather use the ranges around the run values while people figure out how to solve those issues.
So, what do the numbers tell us about the Mariner fielders this year? Using a variety of inputs, we can safely come to the following conclusions:
Raul Ibanez – disastrous left fielder, -15 to -25 runs
Ichiro – above average center fielder, +0 to +10 runs
Jose Guillen – below average right fielder, -5 to -15 runs
Adrian Beltre – above average third baseman, 0 to +10 runs
Yuniesky Betancourt – below average shortstop, 0 to -10 runs
Jose Lopez – average second baseman, -5 to +5 runs
Richie Sexson – terrible first baseman, -10 to -20 runs
Kenji Johjima – who knows? – we don’t really have any idea how to evaluate catcher defense properly.
If we used the median value of those ranges, we’d come out with Ibanez at -20, Ichiro at +5, Guillen at -10, Beltre at +5, Betancourt at -5, Lopez at 0, and Sexson at -15, for a total of 40 runs below average. Toss in the part-time players (Broussard’s nothing to write home about with the glove either) and it ties pretty closely to the -50 that we see from the team wide total.
The controversial numbers, I’d expect, will be the “above average” ratings for Ichiro and Beltre, both of whom have stellar defensive reputations. Ichiro, I’m pretty comfortable with – he was consistently in the +10 to +20 range as a right fielder, and now that he’s being compared to a better crop of defenders in center, we’d expect his relative performance to decline. As good as Ichiro is defensively, he’s not head and shoulders ahead of guys like Curtis Granderson and Corey Patterson, who are also terrific defensive players in their own right. There’s nothing wrong with being a +5 center fielder – this still makes him one of the best defenders alive.
Beltre, I think, is just getting hurt by stiffer-than-normal competition at third base. Beltre’s really good, no doubt, but I’m not sure we appreciate just how good Brandon Inge and Mike Lowell are as well. You may not think of third base as a position with good defensive players, but there are some really, really good fielders manning the hot corner in the American League right now. With those kinds of peers in the group, it’s a little less surprising that Beltre’s defensive prowess doesn’t rate as well as we might expect from watching him play everyday.
This team, essentially, had two good fielders, a couple of average ones, a couple of below average ones, and two of the worst guys taking the field on a regular basis in baseball. Richie Sexson and Raul Ibanez just killed the Mariners with their gloves this year, and the rest of the team wasn’t good enough to make up for the organization putting two designated hitters behind their pitchers on a nightly basis.
Okay, that’s enough long-winded writing. If we apply the defensive run values to the position adjusted offensive run values from yesterday’s post, here are the results, and I’m comfortable using these as definitive metrics of value from the Mariners regular position players for 2007.
Stars
Ichiro Suzuki: +30 to +40 runs
Good Players
Adrian Beltre: +10 to +20 runs
Kenji Johjima: +10 runs offensively, ? defensively
Average Players
Jose Guillen: +5 to -5 runs
Below Average Players
Raul Ibanez: 0 to -10 runs
Yuniesky Betancourt: 0 to -10 runs
Jose Vidro: -3 runs offensively, no defensive adjustment
Bad Players
Jose Lopez: -15 to -25 runs
Disasters
Richie Sexson: -30 to -40 runs
Ichiro carried the position players, while Beltre and Johjima added positive value and Guillen provided league average production and consistency. Ibanez, Betancourt, and Vidro were small negatives, Lopez was a big problem, and Richie Sexson was one of the very worst players in baseball.
Add it all up and you have a group of position players that’s simply not playoff caliber. If the Mariners want to win consistently, they have to make some significant upgrades to this group – there are too many holes and not enough stars. It would take one of the best pitching staffs in the league to carry this group to a division title.
I think Bavasi’s going to sit tight on Kent — the market’s too hot for him this year. He’ll be a good pickup for, say, 2010 to play second every day and bat cleanup, after his price has come down a little. Then we can be sure of getting our bid of $120 mil for five years accepted for sure. His veteran leadership has only one direction to go at his age: to the moon, baby, to the moon.
It’s the canny insider’s thing to do.
No bites on this idea the first time, but I am very curious about it, so I’ll try again:
Dave says that we have a pretty good handle on how well a team played on defense as a whole.
Tango says that we need 4 years of individual stats before we can gain confidence in individual defensive metrics.
I’m wondering if it would make sense to do something in between, and try to evaluate the respective contributions of the infield and the outfield to a team’s overall defensive performance. One metric I think should be clear would be to compare across the league the percentage of groundballs each team converts into outs. That performance would be entirely the work of the infield.
Another metric–somewhat less clear, but maybe still useful–would be to compare the number of flyballs converted into outs. While the infield has some responsibility there, my uninformed guess is that flyball out percentage would primarily reflect the performance of the outfield. Certainly, flyballs resulting in doubles and triples would be balls that the infielders should have no chance at.
So, is this a viable project? Would it advance the ball?
202 – The Harball Times does that using their RZR/BIS data. Go here and take a look.
Doc/197: no. I’m saying you need two samples, each of 1600 balls in play (there’s 5 balls in play per game), to get a correlation of r=.80.
The correlation equation is what it is. The equation I posted is an approximation.
I’m also saying that a hitter need 800 PA in each of two samples to get a correlation of r=.80.
So, choose your personal threshhold relevance, and that’ll tell you how much data you need to believe whatever the metric is telling you. Obviously, you know Beltre/2004, so you’re not going to believe some 600 PA for a hitter. Your personal threshhold might require a hitter have a total of 1200 PA perhaps. If that’s you, then your correlation of r=1200/(1200+200) for hitters, or r=.833.
If that’s you, then for a fielder, you need .833=x/(x+400), which means you need 2400 balls in play. For a SS, that’s the equivalent of 480 full games (3 seasons). For a corner OF, that’s 800 full games (5 seasons).
On the other hand, maybe you are happy with an r=.50.
So, decide your tolerance level, and that’ll tell you how much data you need.
Thanks for the link, MarcW. I can’t quite tell from the link if THT’s RZR and OOZ numbers are aggregated from individual estimates, or if they are truly derived at a team level. It does appear that THT has done some interesting team-level comparisons of outs made on groundballs vs. flyballs/line drives, but I don’t see that those numbers are up for 2007.
What strikes me from the RZR/OOZ numbers is that the Seattle infield and outfield were both below average, but the infield was worse than the outfield. In fact, while the outfield’s RZR was below average, its OOZ was above average. Again, I don’t know if that’s just Ichiro stealing from the other outfielders, or if that’s the outfield as a whole showing greater range, but either way, these numbers seem to throw cold water on the UZR numbers suggesting that Ibanez, Guillen AND Ichiro! were each the worst at their position. Their collective performance wasn’t that bad, and the infield was somewhat worse.
Well, yes, Ichiro is a stud according to RZR/OOZ.
But the IF is rather troubling. Now we’ve got both PBP data sources indicating that perhaps Yuniesky’s range really isn’t that great.
THT’s +/- is a pretty cool system using park effects to weight DER. Another good thing to look at along with UZR or whatever; it’s nice to track back to the team-level stats, which was the point of your post, so forget I said anything….
If I were GM, I’d trade HoRam, Sexson, Bloomquist and Vidro for Chase Utley.
That’s a pitcher who has no business being on a major-league roster, a below-replacement-level first baseman, a 25th man, and a broken-down DH with an empty batting average, all for a legitimate MVP candidate.
Just so we’re clear.
Fans’ Scouting Report – Results
http://www.tangotiger.net/scouting/scoutResults2007.html
Thanks to all who participated.