Saturday, March 10, 2007

2006 Reds Pitching Review

Continuing where we left off with the '06 Hitting Review, tonight we'll investigate the Reds pitching in 2006.

'06 Pitching Recap


The 2005 Reds had awful pitching. It's a point that just can't be emphasized enough. They led the league in runs allowed, ERA, hits allowed, and home runs allowed, while they were second to last in strikeouts, last in complete games, last in shutouts, and, for what it's worth, last in saves. So to say that the Reds' pitching needed to improve in 2006 is a tremendous understatement.

Fortunately, it did improve. After allowing an unbelievable 889 runs in 2005, the Reds' staff allowed "only" 801 last season, which was 10th best in the 16-team league. Not quite above-average, but 88 runs is huge improvement, representing somewhere around 9 wins difference (hitting lost 71 runs off the 2005 totals, of course, which cut the overall team improvement to somewhere around just 2 wins).
Reds ERA over the past five seasons, broken down by starters and relievers (ref: ESPN). Dotted lines indicate league averages for starters and relievers.

The largest improvement on the team was in the starting rotation, which saw its ERA drop 0.80 runs. In fact, thanks to strong performances by Bronson Arroyo and Aaron Harang, as well as a decent late-season showing by Kyle Lohse, the Reds starters were (slightly) better than league average for the first time in five years. Nevertheless, the bullpen also showed notable improvement, dropping 0.37 earned runs per nine for the second consecutive year behind the surprising Todd Coffey, as well as David Weathers, Kent Mercker, Bill Bray, and (though I barely remember him being around) Matt Belisle.

Graphically Speaking

Again, I'll return to using graphs rather than tables to learn about the '06 Reds. I chose a cutoff of 24.3 innings for the following work, as there's a big drop-off from that total (Franklin) to the next guy on the list (Standridge).

How they got hit...
Vertical and Horizontal lines indicate league averages for OBP-allowed and hr/9 allowed.

This graph is a little unorthodox for evaluating pitchers, but I wanted to start with an analog to the OBP/ISO graphs for hitters. I think it nicely shows the two main ways a pitcher can be hurt: letting opponents get on base, and letting them hit for power. Clearly, the best place to be is in the lower-left, but pitchers can be successful in in the lower-right or upper-left (see Arroyo). If you find yourself in the upper-right, however, you're in bad shape.

Two principle observations:
  • The Reds had four pitchers in the lower-left, with Arroyo just outside the bounds. I'd rate those five guys as our five most reliable returning pitchers.
  • The Reds had a lot of players who allowed home runs at an unusually high rate last year. Part of this is undoubtedly the effect of Great American Ballpark, which routinely has a park factor for HR's among the highest in baseball, despite being only a slight hitters park overall. But we also just have a lot of guys who allow a lot of fly balls.

Walks vs. Strikeouts...

Strikeout rate vs. walk rate. The horizontal and vertical lines represent league averages for strikeout and walk rates. Observations:
  • The best place to be in this graph is in the upper-left: low walks, high strikeouts. Gratifyingly, the Reds had a number of pitchers in that part of the graph, including Harang, Arroyo, Coffey, Bray, and Lohse.
  • I was surprised to see Hammond up there, but his peripherals were actually pretty decent last season (high HR-allowed rate though). It's easy to forget, but Hammond had an outstanding May (0.79 ERA, 7.2 k/9, 0.8 bb/9, 0.8 hr/9) after a miserable April and a rough June, the latter failure leading to his release.
  • The Reds also, unfortunately, had a number of guys in the low-k, high-bb range, most notably David Weathers and Matt Belisle. Weathers, in particular, is concerning. Despite actually improving in ERA in 2006, he saw surges in his walk and home run allowed rates, as well as a drop in his strikeout rates last season. I'm very worried about him as we look toward 2007.

Clutch and choke pitchers for 2006

Unlike among hitters, VORP is not adjusted (afaik) for pitcher role, and therefore serves as an excellent comparison to WPA for pitchers:
VORP (based on classic stats) and WPA (based on win probability changes) for Reds' pitchers in 2006. Regression line is for Reds' pitchers only, as I couldn't easily get WPA and VORP for all NL pitchers.

Observations:
  • WPA, at least for Reds' pitchers, seems to track much more closely to classic scorebook stats (which make up VORP) than it does for hitters.
  • Nevertheless, Weathers, Mercker, and Bray all come out as having a greater impact on game outcomes than you'd predict from their stats. It's good to see a young guy like Bray performing well in that situation. It's also impressive to see Weathers doing so well after his mid-season swoon--he was easily our most successful reliever the last two months of the season, and won my "impact pitcher" awards both of those months.
  • Harang (sort of) and Ramirez seem to be the most notable underperformers. I noticed last season that Ramirez, even before his August implosion, was rarely bringing home much in the way of wins or WPA. This is reflected in his record: the kid went 4-8 despite turning in 10 quality starts. Perhaps we're seeing an indication of a lack of run support?

Surprises and Disappointments:
This graph shows ERA vs. PECOTA-predicted ERA from the 2006 BP Annual. While I didn't agree with all the predictions, they are generally close enough to what I expected to make for a useful comparison.

Observations:
  • Weird pattern, and I think it shows something about the Reds' pitching in 2006. While some pitchers performed expected levels, we had a number of major surprises and, unfortunately, a number of huge disappointments.
  • First, the surprises:
    • Arroyo, of course, leads the list, with an outstanding 3.29 ERA compared to a mid-4's PECOTA projection.
    • Another surprise came from Todd Coffey, who was absolutely brilliant in April (0.60 ERA) and May (1.80 ERA) before struggling from June through August (5.84, 5.59, & 5.79 ERA's). He did put together an excellent final month of the season (2.45 ERA), which helps me think he might yet be able to put up outstanding numbers again.
    • Weathers, Belisle, and Franklin also did better than expected...in Franklin's case, though, that's not saying much.
    • Harang was listed as a surprise, but I think he was underrated by PECOTA last year...and this year, for that matter.
  • As for the disappointments...
    • I think Claussen's implosion/injury was the biggest story that was forgotten last season. He began the season as our #3 starter after a very solid 2005 campaign (4.21 ERA in 167 IP), but was completely ineffective and, ultimately, was placed on the disabled list. I think it may have been a mistake to cut bait with him this offseason--he's still young (28 in May) and has had success in the past. The question is whether he'll be able to come back from his surgery. Apparently the Reds didn't think he could--we'll see if he does anything for the Nationals this year.
    • The Reds also received catastrophically bad performances from Rick White, Chris Hammond, Dave Williams, and Joe Mays. Williams was a fairly big disappointment given his status as the guy we got for Casey, although I mentioned in my early-season profile on him, his peripherals in 2005 were far worse than his respectable 4.41 ERA would indicate. Peripherals also hinted at some loss of skill for Chris Hammond in the 2005 season, so there's a way in which his implosion was also predictable. It's clear to me that Dan O'Brien's team didn't pay much heed to DIPS, and that was to their detriment.

The winds of fortune...FIP vs. ERA

Line indicates a perfect match between FIP and ERA.

Observations:
  • I have to admit that I'm a little disappointed in how well FIP predicted ERA. Compare this, for example, to the tighter fit between PrOPS and OPS (another reason why I think PrOPS is such a good stat). I've been meaning, for quite some time, to revisit and critically evaluate FIP. I rely on it a lot in my player evaluations, but sometimes I wonder about how good it really is.
  • Nevertheless, we can see some things pretty clearly:
    • Some players, most notably Arroyo, Belisle, Weathers, Mercker, and Michalak, grossly overperformed their peripherals. Mercker and Weathers had BABIP's below 0.250, which generally only happens with some outstanding luck. ... I continue to be really worried about Weathers' prospects this season.
    • Arroyo, it should be said, still looks to have had a good season based on his peripherals (low-4's ERA, tons of innings). Just probably not a should-have-been Cy Young contender.
    • A number of players may have pitched better than their ERA's would indicate, including: Kyle Lohse (I continue to be bullish on him this season, as long as he gets his hamstring issues worked out), Elizardo Ramirez, Chris Hammond, Rick White, and Brandon Claussen. Of course, White and Claussen were still predicted to have terrible ERAs, just not as terrible as they turned out to be.
Finally, we'll close with a stats table describing the Reds' 2006 season, sorted by VORP.
Player IP H/9 K/9 BB/9 HR/9 BABIP ERA FIP VORP WPA OBPa
Arroyo 240.7 8.3 6.9 2.4 1.16 0.270 3.29 4.14 64.9 3.16 0.293
Harang 234.3 9.3 8.3 2.2 1.08 0.312 3.76 3.64 50.2 1.92 0.308
Coffey 78.0 9.8 6.9 3.1 0.81 0.320 3.58 3.77 19.8 0.89 0.335
Weathers 73.7 7.5 6.1 4.2 1.47 0.227 3.54 5.28 18.9 1.57 0.309
Milton 152.7 9.6 5.3 2.5 1.71 0.270 5.19 5.35 10.2 -0.18 0.317
Lohse 63.0 10.0 7.3 2.7 1.00 0.323 4.57 3.85 9.5 -0.07 0.327
Belisle 40.0 9.7 5.9 4.3 1.13 0.299 3.60 5.12 9.0 0.15 0.361
Schoeneweis 14.3 5.7 6.9 5.0 0.63 0.205 0.63 4.26 8.6 0.77 0.300
Guardado 14.0 9.6 10.9 1.3 1.29 0.361 1.29 3.07 4.5 0.31 0.310
Mercker 28.3 8.9 5.4 3.5 1.91 0.247 4.13 5.83 4.3 0.63 0.317
Yan 15.0 7.8 4.8 4.2 2.40 0.205 3.60 6.62 3.2 -0.33 0.317
Michalak 35.0 10.8 2.6 4.1 1.54 0.283 4.89 6.33 2.9 -0.35 0.377
Bray 27.7 10.7 7.5 2.9 0.98 0.341 4.23 3.83 2.9 0.41 0.341
Cormier 14.0 13.5 3.9 2.6 1.93 0.340 4.50 6.00 2.5 -0.69 0.379
Franklin 24.3 10.0 6.7 5.9 1.11 0.329 4.44 4.57 1.5 -0.47 0.391
Ramirez 104.0 10.6 6.0 2.5 1.21 0.316 5.37 4.65 1.0 -1.75 0.344
Johnson 8.7 11.4 4.2 0.0 1.04 0.313 3.12 4.14 0.9 -0.13 0.316
Kim 6.7 9.5 5.4 0.0 4.05 0.190 5.40 4.12 0.6 0.00 0.250
Germano 6.7 10.8 10.8 4.1 1.35 0.389 5.40 7.87 0.5 -0.16 0.387
Gosling 1.3 6.8 6.8 6.8 6.75 0.000 13.50 15.97 -1.0 -0.07 0.429
Standridge 18.7 8.2 8.7 6.8 0.96 0.294 4.82 5.09 -1.2 -0.68 0.372
Shackelford 16.3 9.9 8.3 5.5 2.20 0.292 7.16 6.77 -1.8 0.19 0.380
Claussen 77.0 10.9 6.7 3.3 1.64 0.321 6.19 5.39 -2.7 -0.76 0.362
Hammond 28.7 11.3 7.2 1.6 1.57 0.337 6.91 4.40 -3.3 -0.41 0.328
Majewski 15.0 18.0 5.4 2.4 0.60 0.468 8.40 3.88 -3.7 -1.98 0.468
Burns 13.3 20.3 6.1 2.0 1.35 0.519 8.78 4.72 -3.8 0.01 0.500
White 27.3 11.2 5.6 1.6 1.65 0.322 6.26 4.90 -4.1 -0.86 0.339
Mays 27.0 13.3 5.3 4.0 1.33 0.367 7.33 5.07 -4.6 -0.76 0.400
Williams 40.0 12.2 3.6 3.6 2.03 0.302 7.20 6.84 -6.6 -1.10 0.381

7 comments:

  1. Thumbs up research and analysis J. Your site is by far my favorite Reds site to visit. Not that the others are bad of course, I love them, but yours just is a complete different way of looking at things than the others....

    ReplyDelete
  2. J just recently discovered this site and wanted to say keep up the great work. One of the most informative sites in stat comparison I have seen. Love it

    ReplyDelete
  3. Nice work. Where would we be without blogs like this? You won't get this kind of analysis from ESPN, yahoo or mlb.com. Awesome.

    ReplyDelete
  4. I'm glad to see Lohse's numbers on the positive side. I expect good things from him this year.

    Milton, on the other hand...

    ReplyDelete
  5. Yeah, I have to say, if we have more than one reasonable candidate for the #4 and #5 starting slots aside from Milton, I'd seriously consider trying to move him to the bullpen. His performance is dramatically better during the first trip through the batting order, so he might turn into an outstanding setup-type guy. ... 'Course, the flip side to that is that the Reds already have three pretty good lefthanders in their bullpen (Stanton, Bray, and Cormier), so maybe he wouldn't help much.

    Just think though--this is the last year of Milton's contract! :)
    -j

    ReplyDelete
  6. Hi Ken, great questions.

    In response to your first point, I didn't consider PERA because it includes h/9 as a "peripheral" in its calculations. There are two reasons I don't like that. First, h/9 is inversely correlated with k/9 and (to some extent) positively correlated with hr/9, so there's redundancy in there (though this might be addressed with their equation, whatever it is). Second, h/9 is influenced by BABIP, and I'm trying in that graph to explicitly get away from BABIP, because a major source of its variation (~40% IIRC) is due to randomness (luck), with the other parts being defense (~30%), park effects (~20%), and pitcher skill (~10%) (%'s are from an article by tangotiger and others...I can probably scrounge up the link if needed).

    I am planning to sit down and see if I can work up an alternative to FIP that uses k, bb, and hr rates via a regression approach. We'll see--I have my doubts about whether it can be improved.

    Your second point, taking into account park factors, is an interesting point and one that I didn't consider. 4-year regressed-to-mean park factor for runs at GABP is only 1.01 (http://gosu02.tripod.com/id103.html) so it probably wouldn't cause that big of an effect, though given that many of our pitchers are fairly severe fly ball pitchers, the 1.08 HR park factor might be inflating their performance more than is typical.

    I think one of the more interesting things about that first graph is how severely pitching ERA rose this past season. I'll be interested to see if it drops back down this year.
    -j

    ReplyDelete