|
|
February 2009 - Posts
-
In Baseball-Reference box scores, you can find a goofy little stat called "wWE," which means "Winner's Win Expentancy." Quite simply, in their words:
The current probability (after the play) of the eventual winner winning at this point in the game.
wWE is a Win Probability Added stat, which is a group of stats that attempt to analyze the progression of probability of winning throughout a game, allowing us to judge the value of certain plays. Again, in their words:
These statistics generally look at the game context at the start and the end of the play and compare the batting team's probability of winning the game in both situations.
For example, in the top of the eighth, the visiting team might be down five with one out and runners on first and second. The batter then hits a home run to bring the visiting team to within two runs, still with one out, but now with no runners on base. Prior to the home run, the batting team had about a 3% chance of winning which improved to 10% following the home run. This change of 7% is credited to the batter and debited to the pitcher. Compute these for every play and every game from 1956 on and you have win probability added stats.
This stat does something that a lot of people criticize many advanced stats for not doing. It considers the context of the game. A home run with the score tied is worth much, much more than a home run with a ten-run lead. In fact, a walk or run scoring ground out in a tie game is worth more than a home run in a blowout.
***
Now, looking through game logs, we have a very valuable reference for judging plays: we can judge the value of crucial plays throughout the season; which clutch plays were most clutch; which winning plays were least expected, most crucial, etc.
My survey in the coming week is more mundane and simple: Working through box scores, I am looking at each Brewers win, and evaluating the point at which the Brewers kept the lead for good -- sometimes a game-winning hit can occur in the first inning, other times a game-winning hit can occur in the 12 inning. The beauty of wWE is that we can capture the value of game-winning hits across the spectrum of game situations.
Obviously, plays the end the game immediately -- walk-off hits -- will usually be more valuable because they end the game immediately, making it impossible for the other team to come back. But inbetween the 1st and extra innings, we can find a lot of unique plays.
Here are the Brewers' April victories, and the Probability shift of the winning plays:
April 19, 2008 (@Cin): Bill Hall (T10); double +45%
April 23, 2008 (vs. PHI): Prince Fielder (B8); home run +39%
April 17, 2008 (@ StL): Prince Fielder (T10); home run +35%
April 22, 2008 (vs. STL): Gabe Kapler (B12); single +30%
April 26, 2008: (vs. FLA) Prince Fielder (B8); home run +26%
April 8, 2008 (vs. CIN): Rickie Weeks (B10); single +18%
April 13, 2008 (@ NYM): Rickie Weeks (T6); home run +17%
April 12, 2008 (@ NYM): Rickie Weeks (T5); home run +15%
April 29, 2008 (@ ChC): Bill Hall (T5); single +14%
March 31, 2008 (@ ChC): Tony Gwynn (T10); sacrifice fly +13%
April 2, 2008 (@ ChC): Rickie Weeks (T1); home run +12%
April 5, 2008 (vs. SF): Mike Rivera (B2); single +10%
April 4, 2008 (vs. SF): Prince Fielder (B1); single +8%
April 6, 2008 (vs. SF): J.J. Hardy (B2); double +7%
April 18, 2008 (@ Cin): Prince Fielder (T1); sacrifice fly -1%
Series Victories
@ Cin (April 18-19): +44% (22% shift per win)
@ NYM (April 12-13): +32% (16% shift per win)
@ ChC (March 31, April 2): +25% (12.5% shift per win)
Vs. SF (April 4-6): +25% (8.3% shift per win)
Single Victories (lost series / two game series)
Vs. PHI (April 23): +45%
@ STL (April 17): +35%
Vs. STL (April 22): +30%
Vs. FLA (April 26): +26%
Vs. CIN (April 8): +18%
@ ChC (April 29): +14%
The Brewers' series at New York in April of 2008 probably featured the most consistently crucial winning plays, right in the middle of very close ballgames. Both were Rickie Weeks home runs, one in the 5th, the other in the 6th, both providing the Brewers' the lead for good. Less than a week later, the series at Cincinnati featured the month's most-clutch, most crucial hit, a Bill Hall double in the top of the 10th (worth a 45% shift in winning probability).
One of the features of the Brewers' first half -- prior to acquiring Sabathia -- is playing above their expected win/loss record; the Brewers played roughly 4 wins above their Pythagorean and Expected W-L records, based on runs allowed and runs scored, before acquiring Sabathia in mid-July, and I think we can see why in April: although the Brewers had a few series in which they only won a single game, five of those single-game victories were above the median probability shift (the median was 15%).
In these single-victories, the Brewers stole away some very, very close games, winning in late game situations and extra innings situations; it is my hypothesis that these trends of stealing away close-game victories -- even in the midst of losing the overall series -- helped to shift the Brewers' W-L. (It also helps that the Crew won most of their blown saves in April).
Of course, April also featured some mundane victories -- including the entire sweep of the Giants at Miller Park, where the average probability shift on winning plays was under 9%. This means that the Brewers won those games early, never to relinquish the lead; those games truly were never in question.The other mundane victories occurred at Chicago to open the series, where the Tony Gwynn sacrifice fly and Rickie Weeks' game-opening home run served as rather low leverage winning plays.
Outside of the two opening series (one at home, one on the road), the Brewers had a rather hectic April, winning games in close situations, and especially late situations. *** Citation Baseball-Reference. "Win Probability Added (WPA) Stats." Win Probability Added Glossary. October 28, 2008. Accessed February 28, 2009.
-
Today we have a quick morning coffee and baseball stat: which Brewers starting pitchers allowed balls-in-play most frequently in 2008, and who had the best defensive support? I calculated pitching BIP% by reading each pitcher's Baseball-Reference 2008 splits, calculating (PA-HR-BB-K-HBP)/(PA). I broke the trend with Gallardo, calculating his entire career, rather than an injury-shortened 2008. There is simply no meaning to analyzing 100 PA of work from a pitcher.
BABIP / BIP% Looper: 842 PA, 653 BIP; .296 BABIP / .776 BIP% Suppan: 780 PA, 589 BIP; .306 BABIP / .755 BIP% Bush: 763 PA, 567 BIP; .238 BABIP / .743 BIP% Villanueva: 464 PA, 320 BIP; .302 BABIP / .690 BIP% Gallardo: 563 PA, 386 BIP; .302 BABIP / .686 BIP% Parra: 749 PA, 499 BIP; .333 BABIP / .666 BIP% McClung: 456 PA, 297 BIP; .284 BABIP / .651 BIP% Isn't it odd that the progression from more-to-less balls-in-play almost perfectly follows the progression from low-to-high top-speed velocity and pitching style for Brewers' pitchers? The sinkerballers, who throw slightly below average top-speed, or around average top speed, allow more BIP; but by contrast, a player with a slightly below average top speed who throws a fastball and change up (Villanueva) allows less frequent BIP% than pitchers with similar velocity range, but different approaches (i.e., Looper, Suppan, and Bush). The hottest top-speed on the team -- Gallardo, McClung, and Parra -- each allow the fewest BIP% on the team, although Gallardo's is very close to Villanueva, which I attribute more to pitching style and less to top-speed velocity. A more in-depth study is in order -- how closely to velocity and/or pitching style affect BIP%? I mean, the answer is as plain as the nose on your face that how a pitcher pitches, and how fast he pitches, will affect balls-in-play -- but we could use available pitch f/x data to compare BIP% between pitchers with different primary / secondary pitch combinations (i.e., change-up / fastball, fastball / sinker, fastball / slider, sinker / slider, fastball / curveball, etc.). This is probably another prime area where statistical analysis supports baseball common sense -- how you pitch, and how fast you pitch, affects opposing batting approaches.
-
Dave Bush had one hell of a 2008 campaign -- his best of his career. Deep in the shadows, though -- behind ERA, behind WHIP, behind W-L, behind all that good stuff, we might be concerned by one simple fact. 2008 was one hell of a lucky season for Dave Bush. You know that recent obsession with balls-in-play I've written about in the last couple of weeks? Well, the same goes for pitchers: if pitchers change the percentage of their balls-in-play, or the defense behind them changes on those balls in play, everything else can change for the pitcher as well. To that effect, Bush's 2008 BABIP-against? .238. That's right, .238; a full 5% better than average defensive efficiency. For a pitcher such as Bush, the fact remains that BABIP will be especially important because he does not consistently strike out 8-10 batters per 9 innings. In fact, the 2008 BABIP is all the more significant because his K/9 IP dropped in both 2007 and 2008. Luckily, we can hang on to a couple of things: (1) Dave Bush doesn't walk anybody. Ever. This will keep a consistent factor in his approach even if his hits fluctuate greatly from year to year; you don't want a pitcher putting a ton of guys on base through things he can control (like walks), when there will be a fluctuating amount of men on base through means that that pitcher cannot control -- defensive support. So, here goes: H/9 IP (BABIP) Career: 9.03 (.282) 2008: 7.93 (.238) 2007: 10.48 (.323) 2006: 8.61 (.286) 2005: 9.38 (.280) 2004: 8.80 (.278) 2008 seems to be the logical shift in a series of volatile years for Bush; while his defensive support was virtually nowhere to be found in 2007, it showed up with flying colors in 2008. Add to that a slightly shifting number of balls in play, and you clearly hit levels that are absolutely all over the place. A correction might be in order; but let's think positively for a moment. (1) Bush threw his fastball approximately 3% less frequently in 2008 than in 2007, and nearly 4% less frequently than his career usage of that pitch. By contrast, he distributed those extra pitches to his change up and sinker, rather than his curveball; he used his change up over 2% more frequently than both his 2007 and career usages of that pitch. Similarly, he used his sinker more frequently (FanGraphs calls it a "cutter"), by about 3%, and used his curveball 3% less frequently. (2) His sinker, curveball, and change up were slower than his 2007 and career velocity averages for those pitches. Two conclusions: Bush's pitching sequences were notably different in 2008, which probably made him a slightly more deceptive pitcher, especially coupled with shifts in velocity for some of his pitches. When Bush gained mechanical confidence by mid-season, and began relying on his change up (a well-publicized shift by Brewers' media outlets), his results changed for the better -- way, way, better. Bush's change up, anyhow, is a unique pitch -- it is within the velocity range that a slider would serve for a power pitcher; further, it seems to be a change-up off of his sinking fastball and his 4-seam fastball alike (meaning that throwing two different fastball-change up combinations creates two different velocity differentials for batters, and effectively two different outcomes for that change-up; changing off a fastball and changing off a sinker are two different approaches, and there is no reason to suspect that Bush does not use his change up in both ways). Bush's defensive luck probably will correct itself in 2009. These things have an odd sort of determinism about them; average defensive efficiency has the sort of powers over pitchers that a vital, creative force has on the universe. What seem to be coincidences probably are not coincidences when the hand of God is entered into the equation; and yet there is a tension in Bush's 2008, between lucky efficiency and noted pitching approach changes. I suspect the struggle between pitching approach and luck continues in 2009 for Bush; as an optimist, I see great things happening when Bush sticks with his change up. It adds a third velocity level in his pitching approach, between his fastball and curveball, and off of his fastball and sinker. And hopefully, it inspires consistently high levels of defensive efficiency.
-
While I'm on this kick about pitchers strengthening their arms by pitching smaller workloads more frequently, I'm becoming more and more intrigued at the thought of utilizing a modified four man rotation with multiple swingmen to solidify middle relief. A perfect example of such a pitching staff -- capable of supporting four starters and three swingmen -- is the 2009 Brewers staff, with Gallardo, Parra, Bush, and Suppan certainly the best candidates for starting on our team, and Looper, Villanueva, and McClung with experience as starters and relievers (in some way or another). Here's the basic idea. Take Dave Bush. In 2008, Bush pitched in 31 games, starting 29, pitching 185 IP and throwing 2741 pitches (14.82 NP/IP). Now, the idea of a modified four man rotation would not be to pitch all four starters at full tilt, but rather to utilize their usual pitch workload and simply redistribute it. So, in Bush's case, a redistribution looks like 2741 pitches over 40 games -- instead of 31 -- which amounts to approximately 68 NP/G, or approximately 4.59 IP/G. That looks pretty rough, but here's the kicker -- the secondary rotation, the rotation of swingmen, would function in a manner similar to a starting rotation, and feature an extended middle reliever on most days (in order to rest guys consistently pitching in relief on two days' rest, the remainder of the bullpen would need to take IP approximately once every 10 games or so to give the swingmen some extra rest). While most starters pitch between 86 and 91 pitches per start, and then utilize four days of rest, the swingmen would have their workload constructed based on two days' rest -- meaning that their workload would probably stand around 45 pitches (approximately 3 IP of work). So, in the case of Bush pitching approximately 4 and two-thirds innings, a swingman would then be able to take over and pitch approximately 2 and two/thirds or 3 IP in relief -- thus covering approximately 7 IP of the ballgame. Specialized relievers could then handle the remainder of the game. The whole point of this exercise is to redistribute pitching workloads and the quality that comes along with those workloads. (1) What would occur if a team stopped concerning themselves with searching for a suitable fifth starter? I do believe there is a qualitative difference between running a rotation 1-4 continuously, versus utilizing a fifth starter consistently. (2) If resources are scarce (in terms of quality 5th starters), those resources should be manipulated in a manner that is advantageous to a team's performance -- it might actually be more useful to pitch certain pitchers as extended swingmen in the bullpen, rather than as fifth starters. (3) The average starting pitcher no longer pitches over 6 innings as it is; redistributing the workload between a team of starters and a team of swingmen would have the advantage of recognizing that fact, changing pitching roles, and subsequently changing middle relief patterns and strategy. (4) As much as teams struggle to find good 5th starters, teams also struggle to find viable middle relievers. By moving low rotation starters -- who might not be best suited as starters themselves -- into the bullpen, the issue of low rotation quality and middle relief stability is simultaneously managed. The average swingman's workload for this plan would probably not be terribly taxing, either. Over a season that is approximately 180 days long, with extra bullpen rest days thrown in every 10 games or so, the average time period for swingman appearances would be one appearance every 3.6 days. With days off accounted for, even the starters would pitch one appearance approximately every 4.4 days. With these factors considered, the average swingman workload might look something like 49 G, 131.7 IP, 2107 NP. A fair workload if you factor in the pitcher's previous record as a starter, and factor in pitching endurance and the distribution of smaller pitch workloads, more frequently.
I know this probably strikes some as an awfully bad looking idea, especially since we are conditioned to yearn for the days when starters completed most of their starts, and pitched tons of innings for their respective teams. But the pitching trends are working in the other direction, and rather than rejecting that trend and trying to find four pitchers who can complete over 50% of their games in this era of strategy and increasing athletic ability and training resources, I think that recent trends can be manipulated in the other direction: that of redistributing pitching workloads and SP/middle relief roles entirely. While it might be against all sentimental value about baseball, it might also be a pretty good way to use a small core of relatively few pitchers to get a team through 7 innings of a ballgame.
-
(1) I am intrigued by Gallardo's 2009 campaign, and what it promises. The young ace-in-training has more talent than any pitcher on the Brewers' staff, taking into account both his control and his stuff, as well as his results thus far. He also accmulated the largest bag of question marks in the rotation, given his freakish 2008 and his low IP total, raising questions about his health and his stamina to complete an entire season in 2009. Part of me thinks that it is rather clear that Gallardo can and will be an above average pitcher even in limited time in the rotation -- but my real concern is whether or not questions about health and stamina will eventually override that performance. On that note, I believe that the fact that Gallardo did not play winter ball will impact his ability to place innings on his arm in 2009. Normally I agree with arguments for watching a young pitcher's innings and pitch workload. But Gallardo doesn't present as a normal case after 2008. What rests in my mind is conventional evidence -- uttered by some former players (most notably Tommy John) and coaches; evidence that states that if a pitcher wants to strengthen his arm, rest is not the way to do it. I would, in that regard, feel better about Gallardo's chances to succeed in 2009 if he had pitched a tour in a winter league bullpen, perhaps pitching long relief, working on pitching small amounts or small workloads frequently. I don't believe Gallardo needs to be rushed right in to the starting rotation. The Brewers have enough starting pitchers and enough days off in April and May to start Gallardo periodically, and then build up his stamina through other more frequent appearances from the bullpen. It'd also be advantageous to control his appearances early in the season, to keep him from pitching in cold weather or inclimate conditions. That said, I don't believe stamina will keep Gallardo from pitching at an above average level. I am just not sure that we ought to expect a full ace season from him, either. Between the bullpen and the rotation, Gallardo can eat valuable innings for the Brewers, and build up that arm for a full 2010 campaign. I firmly believe that Gallardo's treatment by the coaching staff, and placement in the rotation, must be towards the future. I believe Macha has already captured this somewhat by not pitching Gallardo first in the rotation; now it's time to move beyond that and put him in a situation where his stamina does not harm him. Manny Parra's 2008 would be a perfect example of what not to do -- simply allow a young pitcher to take a mound continuously, without any clear plan in place to limit IP and pitch workloads in the bullpen or minors. It doesn't require much -- just use Gallardo as a swingman in April, and begin pitching him more regularly into May, to prepare him for four full months of starting. I do not believe that being preventive or pre-emptive with Gallardo's arm can be a bad thing. (2) Yovani Gallardo: 30 G, 24 GS, 159.7 IP, 144 H, 73 R, 137 K/65 BB; 14 HR, 3 HBP; Average defensive efficiency (between .295 and .305 BABIP). 4.11 runs average, 5.32 IP/G. 15 quality starts (.625).
I see no reason to expect Gallardo to receive anything other than average defensive efficiency, given the fact that he will probably rely on the defense less than any other Brewers' starter (due to his strike out rate). His control is slightly affected by a larger workload, but with an average defense behind him, he is not punished by a slight dip in control numbers (see K/BB). As an extension of not expecting an ace season out of Gallardo, I also do not believe that the lack of an ace season will harm the Brewers' chances to compete. This season is much more about quality-depth than ace-quality in the rotation, and the Brewers should take advantage of this situation by easing Gallardo into a starting role, and allowing him to pitch four full months in the rotation.
More Posts Next page »
|
|