This is the second annual forensic handicap of the first round of the MLB playoffs. Last season we went 2-1 for +1.71 units in the divisional series'. This was another positive regular season (which ends for me on September 1 - after the September callups arrive and teams stop going all-out to win, I bow out) in the bases forum at MajorWager and hopefully that trend will continue into the playoffs in advance of five months of annual baseball hibernation (and withdrawal).
Below, I outline some methods of handicapping the hitting, starting pitching, relief pitching and defense of the playoff teams, put it all together, come up with some series win probabilities and then contrast them with the current series prices. This isn't exactly how I do things in the regular season (someday I'll give all that away, but not today!), but the principles are in line and the methods outlined below are solid.
Let's explain our methods by going step-by-step through the Milwaukee vs. Philadelphia series. Then, we'll dive into the other three series without the lengthy explanations.
**Hitting**
As we did last postseason, we are going to be using runs created per 27 outs to break down the hitting prowess of each player and team. As mentioned in this space last season, runs are made of run elements, like singles, walks, homeruns, hit by pitches, doubles, stolen bases, etc., each of which has a historical value toward making up or contributing to a run. By breaking down each player in terms of the elements that he contributes, and then reconstructing a team figure based on the projected lineup, we are able to make projections that keep a couple of important goals in mind. First and foremost is the elimination of the luck factor - some teams may have aberrant event clumping - imagine that team A gets 9 hits, all singles, in a given game, one per inning. That team will be shutout. Imagine then that team B gets 9 hits, all singles, in a given game, but the hits all happen in the sixth inning. Team A and team B put together the same number of run elements, but team B was fortunate enough that its run elements were sufficiently clumped together so as to score, probably, six or seven runs. Team A performed just as well offensively as team B, and their runs created per 27 outs would reflect that fact. Enough event clumping (or a drastic enough lack of event clumping) can make season runs scored totals for some teams misleading, insofar as the goal is to project the offense.
Another factor that our approach can take care of here is lineup construction. The Red Sox season totals would be too high, because they would include a partial season of Manny Ramirez; the Angels season totals would be too low, because they would not include the first part of Mark Teixeira's season. Naturally, this approach accounts for injuries, too. So - our approach is to project each lineup and assign a runs created per 27 outs score for each player. Then, the figures for the team are averaged and adjusted for the hitting environment in the team's home ballpark. One final adjustment is made so that the team number reflects runs per 25.5 outs (because in roughly half the games, there is no bottom of the ninth). One final note: for players with fewer than 350 plate appearances in 2008, 2007's stats were added to produce a more reliable rate. Also, the pitcher's batting slot for National League teams is given a generic 2.6 runs created per 27 outs - which is a weighted average of pitchers and pinch hitters.
Let's take a look at the Brewers starting lineup: Jason Kendall (4.0 runs created per 27 outs), Prince Fielder (6.4), Rickie Weeks (4.8), J.J. Hardy (4.9), Bill Hall (3.6), Ryan Braun (6.0), Mike Cameron (5.5), Corey Hart (4.5), Pitcher (2.6). Those stats gives the Brewers an average of 4.69 with that lineup. We must correct that figure for Miller Park, which favors hitters very slightly, and our new figure is 4.67 runs per 27 outs - or 4.41 runs per the 25.5 outs featured in a typical game. This is not good, and, as you will see, it is the worst figure among Major League playoff teams. There are a couple of possible lineup changes: if Counsell plays over Hall, the figure dips ever-so-slightly to 4.39; if Durham plays over Weeks, the figure rises to 4.52, which is significant.
Let's take a look at the Phillies starting lineup: Carlos Ruiz (2.9 runs created per 27 outs), Ryan Howard (6.7), Chase Utley (6.8), Pedro Feliz (3.8), Jimmy Rollins (6.1), Pat Burrell (6.0), Shane Victorino (5.4), Jayson Werth (6.4), Pitcher (2.6). For the Phillies, the lineup average is 5.19, which gets adjusted downward for park factor to 5.07 runs created per 27 outs, or 4.79 runs per game (25.5 outs). Like Milwaukee, Philadelphia has at least one lineup spot somewhat unsettled that could have an effect: if Dobbs plays third instead of Feliz, the Phillies offense rises from 4.79 to 4.93. As the lineups look from here, however, the numbers give the Phillies a 0.38 runs per game advantage.
**Starting Pitching**
For reasons outlined in this space last year, we are going to use a statistic called xFIP, or expected fielding independent pitching, which is compiled by and available at hardballtimes.com. In as few words as I can describe it, xFIP is an ERA-like stat that attempts to correct for the quality of defense behind the pitcher and the rate at which the flyballs that the pitcher allows end up going over the fence. There are a couple of reasons why these adjustments are necessary. First, the rate at which balls in play (that is, anything but a walk, strikeout or homerun) become base hits is relatively stable over time and between pitchers; that is, pitchers tend to regress toward the league mean over time. By correcting for quality of defense, we put the blame (and the credit) where it belongs. The other adjustment is based on the fact that the rate at which flyballs become homeruns is also steady - the league average is around 10% - and pitchers (good ones and bad ones) regress to that percentage over time. By adjusting for defense and homeruns per flyball allowed, xFIP is essentially adjusting for luck and circumstance.
Let's take a look at the xFIP numbers, along with innings pitched per start, for the Brewers starting pitchers:
Gallardo (4.09 xFIP, 6.0 IP/GS), Sabathia (3.25, 7.2), Suppan (4.91, 5.7), Bush (4.91, 6.2)
And the Phillies: Hamels (3.78, 6.9), Myers (4.01, 6.3), Moyer (4.72, 5.9), Blanton (4.84, 6.0) Bullpen
The sixth and seventh men in the bullpen rarely see the light of day in the playoffs, and for that reason we'll ignore them. Not only that, but I've done a quick and dirty study on the amount and import of relief inning assignments and, not surprisingly, the better pitchers get more work (in both quantity and quality) than the poor ones. As I mentioned last year - to determine the strength of a playoff bullpen, weigh it according to the following scale: 37%, 28%, 20%, 10%, 5%. Using the cumulative season stats is a significant mistake because they include everybody, and also because they weigh innings pitched by those pitchers who pitch low leverage innings as much as those of pitchers who pitch higher leverage innings. For handicapping purposes, an inning of Francisco Rodriguez has a greater impact on winning or losing than does an inning of, say, Jason Bulger.
Looking at the Brewers bullpen (or at least my best guess as to who the manager has the most faith in), we see a pecking order of Torres, Mota, Shouse, Gagne, Villanueva. So, we are handicapping the Milwaukee bullpen as 37% Torres, 28% Mota, 20% Shouse, 10% Gagne and 5% Villanueva. Using xFIP, as we did with the starting pitchers, we get: Torres (4.25 xFIP, .37 weight), Mota (4.40, .28), Shouse (3.42, .20), Gagne (4.77, .10), Villanueva (3.96, .05). This composite Brewers bullpen xFIP, then, is 4.16. This is a poor bullpen for a playoff team, but not historically bad as it is sometimes portrayed in the media.
Here are the Phillies' bullpen xFIP's and relative weightings: Lidge (3.06 xFIP, .37 weight), Madson (3.66, .28), Romero (4.28, .20), Seanez (4.68, .10), Condrey (4.24, .05). The composite Phillies bullpen has an xFIP of 3.69. This is a very good bullpen in the playoffs, wherein Lidge, Madson and Romero are likely to see most of the important innings.
**Defense**
This will seem lazy, but we are simply going to lift the defensive team stats from hardballtimes.com. According to the site (using a process and numbers that I trust), the Brewers' defense was worth a whopping 60 runs relative to average, or 0.37 runs per game. The Phillies clock in at a slightly more modest 0.32 runs per game. These are both excellent defensive squads. Remember, incorporating defense is that much more important because we are using defense independent pitching measures.
**Bringing it all together**
So, what do we have? We know that the Brewers offensive projection is 4.41 runs per game, whereas for the Phillies it is 4.79. We know, for example, that in game one, Gallardo is matched up with Hamels. Gallardo is projected at 4.09 xFIP for 6.0 innings, which is 2.73 runs, leaving 2.75 innings for the bullpen, which rates at 4.16, so the bullpen is projected to allow 1.27 runs. When we add them together, we get 4.00. Now, we have to make a couple of adjustments to the 4.00. First, we multiply that number by 1.08, because our xFIP numbers are based on an earned run scale, and we have to get them onto a runs scale (i.e. adjusting upward because the ERA scale is artificially low because it does not account for earned runs): this gives us 4.31. Now we must account for the defense, so for the Brewers we subtract 0.37, leaving us with 3.94.
So here is what we know: with the Brewers starting Gallardo, against an average opponent, we can project 4.41 runs scored and 3.94 runs allowed. After putting this through Bill James' Pythagorean theorem, we can get an idea of the strength of the Brewers, with game one's lineup and pitcher - the formula is RS^1.83/((RS^1.83)+(RA^1.83)). The result: the Brewers are a .551 team in this situation. Now, we can do the same for the Phillies. As discussed earlier, the Phillies' offense rates at 4.79, and with Hamels (using the same calculations and adjustments as above with Gallardo and the Brewers), we get a runs allowed projection of 3.63. Yeah, Hamels and the top of the Phillies bullpen are good. Putting those numbers through James' formula above, the Phillies are a .624 team in this spot.
Let's go through the other starting pitchers: with Sabathia pitching, the Brewers rate as a .641 team, with Suppan they are at .490, and with Bush .485. For the Phillies, we get Myers at .604, Moyer at .551, and Blanton at .541. Now, to handicap game one, we use another of James' formulas (I won't write it out here, but feel free to google it) called Log5, which gives you the probability of one team with a given winning percentage beating another with a given winning percentage. The formula tells us that a .664 (we've added .040 to the Phillies in game one with Hamels because they are at home) team should beat a .551 team 61.6% of the time.
Let's put it all together one last time: The Phillies have a .616 chance of winning game one with Hamels vs. Gallardo. The Phillies have a .503 chance of winning game two with Myers vs. Sabathia. The Phillies have a .522 chance of winning game three with Moyer vs. Suppan. The Phillies have a .516 chance of winning game four with Blanton vs. Bush. The Phillies have a .525 chance of winning game five with Hamels vs. Sabathia. So the only game that is significantly more than a coin flip is game one, but the Phillies have small advantages throughout the balance of the series that certainly could compound. There are ten ways to win a five game series, for example a team can win games 1, 2 and 3, or win games 1 and 2, lose game 3 and win game 4... you get the point. There are ten permutations, and we can calculate the probability of each using our expectations above. The chance of a Phillies sweep is .616*.503*.522, or 16.2%. The chance of the Phillies winning games 1, 3 and 5, but not 2 and 4, are 4.1% (.616*.497*.522*.484*.525). By taking all ten ways that the Phillies can win and adding up the probabilities of those scenarios, we get the Phillies probability of winning the series, which in this case is .568, meaning that the Brewers have a .432 shot. If we convert those to a "juice-free" moneyline, we get +/- 131. Pinnacle currently has Milwaukee at +162, so that's a play. So while the Phillies are certainly more likely to win, there is significant value in the Brewers at +162.
**Los Angeles vs. Chicago**
Okay, before we go any further, I have to point out that this is the one series where I am going to lose credibility. You will think me a heretic. But here it is: the wrong team is favored. The Cubs have been a much better team to this point in the season, but as we walk through the steps I hope that you'll be able to see what I see: that the Dodgers are very well built for the playoffs. None of this is subjective - the numbers don't lie.
The Dodgers Offense: Martin (5.8 runs created per 27 outs), Loney (4.8), DeWitt (5.1), Furcal (5.2), Blake (5.8), Ramirez (9.3), Kemp (5.2), Ethier (7.1), Pitcher (2.6). After our adjustments, we get 5.23 projected runs. The Cubs Offense: Soto (5.8), Lee (5.2), DeRosa (6.7), Theriot (4.8), Ramirez (7.0), Soriano (6.0), Edmonds (4.8), Fukudome (5.3), Pitcher (2.6). Adjusted, that comes out to 4.96.
The Dodgers rotation: Lowe (3.43 xFIP, 6.2 IP/GS), Billingsley (3.74, 6.1), Kuroda (4.02, 5.9), Maddux (4.18, 5.9). The Cubs rotation: Dempster (3.94, 6.3), Zambrano (4.68, 6.3), Harden (3.70, 5.9), Lilly (4.39, 6.0).
The Dodgers bullpen rates out at a superb 3.35 xFIP, thanks to a weighted average of Broxton (2.91), Saito (3.00), Park (3.96), Wade (4.07), Beimel (4.77). The Cubs bullpen, while excellent, falls short of the Dodgers standard at 3.68, based on Wood (3.12), Marmol (3.87), Howry (4.36), Cotts (3.56), Gaudin (4.30).The Dodgers defense rates a .06, the Cubs a bit better at .17.
What does this mean on a game-by-game basis? The Dodgers have a .527 chance of winning game one with Lowe vs. Dempster. The Dodgers have a .563 chance of winning game two with Billingsley vs. Zambrano. The Dodgers have a .542 chance of winning game three with Kuroda vs. Harden. The Dodgers have a .584 chance of winning game four with Maddux vs. Lilly. The Dodgers have a .527 chance of winning game five with Lowe vs. Dempster.
Put all this together, and the Dodgers have a .591 chance of winning the series. So a fair price would be Dodgers -144. You can get +184 on the Dodgers at Pinnacle as I write this. Why??? Here are three reasons:
1. In the regular season, the Dodgers gave Juan Pierre and his awful .327 on-base and .328 slugging percentages 119 games worth of at bats. They have Manny now. With Pierre in the lineup instead of Ramirez, the offensive projection would fall from 5.23 to 4.76. This is huge. When you look at the season stats, including the modest 84 wins - those are mostly the Pierre Dodgers, whereas these Dodgers are the Manny Dodgers. 2. In the regular season, the Dodgers played Angel Berroa, who is a rare Major Leaguer in that he's a worse hitter than Juan Pierre, for 84 games. Now, they have Rafael Furcal, who played only 36 games this season and is thus barely counted in the season stats, back. Add Berroa to that Pierre-in-lieu-of-Manny offense, and you get 4.45 - contrast that with 5.23 in its current form. The season stats are meaningless, because they are bogged down with the numbers of players who will be on the bench in the playoffs. 3. The Dodgers bullpen is top heavy - with Broxton and Saito at the top - and that is a big part of the formula for out-of-nowhere playoff success.
This series is about the Dodgers as they are composed right now. Virtually every pitcher that will pitch an important inning (either starter or relief) is of ace or near ace quality, the lineup is, for the first time, at its best, and Cubs are using their best pitcher only once, in game three. These are not the 84 win Dodgers - that team no longer exists; in its current form this is a very, very good team, made only better by the playoffs' disregard for depth issues. This projection obviously puts a fair bit of faith in Joe Torre to play the right players (because the wrong ones are really, really bad), something with which I am not entirely comfortable.
**Boston vs. Los Angeles**
The Red Sox offense: Varitek (2.9 runs created per 27 outs), Youkilis (8.3), Pedroia (6.2), Lowell (5.0), Lowrie (4.6), Bay (6.8), Ellsbury (4.6), Drew (7.0), Ortiz (7.0), for an adjusted team figure of 5.32.
The Angels offense : Napoli (6.1), Teixeira (7.7), Kendrick (5.4), Aybar (5.1), Figgins (4.6), Anderson (5.7), Hunter (5.2), Guerrero (6.4), Matthews (3.8), for an adjusted team figure of 5.13.
The Red Sox rotation: Lester (4.19 xFIP, 6.4 IP/GS), Matsuzaka (4.82, 5.8), Beckett (3.35, 6.5); the Angels three-man rotation (this is the "long series", spread over more days, therefore there can be three-man rotations without necessitating pitching on short rest) is Lackey (4.00, 6.8), Santana (3.64, 6.8), and Saunders (4.76, 6.4).
The Red Sox bullpen rates out at 3.49, thanks in large part to Papelbon (who, saves record notwithstanding, is superior to Rodriguez): Papelbon (2.51), Delcarmen (3.82), Okajima (4.24), Masterson (4.28), and Lopez (4.34). The Angels bullpen is fine too: Rodriguez (3.71), Arredondo (3.75), Shields (3.75), Oliver (4.20), and O'Day (4.05). The Red Sox defense rates at .04; -.03 for the Angels, for virtually no difference.
What does all of this mean?
The Angels have a .523 chance of winning game one with Lackey vs. Lester. The Angels have a .597 chance of winning game two with Santana vs. Matsuzaka. The Red Sox have a .692 chance of winning game three with Beckett vs. Saunders. The Red Sox have a .561 chance of winning game four with Lester vs. Lackey. The Angels have a .597 chance of winning game five with Santana vs. Matsuzaka.
All of this adds up to the Red Sox as a slight favorite, with a .513 chance of winning the series. How is this possible when the Angels should be favored in three of the games? That is because of the total mismatch of Beckett vs. Saunders at Fenway in game three, and the near coin-toss in game one. Insofar as Beckett is healthy and ready to go in game three, the +117 available on the Red Sox at Pinnacle (as I write this) has value. There is a pretty serious caveat here: the Red Sox have injury concerns and the above is based on the assumption that they will be healthy. It is possible that we could see Mark Kotsay in the lineup rather than either Mike Lowell or JD Drew (which would be bad for the offense), and just how much of Beckett we get is apparently an open question. If one of the hitters is gone, or if Beckett is either limited or absent, then there probably isn't any real value in the series line either way.
**Chicago vs. Tampa Bay**
This one is, of course, done on short notice, so it contains some educated guesses on the White Sox rotation. The White Sox offense will look like this: Pierzynski (4.2 runs created per 27 outs), Konerko (4.5), Ramirez (5.5), Cabrera (4.3), Uribe (4.4), Wise (3.7), Griffey (5.3), Dye (5.4), Thome (6.3). This rates out to an adjusted level of 4.48, which is a mediocre American League playoff offense. Playing Swisher over Wise would help, but Ozzie Guillen appears unlikely to do that.
The Rays, for their part, feature Navarro (5.0), Pena (6.7), Iwamura (4.7), Bartlett (4.5), Longoria (5.8), Crawford (4.6), Upton (5.7), Hinske (4.9), Floyd (5.5). The White Sox rotation (or my best guess at who Guillen will go with) is Vazquez (3.96 xFIP, 6.3 IP/GS), Buehrle (4.10, 6.4), Floyd (4.62, 6.3), and Danks (3.94, 5.8). The Rays have announced that they will go with Shields (4.06, 6.5), Kazmir (4.23, 5.6), Garza (4.59, 6.2), and Sonnanstine (4.47, 6.0). The White Sox bullpen, weighted as described above, rates a 3.65 with the help of Jenks (3.88), Linebrink (3.77), Dotel (3.38), Thornton (2.75), and Carrasco (4.16). The Rays fall short, at 3.95: Wheeler (4.53), Balfour (3.04), Howell (3.63), Bradford (3.97), and Percival (5.98). The White Sox have an average defense (.01 runs per game), whereas the Rays sport a better one (.15).
Let's bring it all in:
The Rays have a .591 chance of winning game one with Shields vs. Vazquez. The Rays have a .591 chance of winning game two with Kazmir vs. Buehrle. The Rays have a .521 chance of winning game three with Garza vs. Floyd. The White Sox have a .527 chance of winning game four with Danks vs. Sonnanstine. The Rays have a .591 chance of winning game five with Shields vs. Vazquez. This would give the Rays a .604 chance at moving on to the ALCS. Pinnacle doesn't have a line up yet, but The Greek has the Rays favored at -160/+135, which (not surprisingly for a 25 cent line!) doesn't offer value either way as our .604 projection (.604 converts to -153).
So, we have three plays:
**Brewers +162, Dodgers +184, and Red Sox +117.**
10-01-08 Spraguer Poster-at-large at MajorWager.com You may e-mail Spraguer through the editor of MajorWager at: editor@majorwager.com
If you would like to make or read comments about this article, you may do so by visiting the Mess Hall forum at MajorWager where a thread has been started. Please click HERE |