Hamels vs Martinez
Primum non nocere – the Latin phrase meaning “First, do no harm.” That sentiment is drilled into physicians from the beginning of medical school, but perhaps it should be emblazoned on every dugout wall in baseball as well. In the ALCS, we saw two managers over-analyze situations, causing harm to their own team time and time again, and now it appears that Charlie Manuel may be on the verge of following in their footsteps.
Manuel will send Cliff Lee to the hill on Wednesday for Game One of the series, as he should. However, he has not been willing to disclose the rest of his playoff rotation, and reports are surfacing that he may go with Pedro Martinez in the second game in lieu of Cole Hamels. The rationale appears to be based on recent performance – Hamels has struggled in the first two rounds of the playoffs, while Martinez was able to hold down the Dodgers in his NLCS appearance.
However, at some point, you have to step back and realize that recent performance simply cannot overcome what we know about the respective abilities of the two pitchers. Hamels is the better pitcher by a significant margin. It’s not even close.
The appearance of Hamels’ struggles this year are mostly just noise. His 2009 FIP of 3.72 is exactly equal to his 2008 FIP, as his walk, strikeout, and home run rates are nearly identical. The ERA jumped by over a run per game due to a 55 point increase in his batting average on balls in play. He was a bit lucky last year and a bit unlucky this year. Overall, he’s the same guy he was a year ago when he led the Phillies to a World Series title and was being crowned the new young ace of the National League.
Martinez simply isn’t in that league anymore. He did a nice job down the stretch for the Phillies, but even in a limited sample of nine starts, he wasn’t as good as Hamels. He pounded the strike zone and missed some bats, but his fringe fastball up in the zone led to a ridiculously low 29.5 percent ground ball rate. Not surprisingly, he gave up a bunch of home runs, which is consistent with his performances over the last several years. He throws enough strikes and has a good enough change-up to be effective, but he’s prone to throwing a couple of meatballs each game.
To decide to go with Martinez over Hamels, you have to significantly undervalue the gap in talent and overvalue everything else. Pedro may have a history with old Yankee stadium, and you might prefer a guy with his personality on the mound, but that stuff doesn’t make up for the fact that Cole Hamels is a far superior pitcher to Pedro Martinez right now. Mike Scioscia made a similar mistake by going with Joe Saunders due to non-talent reasons. It’s not a good idea.
The Yankees have it right – throw your best pitchers as often as possible. Hamels is the second best pitcher on the staff. He should start game two. It doesn’t have to be any more complicated than that.
Dave is the Managing Editor of FanGraphs.
Not to mention, the Yankees are Pedro’s daddy.
Good article. Do you think Manuel will go with Blanton and Happ out of the pen again this series?
I looked into the hitting difference. With wOBAs of 0.177 and 0.099 I get about 0.068R/PA, at most 0.2 runs over the course of a game. Hamels has an edge of 0.5 runs based on FIP (6 innings). This can be looked at two ways, Hamels is a 0.5 run improvement over Martinez in NYS, but a 0.7 run gain in Philly.
Then again against the Yankee 3-man rotation I don’t think any pitcher batting would make a difference.
Obviously you’ve never seen Brett Myers bat against CC. 🙂
that’s wierd that his wife’s name is CC.
oh, no he didn’t….
While your sentiment is right, i wonder if it will have a huge difference. As long as say, Blanton throws once hypothetically and Lee, Martinez and Hames throw twice (you could switch Blanton and Martinez), the end result is the same.
Now lets do something that is simplistic (i would love to do the math deeper but have all of 11 minutes to write this properly)
Now lets say philly is a slight underdog in every game (53-47 or so, if they throw their ace against the yankess ace, their #2 vs the yankees #2 etc.)
That would make them a 56.5-43.5 underdog or so.
Now lets say they do something “insane” and throw pedro game 1 and 5, with lee 2&6, hollywood hames 3&7 and blanton game 4.
Their odds in games 1 and 5 are now say 40% and the other games (since lee is going up agianst burnett etc etc) are true 50-50 games.
What are their odds now?
well they have a 3/8 chance that pedro’s performances are irrelevant (either because they won 4 out of 5 non pedro games or lost 4 out of 5.
their odds of winning 4+ out of 5 are 18.8%
their odds of winning 3 out of 5 non pedro starts and at least one of pedro’s two starts is conveniently exactly 20%
their odds of winning 2 out of 5 non pedro starts and having pedro win both of his starts are only 16% (.4*.4) * .313 or 4.96% (call it 5%)
so that makes it 43.8%, or about the same as the 43.5% chance by matching Ace vs Ace.
I dont know if those are the exact numbers (its a VERY hasty approximation), and even tinkering with the results slightly can make the data say whatever you want (i tried figuring pedro cost his team 7% in each start going up against an ace for -14% versus 5 games of non pedro for +3%*5)
but the short version of this is that which game pedro pitches is unimportant, but what is key is that both lee and hamels pitch 2 games each.
You beat me to it. If they both pitch the same amount of games, it doesn’t matter. Am I missing something here?
“The Yankees have it right – throw your best pitchers as often as possible. Hamels is the second best pitcher on the staff. He should start game two. It doesn’t have to be any more complicated than that. .”
So the Yankees have it right because they’re throwing their best pitchers as often as possible… but the Phillies have it wrong because they’re throwing their best pitchers as often as possible?
They don’t pitch the same amount of games. Games 5-7 aren’t guaranteed to be played. You have to win your way into those games, and by sticking Pedro in game 2 (and potentially game 6), you’re making it far more likely that he pitches two games and Hamels only gets one.
Saunders and Weaver didn’t pitch the same amount of games for the Angels. This is the same scenario.
Also, it’s possible Lee will go twice on short rest, throwing 1/4/7. To double a guy up 2/5 or 3/6 would require short rest, and Pedro and Hamels are both dicey on short rest, but 2/6 would be doable, with a rotation looking like
1: Lee
2: Hamels
3: Pedro/Blanton
4: Lee
5: Blanton/Pedro
6: Hamels
7: Lee
Happ in long relief as needed.
If you want to look at it in terms of higher number of expected starts, that’s fine. But the probability of winning the series is still the same regardless of Pedro/Hamels or Hamels/Pedro. VegasWatch did an extensive post on this last year.
doug you beat me to the counterpoint made by dave.
So we are even
You are dead right.
Dave, if hamels pitches games 3 and 7, and pedro throws games 2 and 6, it doesnt matter if the phillies lose in 6. Even if hamels throws game 6, they still presumably lose game 7 with blanton.
A better way of thinking about this is in soccer, there is the penalty shootout. If you take each kick as an independent event and dont treat it as (“Well Jon, the pressure is on him now, and he doesn’t like the pressure!”), it doesnt matter where you kick from 1-5 in the penalty lineup. Either way you still need to make more spot kicks than your opponent. So if you have your best kicker going fifth (which happens alot for some questionable reasons), but your team misses its first 3 and your opponents make the first 3, you lose. But you didnt lose because your best kicker was slated 5th instead of 1st. IF your best kicker was slated first, youd be down 3-1, and chumpy McChumpalot is now kicking 5th and would miss anyway.
The only way there is ANY validity to your argument dave is if you believe in “momentum”, and the advantages of “pitching with a series lead”, and if that is the case you might want to bring in Joe Morgan as a guest-writer.
@UnrepentantGunner:
Yeah. I agree. I think it’s quite short-sighted to just look at higher number of expected starts when it completely, totally, 100% doesn’t matter if they are to start the same number of games over a seven game series.
one other thing to ponder. if the yankees played the phillies 162 times how many would they win? I used the 53% number because thats about 86 wins, which sounds abuot right. So over 162 games winning 86-76 is a big deal, but as always, small sample sizes lead to crazy retroactive explanations. I can’t wait to hear about the winning team’s “heart”, when i cant put the yankees as more than a 57-58% favorite
An average American League team beats an average National League team 57% of the time, so 53% is way too low.
well, thats great, except for, you know, these aren’t average teams. i am doing my best to ignore sample size issues but looking at this crudely, by say market cap, (not a great predictor of wins by any means), you have the average NL payroll a full 9 million lower than the average AL payroll. I think that right there accounts for at least a decent piece of the difference, espeicially given the fate of the pirates padres and marlins which were overwhelmingly composed of players making 400k, and thus a reasonable chance to be replacement.
any justification for the yankees being more than a 53% favorite in the game would have to be based off of payroll differences or other measurable ideas such as wins above replacement. But in reality, we know a decent piece of the reason why the NL comparitively is awful, and that takes the form of $.
With that said, the Phillies have both a healthy Payroll, and some very valuable players, and can’t be considered underdogs just because they share the same league as the Nationals and Pirates.
UnrepentantGunner – I would expect the average AL payroll to be higher, as you have one additional starting player per team. Coming into 2009, there were big money DH’s like Ortiz, Thome, Matsui, Burrell, Vlad, Hafner, etc. On NL teams, they would all be replaced by low money bench players. That probably explains most of the payroll difference. Any additional difference is likely accounted for by the Yankees payroll being so out of line with everyone else’s.
The DH advantage goes away in the NL ballpark, so that should neutralize the payroll advantage in 50% of interleague games.
Ed that is a fair and valid point to some degree, although certainly the bigger NL teams have some of these bench-type players as well (Thome and Sheffield come to mind). I’d have to think these players also matter in inter-league matchups, especially at AL stadiums, but it might also make a smaller but still meaningful difference, as Matsui >> Stairs/Dobbs.
sorry to nitpick, but i believe both Shef and Thome’s salaries were still being covered by their AL counterparty.
Great piece, couldn’t agree more. It’s amazing how managers buy into the “hot hand” in so many cases. Of course when they luck into something (like Mathis’ ALCS performance), they get a ton of praise for eschewing the word of the stat-heads, despite the concession of said people that anything can happen in a small sample.
I like Carson’s writing, and I think he takes too much flak, but Dave Cameron is still the undisputed king of baseball analysis… screw it, sports analysis in general.
Dave, I could not have said it better myself! Excellent piece! The only thing I might add is that I think Pedro is a master at mixing up his pitches (and not mixing them up in order to “out think” the batter) such that his overall true talent is probably better than what a scout would think if they graded each of his pitches separately. If I had to put a number on Pedro’s overall true talent, I would probably say a little above an average pitcher, but unfortunately I too am heavily influenced by his last outing against the Dodgers and I can’t get that out of my mind. My certainly level on that “a little above average” is not very high. If he pitched a thousand innings (at his current true talent level) and it turned out that he was worse than average, I would not be too surprised. Basically I am not sure where Pedro stands, true talent-wise, at this point in his HOF career. Of Hammels, of course, I am very confident of his true talent, which is likely .7 runs or so better than the average starter.
But your basic thesis is 100% correct. Hammels is a very good pitcher and the notion that somehow he is not nearly as good as he was last year, is nonsense. Plus, I think (I am not sure) that a lefty pitcher is better versus the Yankees, especially in Yankee Stadium (there their DH is a lefty and you prefer switch hitter to hit from the right side).
^Good post. Interesting analysis on Pedro…
Cole “Hamels”
(one ‘m’)
The lefty-righty thing is one that shouldn’t go unnoticed. Even if Hamels and Pedro were equal, you’d probably rather have the lefty pitch in Yankee Stadium and play to LF, the larger (relatively speaking) part of the yard. The same is true of Philly, except you’d want to have your righthanded pitcher pitch to avoid the short porch in LF as much as possible. Therefore, Hamels in 2 and 6, Pedro in game 3. That’s if they were equal which, as you point out, isn’t the case at all. This should be a no-brainer.
What’s the deal with not starting J.A. Happ? Worried about the pressure or innings count? His FIP is better than Blanton’s and just a smidge worse than Pedro’s. Couldn’t a case be made for giving him a start, regardless?
I’m not really sure why the Phillies are not starting 3 LHPs against the Yankees, especially with short porch in RF at New Yankee stadium.
The Phillies clearly need Happ in long relief to handle all those Bomber lefties. Think Darren Oliver Game 6, that will be his role. They are a better team using Pedro/Blanton as starters with Happ in the pen than the other way around.
Talk radio in Philly today commented on the recent declining velocity of Hamels’ FB. Because I don’t trust anything that talk radio says without reviewing support to back it up, is anyone here aware of whether Hamels has shown a lack of FB velocity in recent games?
Don’t know what his avg velo was during the previous two playoff series, but after dipping a bit for several starts in late August, early September, it popped back up over his final 3 or 4 reg season starts
http://www.fangraphs.com/fgraphs/4972_P_FA_20091003.png
Hamels’ last 6 outings have been bad. The pitch f/x analysis might tell you something about his control during that time. It is always possible that the manager knows something about the pitcher’s arm health.
Maybe Charlie Manuel feels that Martinez would be best suited to go against Burnett while Hamels is a better matchup against Pettitte. It just seems like a gut move on Manuel’s part, that’s the only thing that makes sense to me.
Ryan,
The order literally has no effect on the Phil’s win probability for the series.
Yeah I know, but sometimes in my own little world filled with irrational, probably non existent confounding variables, I like to think it does sometimes. Kind of like Santa Claus =)