If you followed last night’s in-game roundtable, you got the visceral reaction to Jim Tracy‘s decision to allow Huston Street to face Ryan Howard in the ninth inning with the tying runs on base and two outs. (You got something similar if you follow my Twitter account, @joe_sheehan.) In the interest of analysis, let’s let the data do the talking this morning.

Ryan Howard is the best left-handed batter in baseball against right-handed pitching. There is no one better than him, no one. He hits for average (.319 in 2009, with a .339 BABIP) and power (.372 ISO, with an extra-base hit every six at-bats) and posts an acceptable walk rate (about one UIBB in ten PA) for a very good OBP of .395. The specifics have bounced around a bit from year to year, but his 2009 numbers against northpaws are a good match for his career numbers: .307/.409/.661, .338 BABIP, about a 10 percent UIBB rate. The only hitter you might prefer at the plate with a right-handed pitcher on the mound in Albert Pujols, although the 2009 version of Joe Mauer is in the discussion as well. Howard is simply a devastating force against righties, one of the very best in memory at hitting them.

When a left-hander is pitching, Howard is a bad hitter. In 2009, he batted .207 with a .299 BABIP, striking out in nearly a third of his plate appearances. He had an ISO of just .149 and inferior walk and extra-base hit rates. It was the worst full-season performance of his career against left-handers, but not that out of line with his 2007 and 2008 lines. (In 2006, Howard posted a .368 BABIP against lefties, which boosted his overall line against them to .279/.364/.558. That stands out as a fluke.) For the three-year period covering 763 plate appearances, Howard has batted .219/.308/.379. Whereas his comps are singular-Albert Pujols-when a righty is on the mound, it’s a different story when a lefty pitches. In terms of his overall effectiveness against lefties, he hits a bit like Pedro Feliz and Cristian Guzman did this year, or Jack Wilson before his trade to the AL.

At around 7:45 p.m. Mountain Time last night, Jim Tracy needed one out to get his team a second win in the Division Series, to push them to a deciding fifth game. He had Albert Pujols, more or less, at the plate, and he had the option to turn Pujols into Pedro Feliz just by walking to the mound and tapping his left arm. With one move, he could have dramatically increased his team’s chance of getting an out, winning the game, flying back to Philadelphia on the heels of a dramatic comeback victory.

And as we know, he didn’t do it. Jim Tracy chose to face a batter with an 1100 OPS instead of one with a 700 OPS. That, more than anything else that happened, is why the Rockies lost.

William Burke raised the point in the roundtable that Street had a large reverse-OPS split in 2009. Let’s examine that. Bil was right, factually: Street was very effective against left-handed batters in 2009, allowing a .167/.227/.265 line to them in 111 plate appearances. Right-handers hit .217/.244/.375 in 129 PAs. But while the OPS gap is large, it is entirely due to the difference in batting average on balls in play and what happened on fly balls, neither of which is indicative of Street’s actual skills. Street allowed a .300 BABIP and five home runs to righties, just .195 and two bombs to lefties. Dan Malkiel ran the numbers and found that the underlying rates of batted balls allowed don’t support these splits-they’re a fluke. Moreover, Street’s non-contact data clearly shows him to have been more effective against right-handed batters: he struck out 36 percent of the right-handed batters he faced, and walked just three righties unintentionally all year long. Compare that to 21 percent of lefties struck out with six unintentional walks allowed in fewer PA. Street is clearly much more effective against right-handed batters.

This is supported by his career data. Street has allowed nearly identical batting averages on balls in play in his career to hitters on both sides of the plate. However, he has struck out a higher percentage of righties and walked them at about half the rate he’s walked lefties. He’s also allowed much more power-a .151 ISO and 16 homers-to lefties than to righties (.082 and nine). There is no way to read Street’s track record and conclude that he’s more effective against left-handed batters, and even a cursory look at his pitching style would support that as well.

In the interest of being thorough, let’s note that since he returned to the majors as a full-time reliever in 2006, Joe Beimel has held left-handed batters to a .301 OBP with an extra-base hit in about every 14 at-bats.

Given all of that information, how much do you have to believe in the closer myth to have allowed Street to face Howard with the season on the line? You have to believe, basically, that the skill involved in closing games, getting the 27th out, is so large that it makes up the difference between the very best hitter in the game, one of the best in history, and a mediocre shortstop. Even that may be generous, as it doesn’t consider that Howard’s line against lefties included a lot of ABs against guys like Tom Glavine and Johan Santana, fastball/changeup pitchers who come over the top and don’t specialize in getting lefties out. Beimel is a different animal, as his stats show.

A manager’s job is to put his team in position to win, and he does that by putting his players in the best possible position to succeed. Tracy didn’t do that, and in fact, he aggressively made the situation worse for Street and the Rockies. You can argue that Street “failed” in some sense, but the guy was asked to do something-get Ryan Howard out using his right arm-that hundreds of pitchers have been failing at for half a decade. He didn’t have to be in that situation. I’ve probably used the “it’s called playing the percentages” line a couple of hundred times, but last night is the first time I can honestly say that a team would have been better off with Montgomery Burns in the dugout than with the manager they had. Tracy made just about the maximum possible error you can make, calling to mind a Buffalo Wild Wings commercial more than a guy who’s probably going to get a nice piece of hardware this winter.

It’s funny‚Ķbecause I missed Beimel’s cameo in Game Three, I wrote a couple hundred words in Monday’s piece that I had to immediately retract, words that castigated Tracy for not using Beimel in the ninth inning. As it turns out, the entire paragraph could have been written after yesterday’s game, nearly verbatim, and been completely correct.

There’s one absolute when you face the Phillies-don’t let Ryan Howard face a right-handed pitcher in any kind of game-critical situation. Tracy’s in-game tactics flew in the face of that, and so he doesn’t get to fly to Philadelphia. The team, the organization and the fans all deserved better.

Tracy’s failure overshadowed another entertaining game that featured strong starts by both Cliff Lee and Ubaldo Jimenez, another set of quality at-bats such as we saw Sunday, more Charlie Manuel bullpen juggling, and at least one play I’ve never seen before. Setting aside Howard for the moment, let’s note that the Phillies’ rally started with another hit by Jimmy Rollins and was extended by a walk to Chase Utley. They had to get the inning to Howard for Tracy to fail, and they did just that.

I’m getting a lot of enjoyment watching Manuel, who in my eyes made Tracy look even worse right after the ninth-inning rally. With the Rockies due to send three lefties and a switch-hitter to the plate, Manuel passed on bringing in his closer, Brad Lidge-a move that would have been automatic in all but a tiny percentage of games in the past decade-and started the ninth with lefty specialist Scott Eyre, who basically has never closed. Eyre wasn’t effective, allowing a pair of singles before giving way to Lidge, who retired Troy Tulowitzki to end the series, but he was the best man for the job regardless of his career saves total. It was a beautiful moment; Manuel watched his team get three runs in part because his counterpart made a mistake, and then he himself showed what the right play would have been minutes later.

Because of Lidge’s season-long ineffectiveness, Manuel has been forced to go away from a closer-centric bullpen. Lidge has closed two games in a row, but the first was largely due to circumstance-Scott Eyre’s injury in the seventh was the trigger event-and the second he was hardly used as a capital-C closer, but rather in a matchup setting. I don’t think for a second that Brad Lidge is a reliable high-leverage option; I do think that he can be used tactically against right-handers because his slider is mostly still an effective weapon against them. If Manuel keeps him in a protected role, the Phillies are going to be much better off than if he decides it’s 2008 again. All we really know about Brad Lidge off of this series is that Troy Tulowitzki is lost against him. (Tulowitzki took a hanging slider on 1-1 that he’ll probably be seeing in his dreams until spring training.)

Carlos Gonzalez had a ridiculous series, which doesn’t mean he’s certain to be a star, but he had so many good at-bats against left-handed pitchers that you have to like his chance to become an everyday player. He has every tool-he nearly gunned down Jimmy Rollins at third base in the third inning, in addition to all the bat work-and he showed skills in 2009 that haven’t always been there. Before this series, I would have taken Dexter Fowler every day and twice on Sunday; now, I’m less sure, in part because the Rockies have put Gonzalez in center and Fowler in left, clearly ranking them that way defensively. A team with the contact staff the Rockies have needs a good defensive outfield, and the Rockies have two-thirds of one lined up for the next five years.

Gonzalez had another big game, but Fowler had maybe the best moment, leaping over the back of Chase Utley during the Rockies’ eighth-inning rally to avoid both a tag and an interference call. The move may have startled Utley into a bad throw that allowed Fowler to reach second, a key play in the Rockies’ mooted comeback. I had never seen a baserunner make that kind of play before, and I probably wouldn’t recommend it again as the risk of injuries to both players as well as the likelihood of an interference call are high, but it was the highlight of the night.

Before all the late-inning machinations, Cliff Lee was putting together another virtuoso performance, his second of the series. He wasn’t able to complete the game, and because it all went bad when he left, we got a glimpse of just how big the gap is for the Phillies between the guys who start and the guys who come in later. They need good starts, deep starts, to win. They survived two bullpen failures in this series because the Rockies did a mostly lousy job with runners on base, because they got a critical umpire mistake in Game Three, and because Jim Tracy was awful. They can’t count on any of those things in the NLCS, where they’ll face a better offensive team with a much better manager and see four starts by lefties in the first six games.

If baseball games were six innings long, I’d take the Phillies against any team in baseball except maybe the Yankees. At nine innings, their chances just aren’t the same.