clock menu more-arrow no yes

Filed under:

AZ 2, Pirates 3 - Yanked, Spanked and Planked

Record: 18-13. Change on last season: +5

Well, you may have thought Thursday's game was bad, but last night? Ugh. In every aspect but the actual number of runs scored, we were thoroughly outplayed. Given we were outhit 14-2, as William K mentioned in the comments (and I can't blame the rest of you for staying clear!), it was utterly amazing that we had the winning run at the plate in the ninth innings with nobody out.

We pushed one across without a hit, thanks to two walks and two sacrifice flies, but any last, lingering hope was extinguished as Quinton McCracken came to the plate and, inevitably, grounded out. 0-for-4 today, taking his average for the season down to an ever-more feeble .192. Is Bob Melvin the only person in the world who can not see how inadequate this is?

I was half right in my pregame prediction: Fogg didn't suck, but Ortiz definitely did. For "inadequate" is also the applicable word for his performance: he only lasted 4.1 innings, allowed ten hits and four walks, as well as all three Pirates runs. He was saved from a far worse line by Pittsburgh's total inability to get men home.

The bullpen had a decent outing, allowing four hits and no walks over 4.2 innings - all four hits came to Cormier during his two innings stint, but his scoreless streak remains intact, and is now up to 14 innings. Given batters are hitting .294 off him, this is likely more by luck than good judgement.

Offensively, Tracy had the only RBIs, with a homer and a sac fly in the ninth; the other hit went to Counsell. That was it. Nada. Zip. We'll want to get the taste of this one out of our mouths sooner rather than later, and the good news is, today's game is already underway - however, we're now trying to salvage a split of the series, rather than the hoped for victory, or better yet, a sweep.

But before I get into that, over at Beyond the Box Score, Marc Normandin posted the Pythagenport standings for the NL West yesterday - and it doesn't look good for Arizona, with a record for the season overall still predicted at below .500.

Pythagenport is a slightly modified version of a formula invented by Bill James, to predict a team's record based on the number of runs scored and conceded. See here for details, though I recommended having a caffeinated beverage close to hand, unless you're a math scholar. History has shown it's not bad over the course of the season.

However, especially at this early stage, I have some serious qualms about it. Elsewhere in the NL, for example, Atlanta are predicted to have 114 wins...and Florida, 137? Ludicrously wrong doesn't even come into it. Nor do I believe the Rockies will win only 41 games.

So if Florida almost certainly won't come within 30 games of the expected total, why should I take a prediction of 80-82 for us as any better than casting pig entrails? Taking the same error spread into account, this tells us we will win somewhere between 50 and 110 games. Very Useful.

Part of the problem is this: Arizona has scored less runs than it has scored, therefore the formula requires that they will win less games than they lose. However, almost 12% of those runs conceded came in just one of our thirty games, the opening 16-6 blowout against the Cubs. Take that wild outlier out of the equation, and our predicted record now becomes 86-76. This demonstrates another problem with the formula at this point of the season: one freak result can have a far bigger impact than it should.

Another issue would be that it seems to completely ignore the already determined results of the first 30 games, rather than taking them into account. Okay, I can believe that based on performance so far, we might be pretty close to .500 for the rest of the season. But at the time of those stats, we were actually at 18-12; give us 66 wins in the remaining 132 games, and we get 84-76 record. If we ignore the outlier, it predicts us to win seventy more, and put us at 88-74. That's certainly within striking distance, given 93 was enough to win the NL West last year.

Marc is, however, savvy enough to be aware of the limitations of run differential and its flaws, which is not the case for all commentators. Witness the horror which is Fox Sports' Dayn Perry. Now, I know expecting credible commentary from anyone about anything on Fox is a bit of a stretch, but the page calls this guy an "expert". When you make that claim, you're setting yourself up.

The not-so-great Dayn is obsessed with run differential. A team gets outscored = they suck. His most recent chart manages to squeeze it into almost every box. As a result, naturally, he hates the Diamondbacks, and hasn't bothered to rank them above 14th this year despite, oh, the most home wins in the majors. He put the Giants - the Giants! - tenth, and still ranks the Dodgers fifth, even after we went into LA and swept them.

Elsewhere on Beyond the Box Score, Dan Scotto offers up his Power 30, which is kinda like the BCS for baseball teams - it combines various lists, third-order winning percentage, ESPN's RPI figure and Dan's own stat, which seems to compare a team's OPS against their opponents. My qualms here are two-fold: firstly, any system which includes idiots like Dayn Perry is already suspect, and secondly, in Scotto's rankings, the Diamondbacks come... twenty-ninth.

Yes, twenty-ninth. He doesn't say who's below us, and admits this is off: "My system's obviously not perfect if it rates the Rockies and everyone ahead of the Diamondbacks, but if the Royals are outproducing them, isn't something wrong? This can't last." Yes, if the facts conflict with your theory, the facts must be wrong. Never mind the possibility that the Diamondbacks' stats are badly skewed by a couple of blowout losses, rendering them useless as a predictive tool.

Have the Diamondbacks over-achieved? Probably. Will they continue to do so? Possibly not. But if they do, it will be their performance in these future games that decides it, not what they did in the first 30. Those are done, over with, and claiming that a single stat or two drawn from them is any kind of reliable guide to future performance is unproven at best.

The problem with all such schemes is that they ignore something very important. At the end of the season, it isn't runs scored that decides who'll be in the playoffs, and it certainly isn't OPS. It's wins, and whether by one run or ten, they all count the same. Any prediction that chooses not to take victories - especially those already obtained - into account, is likely doomed to fail.