Baseball Notes

I work at a place that often gets involved in the political arena, specifically policy recommendation and research, so times like these are often quite entertaining and quite busy. My past couple of weeks have been spent working on plans for various presidential candidates we have invited to come and also hosting an event for Senator John McCain ahead of the Texas Primary. While I am not using all of that as an excuse for the delay in writing, I am using it as segue into what I will discuss today.

You see, when you host an event for a presidential candidate, there are always questions from the guests or from the audience that they would like answered. Inevitably, the two questions always raised are “would you have done X differently if you were president at that time,” and “if you are president, what will happen when you are faced with X problem?”

The candidates are a little more comfortable with the first question because hindsight is always 20/20 and they can come up with a solution that most people would approve and state how much better their solution is than the one that was made. Conversely, they get a little bit more uneasy when it comes to the second question, there are no decisions that have already been made about the hypothetical problem, thus nothing to base their answer on. And who knows, maybe they will be faced with this same problem in office, do they stick with their answer even though it may not be the best one, or decide differently, and risk looking like a liar or a flip-flopper.

I think the same holds true for baseball. It is a bit easier to look back and plug in a different solution/player/strategy than to predict the course of action for a hypothetical game situation or how a season will play out.

And thus is the nature of projections - a lot is usually right and a lot is usually wrong. However, it’s much easier to look back, take numbers we know are facts, and plug in a few new variables to make educated guesses than it is to base future numbers on unknowns.

And with that we turn our attention to the two new aces of the National League: Johan Santana and Dan Haren of the Mets and Diamondbacks, respectively. Fortunately, for the purposes of this research, both of their new teams were involved in tight races towards the end of the season, with the D’backs turning out a lot better than the now-famous collapse by the Mets the last 17 games of 2007.

While we can’t know for sure how these pitchers will perform in 2008, can we at least try to plug them onto their teams last year and see what kind of difference they would have made? Would it have caused Arizona to miss the playoffs? Caused the Mets to make the playoffs? And what is the best way to find out?

Well, also fortunate for us, we know exactly who these two new pitchers will be replacing on their new teams. Haren will replace Livan Hernandez in the rotation (who left for the Twins), while Santana will replace Tom Glavine (now with the Braves). Otherwise, the rotations seem to be the same.

With a little tweaking, and some playing with the numbers, adjusting them from league to league, I think we can tell how Haren and Santana might have affected their new teams had they been pitching instead of Hernandez and Glavine. Comparatively, Haren had 34 starts to Hernandez’s 33, and Santana also started 33 to Glavine’s 34 - so we almost come out event there already.

Here is what I think we should do:

First, we will remove one perfectly average game from Haren’s line and add one perfectly average game to Santana’s so that they will also each reflect 33 or 34 starts (I want to leave in the best and worst games because those are what make a pitcher’s season and define his consistency. See Ron Shandler’s PQS scores for more on that topic).

Second, we will subtract all of the runs allowed by Hernandez and Glavine for their teams last year from the team’s runs allowed total. We will work with both earned and unearned runs here so that the defensive aspect stays constant - it is something pitchers can not control.

Third, we have to add back in to the teams’ runs allowed totals the number of runs allowed by Santana and Haren last year. This is where it gets a bit tricky and where we have to adjust for context. In 2007, the average ERA in the AL was 4.50. In the NL, it was 4.43. So, the AL was about 2% tougher for pitchers than the NL. Keeping unearned numbers the same, we can adjust Santana’s and Haren’s earned run totals by that 2% to get a sensible estimate of what each pitcher would have done in the NL.

We will then check each team’s actual 2007 won-loss record compared to their expected won-loss record using runs scored vs. runs allowed and the Pythagenpat formula: X = ((rs+ra)/g)^.285 for the exponent and then rs^X/rs^X + ra^X for winning percentage. It has been documented that Clay Davenport, who modified the original Pythagorean Theory for win/loss by Bill James believes this Pythagenpat method is an even further improvement, so we will use that one. We will see how many wins better or worse the two teams were in 2007.

Then using the new runs allowed totals and adding them back into their new teams’ 2007 numbers, we can plug these in for runs allowed, adjust for the number of games better or worse they were above the expected outcome, and see where each team would have ended their 2007 regular season. Would the Mets have held off the Phillies? Would the D’Backs have won the division outright? Won the Wild Card? Missed the playoffs?

Here’s the math, starting with Santana:

2007 Mets - 88-74 record - 804 runs scored, 750 runs allowed for Pythagenpat record of 86-76 - two games better than projected
Glavine accounts for 102 runs - subtract from 750 to get 648
Santana accounts for 88 real runs in 2007, 81 earned
Add three earned runs to Santana’s total (an average start for Johan) to make him equal to 34 starts
Santana now has 91 total runs, 84 of them earned
Take 2% away from 84, leaving him with 82 earned runs, 89 total runs
Add 89 back into the 648 left for runs allowed for 737
Pythagenpat forumla:
X = ((804+737)/162)^.285, X = 1.90
W% = 804^1.90/804^1.90 + 737^1.90, W% = .541
New Pythagenpat record: 88-74
New actual record, 2 games better: 90-72

So now the Mets hang on and beat the Phillies (89-73) by one game to represent the NL East in the playoffs. And the 17-game collapse is all but forgotten. Until they get swept by the D’Backs.

And now for Haren:

2007 Diamondbacks - 90-72 record - 712 runs scored, 732 runs allowed for Pythagenpat record of 79-83 - 11 games better than projected
Hernandez accounts for 116 runs - subtract from 732 to get 616
Haren accounts for 91 real runs in 2007, 76 earned
Subtract three earned runs to Haren’s total (an average start for Haren) to make him equal to 34 starts
Haren now has 88 total runs, 73 of them earned
Take 2% away from 73, leaving him with 72 earned runs, 87 total runs
Add 87 back into the 616 left for runs allowed for 703
Pythagenpat formula:
X = ((712+703)/162)^.285, X = 1.855
W% = 712^1.855/712^1.855 + 703^1.855, W% = .506
New Pythagenpat record: 82-80
New actual record, 11 games better: 93-69

The D’Backs had the best record in the NL to begin with, edging out the Rockies for the WC and beating Philadelphia by one game, so it might not look like it would have affected Arizona’s season too much, much less their sweep of the Cubs and then being swept by the Rockies in the NLCS. But, Hernandez did start game 3 of the NLCS, losing it 4-1. Who knows if Haren had started that game what would have happened (especially since Arizona only scored once). But a 2-1 deficit at that stage would have been much less daunting than down 3-0 with another to play in Coors.

So while this is not ground-breaking stuff by any means, don’t be surprised when these guys make a significant difference on their clubs this year, especially if races end up being close like in 2007. It’s impossible to know for sure what will happen this time, but it just goes to show that one guy could make a difference between the playoffs and going home.

If you catch any errors in my math, please let me know.