Playing a Different Game: 2009 Strength of Schedule

by Sky Kalkman on Jun 1, 2009 2:00 PM EDT Comment 7 comments

I've been looking around the web for measures of MLB strength of schedule and have found three results so far. One, Andy Dolphin's site, hasn't started updating for 2009. A second, ESPN.com, doesn't make their methodology obvious at all. The third, BPro, also doesn't explain its methodology, but at least it's easy to figure out how to use the adjustments.

By comparing BPro's second- and third-order wins, you can determine how much credit they're giving to each team for strength of schedule in third-order wins. As a reminder, third-order wins are based on adjusting equivalent runs for SoS and then using Pythag (-pat or -port I'm not sure). Not perfect, but pretty good. Here's each team sorted with the most difficult schedules at the top. Again, because the methodology is a black box, mentally apply large error bars and some skepticism.

Team	Bpro SoS
Indians	1.8
Orioles	1.4
Athletics	1.0
Giants	0.9
Red Sox	0.9
Marlins	0.9
White Sox	0.7
Padres	0.7
Astros	0.5
Nationals	0.3
Rockies	0.3
Twins	0.3
Yankees	0.1
Angels	0.0
Pirates	-0.1
Royals	-0.2
Brewers	-0.2
Phillies	-0.2
Mariners	-0.3
Tigers	-0.3
Braves	-0.4
Cubs	-0.4
Diamondbacks	-0.5
Cardinals	-0.5
Reds	-0.6
Mets	-0.7
Blue Jays	-0.9
Rays	-1.4
Dodgers	-1.5
Rangers	-1.5

As a recent FanShot mentioned, the Indians won't play another AL East team until mid-August and have only four more games left against non-Orioles AL East teams.

To have something to compare BPro's numbers to, I attempted to turn ESPN's strength of schedule data into something meaningful. By pro-rating BPro's numbers to 162 games, I could create an opponents' winning percentagem which is the format of ESPN's numbers (I think). ESPN's distribution of winning percentages were about half as wide as BPro's, so I doubled them to make a comparison more meaningful. The table below is initially sorted with which teams' schedules ESPN thinks were harder as compared to BPro at the top, but you can click on the column headers to re-sort by the ESPN SoS adjustments themselves, too.

Team	Bpro SoS	ESPN SoS	Diff
Dodgers	-1.5	0.0	1.5
Cubs	-0.4	0.8	1.2
Brewers	-0.2	0.8	1.0
Blue_Jays	-0.9	-0.1	0.8
Angels	0.0	0.8	0.8
Rangers	-1.5	-0.8	0.7
Padres	0.7	1.3	0.6
Mets	-0.7	-0.1	0.6
Astros	0.5	1.1	0.6
Reds	-0.6	-0.1	0.5
Tigers	-0.3	0.2	0.5
Yankees	0.1	0.6	0.5
Twins	0.3	0.7	0.4
Rockies	0.3	0.7	0.4
Cardinals	-0.5	-0.3	0.2
Royals	-0.2	-0.1	0.1
Rays	-1.4	-1.3	0.1
Diamondbacks	-0.5	-0.5	0.0
Mariners	-0.3	-0.4	-0.1
Giants	0.9	0.8	-0.1
Orioles	1.4	1.1	-0.3
Pirates	-0.1	-0.6	-0.5
Indians	1.8	1.2	-0.6
White_Sox	0.7	-0.2	-0.9
Athletics	1.0	0.0	-1.0
Red_Sox	0.9	-0.3	-1.2
Phillies	-0.2	-1.5	-1.3
Braves	-0.4	-2.0	-1.6
Marlins	0.9	-0.8	-1.7
Nationals	0.3	-1.4	-1.7

One thing I notice about the ESPN numbers relative to the BPro numbers is that ESPN doesn't favor the bad teams or knock down the good teams as much as BPro does. Perhaps that suggests ESPN better accounts for the fact that bad teams make their opponents look better and good teams make their opponenets look worse.

As for what the two systems agree on:

The Rays and Rangers have had two of the easiest schedules in the majors.
The Indians and Orioles have had two of the hardest schedules in the majors.

One thing I think is missed here (although maybe not) is the number of home and away games played by each team. At one point, the Rockies had played only 20 home games while another team had played 30. That certainly falls under strength of schedule, even though it's not strength of opponents.

Anything else we can take away from these black box systems? Any ideas for creating our own SoS adjustments?

0 recs | Comment 7 comments | Share

Story-email Email Printer Print

Comments

Fanpost

See my fanpost from today as well.

Can some kind of metric be created to measure the difference between each home game and away game/per game as far as wins and losses go? There has to be a way and I am by no means a statistician. It all comes out even in the end, but maybe some sort of number can be applied concerning the difference in away/home games and adjust w/l record accordingly. Would using historical data and averaging out the numbers be helpful?

by backtocali on Jun 1, 2009 3:01 PM EDT reply actions 0 recs

The simplest way to do it would be to realize that home team win 54% of games (I think).

So a home game implies .04 more wins than an average game and an away game .04 wins less than an average game. A team with 30 home games and 20 away games would be expected to win .4 more games than if they played 25 and 25, all else being equal.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Jun 1, 2009 11:22 PM EDT up reply actions 0 recs

What Makes a Baseball Schedule Stronger

Some Hypotheses:

1) General difficulty of opponents. In a shorter stretch of the season, this would lead to greater discrepencies – one team might be playing the Nationals more than someone else in their division so far. Over the course of the season, though, this has limited application. Everyone plays the same teams except for interleague play, where everyone plays a different group of teams from the other league, even within the same division. This seems to be what they already want to do.

2) Roster fluctuations of opponents. Superstars hit the DL (A-Rod, Sizemore) or get suspended (Manny). Good players get called up (Wieters), bad players get released (Mota, we can only hope), and this flux of rosters changes things. Not to mention after the trade deadline. Teams facing the Dodgers on July 20, 2008 had an advantage over those facing them on August 2, 2008. This seems like it would make a difference over the course of the season. Maybe some sort of formula to determine expected runs for a particular lineup.

3) Streakiness. April 2004 was a good time to face Derek Jeter. May 2009 was a bad time to face Juan Pierre. Was the whole team having a generally good streak? Or a terribly bad streak? In the short term this makes a big difference, but in the long term one would thing it might matter more.

4) Lucky off-days. It’s possible for a division rival to avoid a team’s ace while another team faces him 3 or 4 times. The Dodgers, for example, managed to miss Johan Santana when facing the Mets. This could apply to the team as well – do you follow a Thursday night series ending game in Coors field with a day game at Wrigley? Did you just finish a 20 inning marathon game? This makes a difference.

I’d look it up, sure, but frankly it took me too long just to write that. A lot of this is really luck, but I think that interaction effects with the schedule can make it a strength of schedule issue.

by StolenMonkey86 on Jun 1, 2009 8:27 PM EDT reply actions 0 recs

I agree with #1...

I agree with number two, although that’s difficult to adjust for. I totally disagree with number three. I agree with number four, which is the pitcher equivalent of number two, and probably easier to handle. Actually, I agree with the first half of four, not the second half.

Beyond the Boxscore // Calling BJ Upton lazy is lazy.

by Sky Kalkman on Jun 1, 2009 11:24 PM EDT up reply actions 0 recs

How to measure these

1) Well, I think this is sort of the one that everyone’s focusing on. And now prepositions are words that I end sentences with.

2) As a result of a roster move or a trade, who comes into replace them on the roster and in the lineup/rotation? What does this due to the expected skill of a team as measured in 1? Jake Peavy is traded – who replaces him in the rotation?

3) This is probably too much due to randomness to even be counted, based on the examples I gave. However, there are some trends to look at, such as whether individual players perform better in the first or second halves (Rafael Furcal is historically better in the 2nd half, while Orlando Hudson historically has better first halves). Does this even out, or are some teams set up to perform better in the stretch while others are destined to play like the Mets have the last two Septembers? Incidentally, Jose Reyes has a .248/.307/.378 career line in September, .087 below his career OPS.

4) This can be a matter of how the rotation is managed for off-days. Sometimes in an off-day, the fifth starter (or just the crappiest starter) will get his turn in the rotation skipped, and sometimes everyone just gets an extra day of rest. I would think that this would be at least somewhat predictable, and therefore it’s possible that some teams could be set up to face the Jake Peavy, Tim Lincecum and Chad Billingsley while a division rival faces Jeff Weaver, Jonathan Sanchez, and Kevin Correia. In particular, the unbalanced schedule of off days could set up a team using its #1 against another team’s #5, etc. The latter half of this hypothesis is too much of a random effect to measure, so I’ll drop it

by StolenMonkey86 on Jun 2, 2009 5:07 PM EDT up reply actions 0 recs

“Everyone plays the same teams except for interleague play”

This is only true if you are looking at each division separately. If we are looking to see who the best team is in a division, then SoS isn’t that important. But if we are looking to see who the overall “Best” teams are, then it makes a difference. The Royals are going to have a much easier schedule than the Blue Jays, but a harder schedule than the Nationals.

by lookatthosetwins on Jun 1, 2009 10:42 PM EDT reply actions 0 recs