Baseball for the Thinking Fan

Login | Register | Feedback

You are here > Home > Baseball Newsstand > Discussion
Baseball Primer Newsblog
— The Best News Links from the Baseball Newsstand

Thursday, May 02, 2002

Who’s going to make the postseason?

I want to know if the Expos are for real. And I don’t want to wait until late September to find out.

If you look at the straight standings as of today, we’d have:

The Yankees and Twins in a 1-game playoff for the wild card. If the Yankees win, they go to Seattle, and Chicago goes to Boston. If the Twins win, they go to Boston and Chicago goes to Seattle.

Over in the NL, the Expos win the East in a tie-breaker with the Mets (4-3 head-to-head). So the Mets head to Cincinnati. The West would have a 3-game playoff between Arizona, San Francisco and Los Angeles to see who plays Montreal.

If you use David Cornette’s pythagorean projections (linked above), we’d have:

Yankees (104.6 W) at White Sox (97.8 W)
Mariners (96.0 W) at Red Sox (118.5 W).

In the NL:

Expos (98.7 W) at Giants (102.6 W)
Reds (89.8 W) at Mets (102.5 W)

The point of all of this is, when does this matter?

Obviously it matters late in the day on September 29. But leading up to this date, when do the standings start to mean something? The Orioles were 1-0, obviously it didn’t matter. But somewhere along the way it starts to no longer be “early in the season”. With we can now access the major league standings on any given day of the season. It would be a neat study to go through that data and find out things like:

-What are the odds that the team in 1st place with X days remaining in the season ends up in first place. Also, how does this change based on the number of teams in the division.

-A formula that would take X games ahead, with Y teams in the division and Z remaining days in the season and give a % likelyhood of winning the title.

Bill James’ article in the Tigers comment of the 1985 Abstract took on this challenge in a different way, looking at the eventual record of teams, based on their record through a certain number of games. I think things like this would be pretty useful for answering the question of “is this team for real?” I don’t have the time right now to do this study, but I think it’d be pretty interesting to know the answer.

JoeD has the Imperial March Stuck in His Head Posted: May 02, 2002 at 02:12 PM | 11 comment(s) Login to Bookmark

Reader Comments and Retorts

Go to end of page

Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.

   1. Toby Posted: May 02, 2002 at 03:27 PM (#112214)
Can I throw this idea out again? There should ALWAYS be a 1-game playoff for the wild card. The top two second-place teams should battle it out.

That would help alleviate (though admittedly not completely solve) the unbalanced-schedule problem between divisions. Since alleviating that problem would be part of the purpose, the playoff would always involve second-place teams from different divisions, even if the third-place team in a division had a better record than the other two second-place teams.

It would also attach a real incentive to win the division outright.

And it would give us fans a sudden death playoff game every year.
   2. David Geiser Posted: May 02, 2002 at 03:33 PM (#112215)
To answer your question, Joe, no, the Expos are not for real.
   3. Steve Treder Posted: May 02, 2002 at 03:57 PM (#112217)

I agree that some kind of systemic modeling of full-season performance based on early-season results would be interesting to see, and in fact represents something of an underexplored territory within sabermetrics. Why don't we have a Brock2 formula for teams in-season?

But I'm old-fashioned enough to also say that there are just plenty of common sense things that can be brought to bear on the question of "is it still early?"

It's still early when your team hasn't even come close to making one circuit around the league yet, and most teams still haven't played each other.

It's still early when starting pitchers haven't even begun to have to deal with fatigue brought on by mounting innings.

It's still early when individual stats like BA and ERA aren't even close to aligning with season-ending norms.

It's still early when teams haven't had to deal with playing in conditions of searing heat, along with the conditions of bitter cold that they're still dealing with.

And so on. Basically there are reasons why you can't just draw a straight line and extrapolate April performance. Good teams are those that are able to respond to the variety of challenges that the long season dishes out.
   4. . . . . . . Posted: May 02, 2002 at 04:12 PM (#112218)
Way I always understood it, the traditional cutoff is Memorial Day.

Logically it makes sense. Its roughly 1/3 of the way through the season, so a team that went 32-21 by that point will finish the seasons with 92 wins just by playing .500 ball..enough to win plenty of divisions.
   5. WaltDavis Posted: May 02, 2002 at 04:18 PM (#112219)
There are lots of reasons why this isn't accurate, but as a quick and dirty way, we can turn to our friend the binomial distribution. The Expos are 17-10. The chances of a "true" 500 team winning 17 of 27 is 6%. That's low, but it's still well within the 95% confidence interval (and even the 90% one) that we generally accept as "insignificant." (a confidence interval falls on both sides of the expected or null value, so to be outside the 95% CI, the probability has to be 2.5% or less) On the other hand, the Expos have shown themselves to be significantly different from a "true" 400 team.

In fact, even with the best record of 18-9 (2.7% chance), the M's are not quite statistically significantly different from a 500 team.

Quick and dirty. And of course, even if the Expos are currently a 500 (or 600) team, that doesn't mean that injuries, fatigue, etc. won't turn them into a 400 team.
   6. DTS Posted: May 02, 2002 at 04:20 PM (#112220)
Actually, a team that is 32-21 (roughly .600 ball)that plays .500 ball the rest of the way would finish 11 games over .500. Right? Win 86 or 87 games when all is said and done. Makes a bit of a difference, in that that's less likely to win you the division.
   7. WaltDavis Posted: May 02, 2002 at 04:24 PM (#112222)
Whoops, the Red Sox have the best record at 17-7 and are significantly different from a 500 team.

Second whoops, a more accurate quick and dirty test would point out that unless we chose the Expos a priori (aka beforehand), the above test is not accurate. A better idea would be to ask among 30 .500 teams, what are the chances that X # of them would have more than 16 wins at this point in the season. My brain hurts too much to quite figure that out, but the chance that at least one of 30 .500 teams would have 17 wins after 24 games is 29%.
   8. scruff Posted: May 02, 2002 at 06:35 PM (#112226)
David Geiser -- why don't you think they are for real? Other than that they are the Expos? Best offense in the NL 1/6 of the way through the season, by a comfortable margin. An excellent young rotation. 1991 Braves ring a bell? BB are way up here. I think this team has done everything it could possibly do to make me think they are for real. Young teams are prone to taking big steps forward, etc..

Jonah -- I feel your pain. I'm really starting to believe. But I'm also the guy that goes out on a couple of dates and thinks the girl is "The One". I realize I'm setting myself up for a fall, but right now I honestly believe this team is going to win the division. I say enjoy it now, and think as if you are going to the playoffs. Worst case scenario you feel good for a few more weeks.

Walt -- I like the idea of binomeal distribution. Right now at 17-10, the Expos are at a level that is less than 2% for a .450 team. If they take 3 of the next 4 (1 vs. Hou, 3 vs. Ari) they are less than 1% for a .450 team. So it's getting close to the point that we start taking this team seriously as a contender (because a .500 team that gets a little lucky in that division can win it).

The other thing I noticed is this. There is a slightly higher chance that a .500 team will go 92-70 than there is that it will go 17-10. We aren't within 2.5% until we get all the way to 93 wins over the course of a full season. But at some point we have to start taking the team seriously. I agree with dzop that Memorial Day is probably reasonable.

Using your 2.5%, if the Expos win 6-of-8, they'll be 23-12, which is 2.1%.
   9. SM in DC Posted: May 02, 2002 at 06:58 PM (#112229)
You know who just might put Les Senators over the top.. and I hate to say it, 'cause my detest for him has been documented here before -- but Rickey! Free Rickey Henderson!! He's unhappy, and Peter Burgreon needs to have his plug pulled..... Rickey to Canada!!
   10. WaltDavis Posted: May 02, 2002 at 07:28 PM (#112230)

that's one of the reasons why I wrote "There are lots of reasons why this isn't accurate..." A lack of independence (most especially that a win by one team is necessarily a loss by another team) and unequal probabilities across events (injuries, quality of competition, etc.) are the main reasons why it doesn't really work.

According to the binomial distribution, the 95% confidence interval for a .500 team over 162 games is about +/- 13 games so anywhere from 68 to 94 wins. And with 30 teams and the .05 cut off, we're already accepting that 1.5 "true .500" teams every year are expected to be outside that range.

However, I did once run a simulation that at least controlled for the fact that, for example, the Yankees winning percentage and the Rays winning percentage aren't independent and the +/- on a league (of 14 or 16 teams I can't remember) of all "true .500" teams was about 9 wins/losses.

There's a lot of randomness in baseball. Or, for those who prefer not to believe in randomness, there's a lot that's not statistically distinguishable from randomness.
   11. . . . . . . Posted: May 02, 2002 at 10:13 PM (#112231)
boy, i cant believe I flubbed the math on that 21-21 thing....

Football always scares me with its statistical insignificane, so to speak. My god, how do teams evaluate based on 16 games? Its like lotto.

You must be Registered and Logged In to post comments.



<< Back to main

BBTF Partner

Dynasty League Baseball

Support BBTF


Thanks to
Eugene Freedman
for his generous support.

You must be logged in to view your Bookmarks.

Hot Topics

Newsblog2023 NBA Regular Season Thread
(297 - 9:47am, Feb 03)
Last: Fancy Pants Handle struck out swinging

NewsblogHow to Watch the Caribbean Series
(4 - 9:15am, Feb 03)
Last: Jose is an Absurd Sultan

NewsblogEx-girlfriend alleges Mets outfielder assaulted her in Syracuse; warrant, lawsuit target player
(8 - 8:35am, Feb 03)
Last: Lassus

NewsblogMLB Network Exits YouTube TV Ahead of Spring Training After Contract Dispute
(51 - 8:34am, Feb 03)
Last: KronicFatigue

NewsblogOT - 2022 NFL thread Part II
(321 - 7:55am, Feb 03)
Last: AuntBea odeurs de parfum de distance sociale

NewsblogJohn Adams, Who Banged His Drum in the Cleveland Bleachers, Has Died
(14 - 1:12am, Feb 03)
Last: Joyful Calculus Instructor

Hall of MeritRanking Right Fielders in the Hall of Merit - Discussion thread
(42 - 10:46pm, Feb 02)
Last: Howie Menckel

NewsblogMLB expansion: Nashville group led by Dave Stewart makes a pitch for Music City [$]
(25 - 10:40pm, Feb 02)
Last: John Northey

NewsblogOT Soccer Thread - Hi Ho Hi Ho it’s Back to Club Football We Go
(355 - 10:09pm, Feb 02)
Last: Jose is an Absurd Sultan

Hall of MeritRanking Left Fielders in the Hall of Merit - Discussion thread
(94 - 6:37pm, Feb 02)
Last: Howie Menckel

NewsblogOrioles to decline 5-year Camden Yards lease extension, seek to secure long-term agreement
(3 - 3:58pm, Feb 02)
Last: jingoist

Hall of MeritReranking Left Fielders Ballot
(20 - 3:38pm, Feb 02)
Last: Tiboreau

Sox TherapyUpdates
(96 - 2:28pm, Feb 02)
Last: Darren

NewsblogOT: Wrestling Thread November 2014
(2641 - 5:42am, Feb 02)
Last: aberg

NewsblogBaseball Hall of Fame tracker 2023
(740 - 9:25pm, Feb 01)
Last: DL from MN

Page rendered in 0.4429 seconds
44 querie(s) executed