A warning on using Baseball Prospectus' Postseason Odds

[Post-Oak]

As usual, BP is using a pseudo-scientific approach which will cost gamblers money if they blindly follow these "projections".

They run a Monte Carlo simulation of the rest of the season (playing it out 1 million times), but they count ties as 0.5 of a championship for each team.

This is why their projections on the Tigers have been way too pessimistic for a while now. In actual fact, if the Tigers and Twins finish with the same record, the Tigers will be named AL Central Champions.

When both teams are making the playoffs anyway, the heads up record is the tie-breaker. With only a handful of games left, this is an important consideration. To ignore it is an absolute joke.

This also has an effect on the NL West race. For example, if the Padres and Dodgers finished with an identical record, and that record is good enough for the Wild Card (in other words the Phillies have a worse record), then SD will be the NL West Champs.

Any Monte Carlo simulation which does not take this into account is an absolute joke.

I keep seeing sports columnists, as well as posters on this board, refer to BP's flawed projections, so I thought I should point this out.

Edited to add: And of course they are not simulating the extra game which would have to be played when teams tie for
1. the Wild Card
2. the Division, where both teams are not already assured of making the playoffs

[MCS]

Is there any way to see how often a tie actually occurs?

[Post-Oak]

[ QUOTE ]
Is there any way to see how often a tie actually occurs?

[/ QUOTE ]

Well CoolStandings dot com has a better Monte Carlo simulation. They also play the season out a million times, but they have at least some of the proper tie breakers in place. Their simulations definitely take into account the tie breaker case for Detroit and Minnesota (Detroit wins the division if they tie).

I'm not sure if they take into account what would actually happen if Philly, LA and SD all finished with the same record. Off the top of my head, I believe that SD would play at LAD, and the loser would be out of the playoffs. I could definitely be wrong, as I haven't really looked into it (although I know LAD won a coin flip so I know SD would be the visitor).

Of course, with so few games left, the best type of simulation would include the starting pitchers for all remaining games.

[bucrogers]

From the Phillies team section on MLB:

***

As a point of clarification, if three clubs finish the season with the same winning percentage and one team will be a division winner and another will be the Wild Card, the games will be played as follows:

• The two teams tied for the division lead will play the one-game tie-breaker, with the winner being declared the division champion.

• The losing team will then play the club from the other division for the Wild Card.

***

(back to me) - which is the opposite of what Post-Oak thought. Although Post-Oak I SWEAR when the wild card first came out in 95, and AL looked like Ana/Sea/NYY could all end up in a 3 way tie, they said the format then would be the way you said - NYY automatically the wild card, Ana/Sea one game playoff, loser out. I'm a Red Sox fan and I can remember be livid about it. Then I think it was last year when the AL looked liked it could have another 3 way tie (Sox/Yanks/Ana?) they came up with the current plan - which while I think its better, I swear was a flip flop to give the Yankees the best chance to make the playoffs.

I've searched the web everywhere though and can not find confirmation about how I remember the original 3 way tie breaker rule though.

Buc

[fnurt]

[ QUOTE ]
(back to me) - which is the opposite of what Post-Oak thought. Although Post-Oak I SWEAR when the wild card first came out in 95, and AL looked like Ana/Sea/NYY could all end up in a 3 way tie, they said the format then would be the way you said - NYY automatically the wild card, Ana/Sea one game playoff, loser out. I'm a Red Sox fan and I can remember be livid about it. Then I think it was last year when the AL looked liked it could have another 3 way tie (Sox/Yanks/Ana?) they came up with the current plan - which while I think its better, I swear was a flip flop to give the Yankees the best chance to make the playoffs.

I've searched the web everywhere though and can not find confirmation about how I remember the original 3 way tie breaker rule though.

Buc

[/ QUOTE ]

My recollection is the same as yours. The new way is definitely better.

[beetman]

The playoff format was changed in 1998 or 1999 I believe.

[beetman]

Actually it was 2003, scroll down to the part here that begins "In 2003"

http://www.all-baseball.com/ref/wildcard.html

[bucrogers]

Beetman - YOU DA MAN! You have no idea how long I've looked for proof that my memory wasn't slipping in my old age.

Thanks!

Buc

[crockpot]

[ QUOTE ]
Well CoolStandings dot com has a better Monte Carlo simulation. They also play the season out a million times, but they have at least some of the proper tie breakers in place. Their simulations definitely take into account the tie breaker case for Detroit and Minnesota (Detroit wins the division if they tie).

I'm not sure if they take into account what would actually happen if Philly, LA and SD all finished with the same record. Off the top of my head, I believe that SD would play at LAD, and the loser would be out of the playoffs. I could definitely be wrong, as I haven't really looked into it (although I know LAD won a coin flip so I know SD would be the visitor).

[/ QUOTE ]

though coolstandings does take these factors into account, that doesn't necessarily make their results more accurate, since all they look at is RS/RA, not the elements. this leads to flawed conclusions, like the assumption that the cardinals are still the best team in their division.

FWIW, i built a model taking into account SP matchups. it estimates the following tie probabilities:

AL central (Det wins): 19.4%

NL central 2-way tie: 17.5%
NL central 3-way tie: 1.8%

NL west (SD wins): 9.4%
NL west (playoff, Phi WC): 0.5%
NL three-way tie: 1.5%
NL WC: 21.7%

the main difference between these and the coolstandings ones is that coolstandings overrates the cardinals.

by the way, this is a good thread and makes an excellent point. last year i placed a -EV bet on the white sox to win the wild card because i forgot about this factor. currently the lines on WSEX's AL central, for example, seem to be ignoring this factor. if you own any shares of minnesota there, sell them! you can make more just betting on them in individual games.