Postseason Odds from Baseball Prospectus

[Benjamin Grimm]

Hey! Remember last September, when I'd keep posting the Mets' playoff odds from Baseball Prospectus? Wasn't it fun seeing the number go from 99.9 per cent down to zero in a matter of weeks? (I was originally posting the stats in order to get comfort from the high odds, but it went terribly, terribly awry.)

Anyway, I took a peek just now and here's where we're at:

Average wins by position in NL East:  90.0 85.3 81.3 76.7 61.9 90.1
NL East           W    L    Pct3  Avg W  Avg L  Champions  Wild Card   Playoffs      1D Change 7D Change
Phillies          48   42   .526   87.4   74.6   51.72104    5.64145   57.36250       -4.14507   4.42151
Marlins           46   43   .470   80.3   81.7    8.83299    2.19664   11.02963        2.35417  -5.27805
Mets              45   44   .518   85.2   76.8   31.09188    6.49587   37.58774        5.13710  17.12308
Braves            42   48   .518   80.4   81.6    8.35349    2.48949   10.84298       -1.96723 -16.64145
Nationals         34   56   .385   61.9  100.1     .00060     .00000     .00060       -0.00057  -0.00620

I'm guessing that the Mets have a softer schedule than the Marlins do the rest of the way, because their odds are considerably higher despite them being currently behind Florida in the standings.

If this data is correct, the Mets can win the division with as few as 88 wins, and, if they get to 91, would really be in good shape.

To get to 88 wins, they'd have to go 43-30 (.589) and to get to 91, they'd have to go 46-27 (.630).

I think our best hope is to have the Phillies win 84 games or fewer, which seems unlikely, but it would allow the Mets to win with 85 games. (40-33, .548) For this to happen, the Phils would have to go 36-36 (.500).

(My evil plan does kind of rely on the Marlins fading and no longer being a factor. And of course, that might not happen at all.)

[Methead]

"I'm guessing that the Mets have a softer schedule than the Marlins do the rest of the way, because their odds are considerably higher despite them being currently behind Florida in the standings."


Could be the schedule, but the Mets also have a higher run differential than Florida. The Marlins' run differential is actually negative.

[Guest AG/DC]

[Guest John Cougar Lunchbucket]

The Jerry Era looks like Charlie Brown's shirt.

[Guest Vince Coleman Firecracker]

The PECOTA adjusted version has the Mets at 38% to win the division against the Phillies' 49.4%.
Then again, PECOTA had Delgado hitting .265/.343/.471 instead of .239/.318/.433.

[batmagadanleadoff]

At this stage of the season, those BP playoff odds are highly volatile. Thus, the Mets win three of four from Philly and have their odds escalate from slim to good. And that's not so much a knock on the BP system as it is an accurate reflection of the fact that there's still half a season left and for all of the Mets controversy this year, they were never far from the division leaders.

[TransMonk]

ESPN has added playoff chances to their standings page. They have the Mets much lower than BP. I wonder where their numbers are coming from.

[Frayed Knot]

BP definitely takes run differential and schedule into consideration.
ESPN might not look at either.

[Guest Vince Coleman Firecracker]

Frayed Knot wrote: BP definitely takes run differential and schedule into consideration. ESPN might not look at either.

ESPN uses coolstandings' formula to calculate playoff odds. It's a variation on the Bill James Pythagorean theorem combined with, umm, a coin toss? Am I reading that right?

[Valadius]

Nate Silver (creator of PECOTA) has been working on another project this election season - http://www.fivethirtyeight.com. It uses regression analysis to predict what's going to happen in November. I read it every day.

[Benjamin Grimm]

In November, I predict that President Bush will pardon the White House Thanksgiving turkey.

[Guest AG/DC]

I like that that graph is updating.

[Guest Vince Coleman Firecracker]

FWIW, the Mets have passed the Phils in the PECOTA-adjusted projections, and are damn close in the original version.

Not so much the ELO report, though. C'mon ELO, don't bring us down. Bruce.

[Benjamin Grimm]

Average wins by position in NL East:  90.0 85.5 81.4 76.6 61.5 90.1
NL East           W    L    Pct3  Avg W  Avg L  Champions  Wild Card   Playoffs      1D Change 7D Change
Phillies          48   43   .526   86.8   75.2   44.56620    5.58918   50.15538       -7.20712  -9.59211
Marlins           46   44   .464   79.2   82.8    6.04122    1.25933    7.30055       -3.72908  -4.92413
Mets              46   44   .525   86.2   75.8   38.69230    5.76567   44.45797        6.87023  20.37172
Braves            43   48   .523   81.3   80.7   10.69957    2.61042   13.30999        2.46701  -8.23942
Nationals         34   57   .385   61.5  100.5     .00072     .00000     .00072         .00012  -0.00881

The Mets are averaging 86.2 wins to the Phillies' 86.8.

[Guest Grote15]

Benjamin Grimm wrote: Hey! Remember last September, when I'd keep posting the Mets' playoff odds from Baseball Prospectus? Wasn't it fun seeing the number go from 99.9 per cent down to zero in a matter of weeks? (I was originally posting the stats in order to get comfort from the high odds, but it went terribly, terribly awry.) Anyway, I took a peek just now and here's where we're at: Average wins by position in NL East: 90.0 85.3 81.3 76.7 61.9 90.1 NL East W L Pct3 Avg W Avg L Champions Wild Card Playoffs 1D Change 7D Change Phillies 48 42 .526 87.4 74.6 51.72104 5.64145 57.36250 -4.14507 4.42151 Marlins 46 43 .470 80.3 81.7 8.83299 2.19664 11.02963 2.35417 -5.27805 Mets 45 44 .518 85.2 76.8 31.09188 6.49587 37.58774 5.13710 17.12308 Braves 42 48 .518 80.4 81.6 8.35349 2.48949 10.84298 -1.96723 -16.64145 Nationals 34 56 .385 61.9 100.1 .00060 .00000 .00060 -0.00057 -0.00620 I'm guessing that the Mets have a softer schedule than the Marlins do the rest of the way, because their odds are considerably higher despite them being currently behind Florida in the standings. If this data is correct, the Mets can win the division with as few as 88 wins, and, if they get to 91, would really be in good shape. To get to 88 wins, they'd have to go 43-30 (.589) and to get to 91, they'd have to go 46-27 (.630). I think our best hope is to have the Phillies win 84 games or fewer, which seems unlikely, but it would allow the Mets to win with 85 games. (40-33, .548) For this to happen, the Phils would have to go 36-36 (.500). (My evil plan does kind of rely on the Marlins fading and no longer being a factor. And of course, that might not happen at all.)

Funny..88 was the lucky number i was speculating on..Can we have 5-10 stoopid losses played again? Nice work.

[Benjamin Grimm]

Benjamin Grimm wrote: To get to 88 wins, they'd have to go 43-30 (.589) and to get to 91, they'd have to go 46-27 (.630). I think our best hope is to have the Phillies win 84 games or fewer, which seems unlikely, but it would allow the Mets to win with 85 games. (40-33, .548) For this to happen, the Phils would have to go 36-36 (.500).

Now, three days later, for the Mets to get to 88 wins they need to go 40 and 30 (.571). 91 wins would be 43-27 (.614). 85 wins would require 37-33 (a mere .529).

To keep the Phillies to 84 wins, they'd have to go 34-35. (.493)

[Benjamin Grimm]

Here's the latest. Mets at around 50% for making the playoffs, significantly better than the Marlins, with whom they're tied.

Nationals, with the worst record in baseball, have the most infinitesimal playoff odds.

Average wins by position in NL East:  91.3 86.6 82.2 77.4 62.7 91.4
NL East           W    L    Pct3  Avg W  Avg L  Champions  Wild Card   Playoffs      1D Change 7D Change
Phillies          51   43   .535   88.5   73.5   48.39887    9.30102   57.69988        2.94536 -16.35987
Marlins           49   44   .473   81.4   80.6    7.58727    3.03856   10.62583        1.45254   2.17291
Mets              49   44   .532   87.6   74.4   39.85823   10.02838   49.88662         .55925  28.89307
Braves            43   50   .521   79.9   82.1    4.15545    2.03353    6.18898       -2.94667  -7.67118
Nationals         36   58   .395   62.7   99.3     .00018     .00003     .00022       -0.00001  -0.00202

[RealityChuck]

There are all very nice mental exercises, but ultimately silly and meaningless.

Or have your forgotten that the Mets odds of making the playoffs in September 2007 -- from the same sources -- were 99%?

[Benjamin Grimm]

Clearly I didn't forget.

Benjamin Grimm wrote: Hey! Remember last September, when I'd keep posting the Mets' playoff odds from Baseball Prospectus? Wasn't it fun seeing the number go from 99.9 per cent down to zero in a matter of weeks? (I was originally posting the stats in order to get comfort from the high odds, but it went terribly, terribly awry.)

[RealityChuck]

Actually, it's a misnomer and misrepresenation* to call these "odds." It merely is a reflection of how often the teams won in their simulations.

But the simulations are built on so many iffy assumptions as to make them meaningless: that players don't go into slumps, are not injured, that trades are not made, that past performance is a guarantee of future results, etc. All the numbers mean is that the simulations showed these results a certain percentage of the time.

Their connection to the real world is tenuous at best and, at worse, no better than Madame Olga's crystal ball.

*A typical tactic of sabermatricians, who often design models to determine certain characteristics and then claim the models prove or disprove something else.

[Guest AG/DC]

They're not meaningless. They have as much meaning as they represent. I don't know why you want to keep swinging the sweeping axe against those horrid statisticians. "Or have you forgotten...?" It's just obnoxious.

If you really want to put your money on Madame Olga's guess, and I put mine on Baseball Prospectus, I'm certain I'll win more than you.

[nymr83]

statistics are a form of mystical voodoo while baseball is won with "heart" "clutch" and "jeter-ness"

[Guest Vince Coleman Firecracker]

Don't forget me!

[Guest Vince Coleman Firecracker]

RealityChuck wrote: But the simulations are built on so many iffy assumptions as to make them meaningless: that players don't go into slumps, are not injured, that trades are not made, that past performance is a guarantee of future results, etc. All the numbers mean is that the simulations showed these results a certain percentage of the time.

Actually, Baseball Prospectus now works injury probability into their PECOTA projections. Slumps and hot streaks will regress to a player's mean over a large enough sample size. Trades are not accounted for, but I don't think there's anyway to expect a system to project a team using players that aren't on their roster. Past performance is not a guarantee of future results, but it is a fantastic tool for determining the probability of a future event, which is exactly what these odds try to do.

Last year, the Mets were extremely likely to win the division heading into the final few games. That they didn't win the division is not an indictment of the odds, but rather an indication of how extremely unlikely the Mets' collapse was. Seeing how a collapse of that magnitude has only happened twice in the league's history, it seems like the odds were spot-on in their assessments.

[Guest AG/DC]

And that's where we keep losing Reality Chuck. If the outcome isn't guaranteed (of course it isn't) the results are "meaningless." He sees no value in shaving away unlikelhoods and arriving at most likely scenarios.

I see a lot of value, and any good investor would, because while the most likely scenario, taken exactly, still remains highly improbable, the lion's share of outcomes are clustered around that, and would still pay off. (Plus he insists that factors aren't included which are.)

An informed and analyzed opinion from Baseball Prospectus says the Nats will make the playoffs .00022% of the time. An un-imformed and un-analyzed opinion from a crystal ball would give the Nats the playoffs 27.6923% of the time. Which do you want to go with?

[Benjamin Grimm]

Well, with the Mets and the Phillies in a dead heat for first place, I thought it might be interesting, I mean, "meaningless" to take another look at Baseball Prospectus' odds:

Average wins by position in NL East:  92.0 87.3 82.7 78.0 61.3 92.1
NL East           W    L    Pct3  Avg W  Avg L  Champions  Wild Card   Playoffs      1D Change 7D Change
Mets              52   44   .540   89.5   72.5   51.08884   11.22992   62.31876        2.29665  12.99139
Phillies          52   44   .534   88.3   73.7   38.96599   12.66505   51.63104       -1.22309  -3.12348
Marlins           50   45   .469   81.1   80.9    5.27686    2.81910    8.09596       -0.35266  -1.07733
Braves            45   50   .526   81.1   80.9    4.66821    3.00684    7.67505       -0.47074  -1.46060
Nationals         36   60   .390   61.3  100.7     .00010     .00000     .00010         .00000  -0.00013

Mets have a nice edge over the Phillies at this point. (That guarantees a division title, doesn't it? After all, I have no memory of what happened last season!) The NL's leading Wild Card candidate is the third-place Brewers. If the season were to end as BP is predicting, the first round of the NL playoffs would be Mets vs. Brewers and Cubs vs. Diamondbacks. (In the AL it would be Red Sox vs. Angels and White Sox vs. Rays.)

[Guest Vince Coleman Firecracker]

W/L/Pct3/Avg W/Avg L/Champions/WildCard/Playoffs/1D Change/7D Change

Nationals/38/ 61/.398/62.8/99.2/.00000/.00000/.00000/-0.00010/-0.00015

Ouch. They ran the remaining season a million times and the Nats never made the playoffs. Seattle, 7 games further out of first and with basically the same record, got in 108 times.

[Benjamin Grimm]

Three games later, the Mets odds for the division title have increased, but their overall playoff odds have gone down, thanks to the Brewers having taken a bigger bite of the Wild Card odds.

The Nationals have regained a small glimmer of hope.

In the NL West, the Dodgers are now favored over Arizona. The BP playoff picture is now Mets/Brewers, Cubs/Dodgers, Red Sox/Angels and White Sox/Rays.

Average wins by position in NL East:  91.2 86.6 82.3 77.7 62.8 91.3
NL East           W    L    Pct3  Avg W  Avg L  Champions  Wild Card   Playoffs      1D Change 7D Change
Mets              53   46   .540   88.9   73.1   53.22209    6.17341   59.39550         .70964  -0.60971
Phillies          53   46   .530   87.4   74.6   36.74687    7.25895   44.00583         .26875  -8.85860
Marlins           52   47   .469   81.3   80.7    6.16526    1.64830    7.81356       -2.07461  -0.71365
Braves            47   52   .520   80.2   81.8    3.86572    1.16884    5.03456        1.23460  -3.12410
Nationals         38   61   .398   62.8   99.2     .00005     .00000     .00005         .00005  -0.00008

[Benjamin Grimm]

The Mets have played six games (winning four of them) since I last posted an update in this thread.

Here's where we're at now:

Average wins by position in NL East:  91.9 87.1 82.5 78.1 59.8 91.9
NL East           W    L    Pct3  Avg W  Avg L  Champions  Wild Card   Playoffs      1D Change 7D Change
Mets              57   48   .557   90.5   71.5   64.88740    6.90296   71.79036        5.13699  13.10450
Phillies          56   49   .527   87.3   74.7   27.72262   10.59879   38.32140        2.94875  -5.41568
Marlins           55   50   .472   81.8   80.2    5.23053    1.90949    7.14002       -2.26021  -2.74815
Braves            49   55   .524   80.0   82.0    2.15945    1.01150    3.17094       -1.65685  -0.62902
Nationals         38   67   .390   59.8  102.2     .00000     .00000     .00000         .00000    .00000

The Mets' postseason odds have taken a nice jump, partially at the expense of the Phillies. Marlins have pretty much held steady. (Hopefully the next three days will knock the Marlins back a bit.) Their numbers are surprisingly low, considering how close they are to the division lead.

Milwaukee is still projected to be the Wild Card, but the Arizona has reclaimed the top spot in the NL West from the Dodgers. Cubs still project to the best record in the league, so the Mets' projected first-round opponent remains the Brewers.

In the AL, the Red Sox are now favored by BP to win the East, and the Rays are now the leader for the Wild Card. First round ALCS would be Red Sox/White Sox and Angels/Rays. (Too bad Tampa Bay changed its name; we'd have Sox vs. Sox and Angels vs. Devils.)

[Guest AG/DC]

Wow, the Nats' probability has shrunk below five decimal places.