Before we get to Jonathan F’s brilliance, Playoffs are a Crapshoot Part 3, Here are Notes from Matt L:
Last night, we successfully migrated BravesJournal to a new host. This officially gives Ryan and the community full access/ownership to BravesJournal and full administration permissions. Additionally, this will (and already has in most cases) enable us to more thoroughly diagnose and fix bugs, add additional features, speed up the site, and (finally) secure the site to today’s standards.
And now…Playoffs are a Craphsoot, Part 3
In case you missed the start of the series, here’s the first 2 pieces:
The Braves won 60% of their games last year. If they were expected to win 60% of their games in the playoffs, their chances of winning the World Series (as a division winner) would be 34%. But everyone recognizes that their chances of winning a playoff game is considerably lower than 60%. To begin with, there are no teams in the playoffs as bad as the Marlins: the Braves’ 60% winning percentage in the regular season is greatly enhanced by the fact that they played the Marlins 19 times and won 79% of them. (On the other hand, Kansas City didn’t make the playoffs either, and the Braves managed a phenomenal 25% winning percentage against them.)
Since the teams in the playoffs are all good teams, we need a way to measure their chances of winning the World Series that adjusts for this. As I discussed here several years ago, the easiest way to do this is with what are known as Bradley-Terry ratings. These use everyone’s records against everyone else to derive a kind of power ranking that can then be used to derive a head-to-head win probability against any team. I’m not going to explain those again: I’ll just give the Bradley-Terry ratings (the simple version unadjusted for home field advantage) for each of the 10 playoff teams based on the 2019 regular season:
It is of some interest that these 10 teams also had the 10 highest ratings. That is certainly not always the case. The 11th ranked team was the Mets, if you’re interested…. also if you have no interest at all.
So the Braves’ first opponent was the Cardinals. (I had to look that up… somehow it had slipped my mind.) When the Braves played the Cardinals, the chances of a Braves win in any particular game is 52.8%. It should be clear that we’re already close to crapshoot territory. The Braves’ chance of winning that 3 out of 5 series is then 55.2%. Had they survived that round, they then would have (fortunately) played the Nationals. Their chance of winning a head to head game with the Nationals is slightly lower: 52.3%, but in a 4 out of seven series, their winning probability is 55.0%. They next would have played the Astros, with a probability of a single game win at 46.2% and a WS probability of 41.7%. Combining these, even taking into account the favorable odds that came from the gNats knocking out the Dodgers, their championship chances were 0.552 x 0.550 x 0.417 = 12.7%. That is of course essentially equal to their chances if the WS champion was pulled from a hat once the wild card games were completed.
The Dodgers had the best chance, by these rankings, of winning the World Series. We can now calculate the Crapshootiness Index using these rankings. We can calculate the overall probability of the Dodgers winning the World Series (using only the win expectations and the formulas here and in the previous installment) at 27%. This makes the Crapshootiness Index for last year at 0.83: the Dodgers were over twice as likely to win the World Series as pulling their name out of a hat would have estimated. But as we will see in the next installment, even this estimate overstates the odds of a Dodger win.
Thanks for reading Playoffs are a Crapshoot. If you missed our Top-30 Braves Prospects, you can find them all here.
19 thoughts on “The Playoffs Are a Crapshoot, Part 3: Adjusting for Opponent Quality”
Jonathan, for some reason there’s a typo and one win probability didn’t paste into this sentence:
“When the Braves played the Cardinals, the chances of a Braves win in any particular game is .”
What was that probability?
I think Jonathan had an image that did not transfer over to the post. Hopefully he can get to it soon and place it back.
I love the new layout, Matt!
That’s a lot of math!
DOB reports that the Braves have offered four years to Donaldson and others report that so have the Nationals and Twins. If it’s true that JD “prefers” Atlanta, then one must surmise that the Braves have an offer with a lower AAV than the other two. I think the baseline is 4/$92M which is an AAV of the $23M he earned last year.
I believe it. My source told me Braves had went 3 years with 2 options last week.
Greetings from Brussels, home of the Sprouts.
Yeah, a graphic got lost. To get head-to-head probabilities, you take BTa/(BTa+BTb), where BTa and BTb are the Bradley-Terry ratings of the two teams. So head-to-head against the Cardinals, it’s 78.2/(78.2+69.9) or 52.8%.
So — just to make sure I get it, is Bradley-Terry basically the same as the old Bill James Log5?
@JonathanF I added the 52.8% to the post for you. If you have access to the missing image, send it to me & I can add it for you.
AAR: Yep, and the “ballast” discussion from the comments is what we’re going to talk about in the next installment. Thanks… I hadn’t seen that before.
Great series. Not that I have the math needs to understand it of course.
For Smitty the morning after is just as sweet as the night before.
You have a starting pitcher who last year went 18=4. At that even, he was not allowed to pitch the last 18 days of the season, nor will he be able to go to Spring Training or pitch the first 63 days of the new season. 18 plus 63 equals 81 so you can guess from that he was likely violent towards women, just the fourth to be so punished since the new rules went into force in 2015. One of those 4 was a Brave, you will know his name of course.
So what does his club do to address this yawning gap? Sign Gerritt Cole of course, overkill be damned.
used edit feature above, first time, worked perfectly.
Luis Robert signs 6 year “extension” with White Sox for $50M. While he has yet to play in the majors, his contract from signing as a Cuban immigrant was granting him free agency in 2027 so this actually buys out a couple free agent years. Given that he would be starting in the bigs with “only” 1 pre-arbitration year, this is extremely comparable to the proposal I posted about a Christian Pache extension, although it suggests I was about $3M high for each of the buyout years. So, my new Pache proposal is $89.5M over 9 years.
Original proposal here: http://bravesjournal.us/2019/12/22/atlanta-braves-extension-candidates-cristian-pache/
That should have said he was a free agent after 2025, not 2027. AAArgh!
There’s a new thread.