Why You Should Care About Playoff Odds

This morning, David Appelman announced our new Playoff Odds section on FanGraphs, which provides forecasts for the rest of the season and turns those into the chances of each team advancing to each level of the postseason, whether it be capturing the wild card, advancing to the division series, or winning each round of the playoffs. It’s a pretty nifty tool, and there’s a lot to be gleaned from the data, so I wanted to use a few examples from the current playoff odds page to illustrate some of the features now included on the site.

The Importance of Remaining Schedules

Since we rolled out our new Standings page earlier this year, we’ve had expected rest-of-season winning percentage for each team all season long, which is a nice way to look at the relative strength of each team. And since we had a rest-of-season forecast and we already knew what teams had done in the past, it wasn’t that difficult to combine them into an updated final win total. However, there was a piece missing; the actual schedules each team would play over the rest of the year.

A few months ago, that wasn’t a big deal, as the variance in strength of schedule over half a season isn’t going to be very large. However, with just a month to go, there are certainly teams that have much easier competition down the stretch than others, and this difference really begins to show up when looking at the new Playoff Odds data.

For instance, on our Standings page, the Nationals rest-of-season Win% is .539. The Steamer/ZIPS projections haven’t given up on Washington as an above average team, but it sees them as a mid-tier good team, not any kind of huge standout, and unlikely to be able to make up enough ground to make a serious playoff push.

However, on our new Playoff Odds page, the Nationals rest-of-season Win% is .582, the third highest mark of any team in baseball. At first, when saw that gap, we figured there was probably some kind of error in the data, or something had gone wrong in writing the code. However, that really just reflects the fact that the Nationals schedule the rest of the year is a cakewalk.

Here is the expected winning percentage for each match-up the Nationals have over the rest of the season:

Vs Marlins (9 games): 63.2%
Vs Mets (7 games): 59.2%
Vs Phillies (6 games): 57.2%
Vs D’Backs (3 games): 53.7%
Vs Braves (3 games): 50.6%
Vs Cardinals (3 Games): 48.4%

22 of their final 31 games come against three of the worst teams in the National League. This is the easiest schedule any team has to play the rest of the year, and once these specific match-ups are accounted for, the Nationals expected winning percentage goes up significantly. As a result, our FG projection mode gives them a 13% chance of capturing one of the spots in the wild card play-in game, probably a much higher chance than you might expect given their current situation. While we might all be used to looking at games back and games remaining and doing the math in our head, the strength of schedule is a real factor that can make a difference, and that’s a significant advantage of using these playoff odds figures.

Is Making the Wild Card Game Equal to Making the Playoffs?

CoolStandings has been displaying playoff odds for years, and the term has always been kind of self explanatory, since there was no real differentiation in between a division title or a wild card berth once the post-season began. Now, however, there is a huge difference between winning your division and having to play a winner-take-all elimination game for the right to advance to the division series, and it’s a fair question to wonder what “playoff odds” should actually be. Previously, they’ve just been the chances of winning the division added to the chances of winning the wild card, but now, those two things are not equal in value. But, at the same time, the wild card game is part of the playoffs; it could be looked at as a best of one series instead of a best of five, and I don’t know that we want to start using series length to determine what is and what is not considered the postseason.

So, we’ve split the baby and decided to show playoff odds several ways. On the Playoff Odds page, you’ll find POFF, which is just the standard way of showing playoff odds, adding a team’s chances of winning the division and winning the wild card together. However, we also have DOFF, which is a team’s chance of making the division series round, so the math includes the chances of winning the division plus the chance of winning the wild card play-in game, not just reaching it. This is what would have been considered “playoff odds” before the addition of the wild card game.

The changes to the playoffs have made it so that the term “playoff odds” isn’t quite so easy to define, so we’ve just gone ahead and provided the answers to both questions that could theoretically be described as asking what a team’s playoff odds are. And then, for good measure, we’re also showing a team’s chances of winning each subsequent round, so that you can also reference a particular team’s chances of winning their Division Series, Championship Series, and World Series.

Choose Your Methodology

As David noted in the roll-out, there are three different methods to calculating Playoff Odds on the page. The primary method, which is the one I’d imagine most of us will use as our default, is the FanGraphs Projection method, which includes forecasted data from the Steamer and ZIPS projection systems and combines it with updated depth charts maintained daily by our FG authors. This method provides not only the largest sample of data for each player and team, but also includes all recent changes to a team’s roster, whether it be from a new acquisition or an injury that will force the team to reallocate playing time.

However, we also are carrying over the two prior modes from coolstandings, which they’ve set up to run throughout almost all of baseball history. Season to date mode — called “Smart Mode” in their parlance — only uses 2013 data as the inputs, and creates a pythagorean expectation of a team’s true talent level, with recent performance weighted more heavily. If you think that teams should only be evaluated on what they’ve done this season, and not on any prior season data, then this is the mode for you.

We also have Coin Flip mode — what “Dumb Mode” was called on coolstandings — which is exactly what it sounds like. Every match-up, even a Tigers/Astros game, would award 50-50 odds of winning to each team. We probably won’t be using it for analytical purposes much, but it’s there if you want to see how things would play out in a universe where every team was exactly the same.

We hope you enjoy these new playoff odds, and find them as useful as we do. We’re excited to have them on the site, and you can expect to see them referenced pretty frequently over the rest of the season.




Print This Post



Dave is a co-founder of USSMariner.com and contributes to the Wall Street Journal.

33 Responses to “Why You Should Care About Playoff Odds”

You can follow any responses to this entry through the RSS 2.0 feed.
  1. tz says:

    Dave,

    Thanks for dropping my productivity down to the “replacement worker” level. Cool stuff.

    tz

    +36 Vote -1 Vote +1

  2. me says:

    interesting piece could be found in contrasting contenders’ ALCS/NLCS% with their WS%. I noticed that the A’s had an 8% chance of making the world series with a 4.1% chance of making it. Maybe I’m not understanding this correctly, but that seems to suggest they’re a favorite against the NL field (4.1 > (8/2)). I wondered whether this is an AL bias, but while it looks like it’s true for most AL teams, it’s not the rule. For example, the Rays show at 11.8 and 5.1. So, even though the projections clearly don’t think the A’s are as good as the Rays, and are therefore less likely to make the World Series, they DO think that the A’s will win a higher percentage of the World Series that they appear in.

    Either I’m a) misunderstanding what’s going on, or b) your projection contains something that looks at World Series games as somehow different from other games. If so, maybe it’s proprietary, but I’m really curious why this is!

    NOTE: bizarrely, the data seems to have changed in the last two hours without any games having been played, as the A’s are now listed at 8.8 and 4.3, which would make my observation moot. I screenshotted the earlier data, since I thought it was so curious, but can’t embed it. No second-place teams are now listed as being WS favorites. ??

    Vote -1 Vote +1

    • me says:

      Sorry, I don’t know why there are question marks at the end.

      Vote -1 Vote +1

    • Keep in mind the data is being updated hourly with potential new depth chart information, so the projections can change without anything actually happening.

      Vote -1 Vote +1

    • OtherSideoftheCoin says:

      There should be some sort of bias for the AL since they have home field advantage. But maybe this is negated by the majority of NL projected playoff teams having a higher winning % than Tampa Bay and Oakland?

      Vote -1 Vote +1

  3. Dustin says:

    Given the shape of the team at the end of last year, it’s very weird to see the Red Sox as WS favorites.

    Vote -1 Vote +1

  4. Sean says:

    The Diamondbacks have an easy ROS schedule, too (.485), yet you give the Nationals TWICE the chance to make the playoffs over the Diamondbacks despite being a further two games behind?

    Vote -1 Vote +1

    • tangotiger says:

      Misunderstanding:

      The “ROS WIN%” is the winning percentage expected for the rest of the season. The Nats are at .585, meaning that their talent (say .520) plus their easy schedule (say .435 opposing teams) gives them an expected .585 winning percentage. (Basically, .520 – .435 + .500 = .585. Not that simple, but good enough.)

      For the DBacks, it’s more like .520 talent against a tough schedule (.535), which is why their rest-of-season winning% is under .500.

      Vote -1 Vote +1

      • Sean says:

        No, I get that. The Diamondbacks rest of season SOS is .485 while the Nats is slightly lower.

        I get that the Nationals have a high expected win percentage since their schedule is easy (I think it’s .470 or so). But that’s not enough to warrant the playoff odds.

        If you look at FanGraphs projected ROS, the Diamondbacks are projected to go under .500 for those 31 games, even though they play 5 games against SD, 7 games against SF, 3 against Tor, and 6 against CO.

        I’m essentially questioning the projections used for the teams, e.g. the Nationals going .585 ROS against a .470 schedule while the Dbacks go > .500 ROS against a .485 schedule.

        Vote -1 Vote +1

        • Sean says:

          Edit: Dbacks go < .500 ROS

          Vote -1 Vote +1

        • Al Dimond says:

          The 7 games against SF may provide a hint here. SF has a really lousy record this year but they haven’t blown up their team and have lots of players that have been better in the past. So they project to be less lousy over the rest of the season.

          Vote -1 Vote +1

        • Ben Hall says:

          Where did you get the strength of schedule number? I don’t see it on the FanGraphs page.

          Vote -1 Vote +1

        • Sean says:

          I had to calculate SOS manually.

          The issue I’m starting to see comes with how FanGraphs rates the Diamondbacks. Apparently Goldschmidt, Hill, and Chavez are going to regress quite a bit in September.

          But that’s not nearly as bad as our pitching. Corbin is going to “lead” the team with a 3.80 ERA ROS. The rest of our staff will apparently be over 4.00 with most of them closer to 5. David Holmberg is slated to pitch more than Trevor Cahill.

          I know issues like these exist across the board, but when I compared AZ to WASH, the regression was far stronger for AZ. Which is weird, because AZ has drastically underperformed (same with WASH) this year.

          Vote -1 Vote +1

  5. swieker says:

    Great work, Fangraphs and Coolstandings. Thanks!

    Vote -1 Vote +1

  6. Philip Christy says:

    I don’t get the point of the Coin Flip mode. Isn’t that basically saying that every team will play .500 ball from here on out? And if that’s the case, then we already know which teams will end up where, in the exact same spots in the division and wild card races as they are now.

    Vote -1 Vote +1

    • Eminor3rd says:

      Yeah, I’m not seeing the utility in this mode, either.

      Vote -1 Vote +1

    • Teej says:

      I’ll let someone else explain its usefulness, but just a note: 50-50 odds doesn’t mean every team will go .500.

      Vote -1 Vote +1

    • tangotiger says:

      It doesn’t mean they will “outcome .500″.

      It means that both teams have an equal chance at winning. You can still get one team go 20-10, and another go 10-20, even if they are equals.

      Vote -1 Vote +1

      • Philip Christy says:

        Yes, but doesn’t this system work off of thousands of simulations? Over a long enough time span basically all teams will be .500.

        Vote -1 Vote +1

    • MobiusKlein says:

      coin flip is nice when evaluating the probability space in general, devoid of assumptions about team strength.

      e.g.: If there are three games left, and team A has to win all three, and team T has to lose all three, you would see 1/2^6 or 1/64 chance.
      Unless team A plays team T for the last three games, and the probability was 1/8. (Or in hindsight 1/1.)

      It’s also a useful way to check your simulations are not crazy wrong.

      Vote -1 Vote +1

  7. There are some pretty large discrepancies between this report and the odds report on baseballprospectus. A few examples:

    Tigers’ WS win % – 25.1 (B-pro) / 17.1 (FG)
    Red Sox’ WS win % – 10.3 (B-pro) / 21.5 (FG)

    Nationals’ Playoff % – 3.3 (B-pro) / 12.6 (FG)
    Reds’ Playoff % – 92 (B-pro) / 82.5 (FG)

    Obviously those two pairs are connected, but does anyone know enough about the differences in calculation to offer some explanation?

    Vote -1 Vote +1

  8. Ned Yost says:

    Obviously “season to date” mode is most useful…

    Vote -1 Vote +1

  9. CubsOfTheNorth says:

    I fear this is when Fangraphs jumped the shark.

    Vote -1 Vote +1

  10. JoeyO says:

    While I hate to be ‘That Guy’, I do feel obligated to point out there seems to be a massive flaw somewhere in your calculations. Specifically with regards to the mid-1% for the Royals, a number that is clearly way out of whack.

    After all, just earlier today Danny Duffy was recalled solely because, as Ned Yost said, “we want Duffy on our playoff roster”

    So yeah, sorry, but apparently this needs to go back to the drawing board until you properly have KC playing October Baseball. The Process says so…

    Vote -1 Vote +1

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Current day month ye@r *