claydavenport.com
My first run (that I’m willing to talk about) of projections for the coming season is now up on the 2014 Projected Standings tab. They have also been used to create a new Playoff Chances Report. And, of course, the individual projections that go into are available, again on the Projected Standings page.
| American League | |||||||
| East | Won | Lost | Runs | Runs A | Champ | Wild Card | Net Playoff |
| Tampa Bay | 90 | 72 | 698 | 618 | 45.8 | 19.1 | 65.0 |
| Boston | 86 | 76 | 723 | 680 | 22.8 | 19.2 | 42.0 |
| NY Yankees | 85 | 77 | 683 | 646 | 21.6 | 18.8 | 40.4 |
| Toronto | 78 | 84 | 720 | 749 | 5.9 | 7.8 | 13.7 |
| Baltimore | 77 | 85 | 693 | 733 | 3.9 | 5.5 | 9.4 |
| Central | Won | Lost | Runs | Runs A | Champ | Wild Card | Net Playoff |
| Detroit | 91 | 71 | 711 | 618 | 60.1 | 14.9 | 75.0 |
| Cleveland | 85 | 77 | 717 | 682 | 24.1 | 19.7 | 43.8 |
| Chicago WS | 79 | 83 | 682 | 701 | 8.2 | 9.9 | 18.1 |
| Kansas City | 77 | 85 | 680 | 712 | 5.9 | 7.5 | 13.4 |
| Minnesota | 72 | 90 | 669 | 752 | 1.7 | 2.4 | 4.1 |
| West | Won | Lost | Runs | Runs A | Champ | Wild Card | Net Playoff |
| Oakland | 88 | 74 | 723 | 655 | 35.9 | 20.7 | 56.5 |
| Texas | 87 | 75 | 731 | 676 | 30.6 | 20.7 | 51.3 |
| LA Angels | 84 | 78 | 712 | 685 | 17.5 | 16.9 | 34.4 |
| Seattle | 83 | 79 | 707 | 690 | 15.2 | 15.7 | 30.9 |
| Houston | 70 | 92 | 676 | 781 | 0.8 | 1.2 | 2.0 |
| National League | |||||||
| East | Won | Lost | Runs | Runs A | Champ | Wild Card | Net Playoff |
| Washington | 87 | 75 | 661 | 612 | 46.2 | 17.0 | 63.2 |
| Atlanta | 85 | 77 | 673 | 641 | 34.3 | 18.4 | 52.7 |
| NY Mets | 78 | 84 | 639 | 666 | 10.7 | 10.3 | 21.0 |
| Miami | 75 | 87 | 616 | 670 | 5.5 | 5.9 | 11.4 |
| Philadelphia | 72 | 90 | 615 | 690 | 3.3 | 3.7 | 7.0 |
| Central | Won | Lost | Runs | Runs A | Champ | Wild Card | Net Playoff |
| St Louis | 90 | 72 | 698 | 619 | 58.0 | 17.4 | 75.3 |
| Pittsburgh | 83 | 79 | 660 | 639 | 21.7 | 21.4 | 43.0 |
| Cincinnati | 80 | 82 | 633 | 640 | 12.3 | 15.5 | 27.8 |
| Milwaukee | 77 | 85 | 654 | 690 | 7.4 | 10.8 | 18.3 |
| Chicago Cubs | 67 | 95 | 598 | 721 | 0.6 | 1.2 | 1.8 |
| West | Won | Lost | Runs | Runs A | Champ | Wild Card | Net Playoff |
| LA Dodgers | 88 | 74 | 649 | 593 | 40.3 | 21.8 | 62.1 |
| San Francisco | 85 | 77 | 659 | 624 | 27.6 | 22.0 | 49.7 |
| San Diego | 83 | 79 | 670 | 648 | 22.3 | 20.6 | 43.0 |
| Arizona | 78 | 84 | 651 | 676 | 8.4 | 11.5 | 19.8 |
| Colorado | 71 | 91 | 655 | 748 | 1.4 | 2.5 | 4.0 |
To build these projections, I:
1) Run a computerized projection scheme, using the last three years of player performance compared against a database of all players’ four year performances. The algorithm attempts to find the most similar players, in terms of age, position, build, and performance, and the top 20 players are noted on the individual player cards.
2) Take those performances, and enter them into a very large spreadsheet, where I fill in expected playing times for all of the players. Every team, every position has to equal 100%. There have to be 162 pitching starts. Generally speaking, a) no position player gets more than 90%, and pitchers are mostly capped at 32 starts; b) rookie starters don’t get more than 80%; c) players I don’t think can hold the job all year certainly get less; d) the playing time estimates from the computer tend to carry a lot of weight. I normally set a sure starter to the 5% playing time level that first passes their projected PA, while innings are usually held under the computer’s values.
All of the statistics in the spreadsheet get rebalanced and weighted. Players on teams with high OBAs will get more plate appearances. Defense trickles back into pitchers hits (and runs) allowed. The league as a whole has to come out equal to the league totals of last year.
Current free agents won’t show up here – no team, no projected playing time. Their projections are still available on the “All hitters” and “All pitchers” downloads.
Getting to some of the players takes a deep depth chart. I’ve prepared some that you can find under the 2014 Spring tab, under “dts”. Every team has three files in there. One is a dt file, which contains the translated statistics, 2009-13, with the computer-only 2014 projection, for all hitters in that team’s system; another is a pdt file, which does the same for pitchers. The “orgdt” file just has the 2014 projections for all players on the team, sorted by position and projected WARP, like the one here for the Nationals. Kind of works as a very deep depth chart for all teams, although I can’t swear that aren’t players showing up on the wrong team (especially for players who have been released – there’s a decent chance they still show up for their old teams). That’s just for these depth charts – I am reasonably certain that every player used in the major league projections is actually a member of their team. The one exception might be Matt Garza, who I have already written into the Milwaukee rotation.
If you enjoyed this article, please consider sharing it!
[…] 75%. They are also projected to lead baseball in wins with 91. They really don't like KC. First Projections for 2014 | claydavenport.com Reply With […]
You might want to tweak your software. The Royals aren’t going to be 9 games worse than they were last year, bro.
Insane! How does the computer project the Royals to get worse??? With that defense and relief corps? No way…
Arttie
Ervin Santana is gone for starters.
Show your projections from last year.
On the Projections page, there are links to the 2012 and 2013 numbers posted in late March.
[…] First Projections for 2014 | claydavenport.com This guy predicts the standings based on math and stuff. I think he is severely conservative. That said, it is still interesting to look at. Personally, I think he is 10-12 games too generous with the Cubs. […]
Clay:
To help people understand how the #1 team is forecasted to “average” 91 wins, can you also show the averages for #1 through #30? That is, take the highest win total for each of your simulations (regardless of team), and show us that average. Then do the same for the second highest and so on.
That’s a really good idea. Also, I’d love to see the “best case” and “worst case” win totals for each team, as produced by your simulation. Or maybe a standard deviation? Thanks, Clay!
this is neat man. but my stats say the orioles are much better then 77 wins. we shall see
He has no team winning more than 91 games… very likely.. lol
[…] Cubs are unlikely to be very good in 2014. However, this signing give the Cubs the ability to hand the ball every day to a guy that gives you […]
Thanks for firing up the hot stove!
Hope and stats are disjunct sets.
First thing I do when I see these specific projections is to toss them into a spreadsheet and total the wins and losses. Only 2 different. Probably just a rounding error. Once I saw someone’s NBA projections where the numbers were so inflated that the one team the analyst was non specific on the win total would have to have lost every one of their games, then somehow manage to lose 19 more…
And the AL is projected to be over 30 games better than the NL. Interesting results. Thanks for putting together.
Who the hell is Clay Davenport??
That is exactly the question we International Men of Mystery want you to be asking.
Sorry, but if you think the Reds will be under .500 your computer has a bad virus.
It’s still way too early for projections like this but I do find great fault with 91 wins being the best record in baseball this year. The AL East looks about right standings wise.
Sadly, only money talks. The Phillies’ owners possess a profitable business–win or lose, at least at the moment. Why change? Until fans abandon the franchise, including cable television viewing, this dismal performance will continue. I am always amazed, despite years of subpar performance, that Phillie fans expect a return to greatness when greatness was only a few years–less than 60 months.
Mariners and Padres more with more wins than Cincy, KC, Mets, and same amount as Pittsburgh? I don’t think so
The Dodgers will score the fewest runs in the NL West, even fewer than the Padres??
I find that to be a real stretch, are we just assuming everyone will be hurt again or what?
I would say that the best team in baseball wins more than 91 games. I would take the over on the Reds winning 80, though I think they finish third. I also would take the over on the Nationals and Dodgers on this sheet. The wins seem light.
Detroit again ugh
I think you’re a bit off on the AL East…the Yankees out scored the Rays in the second half last year with a terrible offensive club. If the Yankees pen comes together they are the best team.
The 2012 projections showed the Marlins winning 90 games. 30% of his projections are off by 20 wins one way or the other. In a scientific community Clay Davenport would never be published with that kind of results. Here in Blogshere, he can tout himself as smart. Amazing
Wow, THE Clay Davenport?! 90 wins out of the Cards? Pull the other one jagbag.
A team that wins 97 last year, has only one appreciable loss, is projected to win 90? I think that’s one of the least objectionable projections.
This is why the game is played on the field and not on spreadsheets. These numbers to most of us should make about as much sense as putting a screen door on a submarine. Only a computer would see KC getting worse than getting better.
Submarines get a lot of design on paper before they get built.
No one doubts that KC was better last year. The question is, did they improve in ways that have staying power or in ways that are ephemeral? Re-signing Bruce Chen is an improvement over the pitchers I had at first had in the slots – but Chen was +23 last year in categories that don’t have predictive power, so I expect him to be 10-15 runs worse than last year. Guthrie was +19. Duffy was +6 in only 25 innings. Hochevar was +29; Crow was +26. They should still be good relievers, but betting on them to repeat their sub-2 ERAs is a bad one. The team as a whole gave up 35 fewer runs than their stats suggested – the highest figure in baseball. Its a good thing to have, but it is a lousy thing to build on.
Dear everyone,
Think of these team projections as the middle point in a wave – the crest represents the best case for that team – everyone stays healthy, prospects come up and hit their stride, career years from your stars. The trough represents the worst case – stars get hurt, players regress/have off-seasons, and no one (prospects or otherwise) steps up to fill the void. The case Davenport is describing isn’t what *might* happen, but what the average of lots of simulations indicates an “average” season looks like – i.e. some players do great, some do worse than expected – some people get hurt, some prospects turn into big league regulars – and overall, the team has average outcomes.
In other words, these standings show what a boring, unexciting season for your team looks like, in which everyone is slightly disappointed, because all your hopes and dreams failed to come true. Are you really surprised that these results are disappointing to you?
I’m a Cardinals fan from Kansas City, and I can tell you, the Royals will have to have a “positive outcome” season to have a chance at the playoffs, and the Cardinals will have to have a “positive outcome” season to get back into the mid-90s in wins. Both those teams have a shot at a “positive outcome” season. But there’s a 40% chance that James Shields or Adam Wainwright gets hurt. Lots of bad things *could* happen. And these projections reflect those possibilities, not just the best case. So take a deep breath, recognize that simulation averages, by their nature, are going to feel bad to you.
I have to copy and keep that, because it is a better expression of what I have wanted to say than I have managed to print.
Take it easy people. If you want to criticize, do so respectfully and with some reasaonble reasons why the projections might be wrong. Of course they’re wrong, projections are always wrong. But Clay D. puts a lot of time and effort into this, certainly more than any of us.
One issue I have is the tightness of some of the ranges. Last year in the AL, runs scored ranged from 598 (white sox) to 853 (red sox), a 255 run spread. The projections forecast a spread of only 62 runs — 669 to 731. Basically, Clay has every team scoring 700 runs, plus or minus a few. It doesn’t seem likely that the teams will be that tightly bunched.
No, they certainly will not be. Keep in mind, though, that the best team in scoring is very likely to be a team that is above average offensively AND outscores their projection by a good 50 runs (probably because one or more players has an MVP year). Invert that for the lowest teams. If 250 is what you take as a standard spread from top to bottom – than I don’t think your projection should be any more than 150. And very possibly less than that.
Keep in mind, these are projections where nobody does appreciably better or worse than expected. Nobody’s park factor goes off in an unexpected direction. Nobody gets unexpectedly hurt and misses half the season (that risk is spread rather evenly across all players). No managerial turmoil distracts anyone. When you look at those best or worst teams at the end of the year, you’ll be able to identify all kinds of surprises from the players…that is, after all, why they play the game.
Of course, as you note, I only had a 62 range…which expanded to 87 when I relaxed the regression to mean components (which were clearly overdone on the first draft).
These projection systems will simply forecast team performance at the median of their parts. Surely the Red Sox were way more talented than the White Sox in 2013, but the Red Sox also had a lot of good fortune and overperformance (Napoli, Drew, Ellsbury, Ortiz, Nava, Carp, Lackey, and Uehara all performed better than expected). These projections always understate the eventual spread between teams. Also, there will be more variance around teams like Boston / New York (injury / age decline issues) than Tampa Bay (younger, less likely to see the roof fall in). We’ve seen this the past few seasons, where Tampa Bay has been a practical lock for 90-95 wins, while Boston and New York have been more sporadic.