DMB News May 2005
May 13, 2005
Written by Tom Tippett
Welcome to the second edition of the Diamond Mind email newsletter for the year 2005. Through these newsletters, we will try to keep you up to date on the latest product and technical information about the Diamond Mind Baseball game, related player disks, and our ongoing baseball research efforts. Back issues are available on our web site.
Topics for this issue:
The April update to the 2005 Projection Disk was released on schedule a couple of months ago. It was sent automatically, and at no additional charge, to everyone who ordered the Projection Disk prior to that date. Projection disk orders received after that date will receive only the updated version of the disk. As has been the case in past years, this is our last update to the projection disk for 2005.
We're still working on the version 9b patch and hope to have it ready in a few weeks. As we get closer to that release, we'll keep you posted via our web site and the DMB forum.
In the last newsletter, we mentioned that we'd started work on an update to the All-time Greatest Players disk, but were undecided about whether to do a small update quickly or take more time to add a couple of hundred additional players. Since then, we've heard from a number of our customers, and their overwhelming preference was for the larger update. So that's what we're going to do.
As noted in February, we've been adding real-life transactions and game-by-game starting lineups to several of our Classic Past Seasons. Within the next few weeks, we plan to release updated editions of the 1934, 1946, 1954, 1955, 1965, 1966, and 1977 seasons. In most cases, the only new or revised content will be the transactions and lineups. In the coming weeks, check our web site for more information about the timing, content and price of these releases.
As you know, we put a lot of time and energy into the projected stats and ratings that appear in our annual Projection Disks. And we put a lot of time and energy into simulating the coming season and writing up our projected standings.
That process includes gathering predicted standings from other sources so we can assess our projections at the end of each season. At the moment, our database includes 61 predictions.
That number includes a few that aren't exactly predictions -- the previous year's final standings, the current year's spring training standings, rankings based on opening day payrolls, and standings derived from the Las Vegas over-under line. One entry is the consensus of several hundred SABR members who participated in their predictions poll. The rest are from individuals or publications.
Nine of the 61 include projected wins and losses for all 30 teams. Interestingly, only 2 of these 9 add up. There are 2,430 games on the schedule, so the wins in any set of projected standings should add up to 2,430. They do for ours and those in Baseball Prospectus Today, but the other five are off by as many as 55 games.
To be fair, two of these nine are over-under betting lines, and they have no obligation to make sure things add up. Their goal is to get equal amounts of money bet on both sides, and if bettors tend to be optimistic about their favorite teams, that would push over/under lines up by a few games.
Other publications don't have that excuse, however, and it's a little disappointing to see a major newspaper put forth projected standings that could never actually happen unless 55 games were magically added to the schedule.
If you read our projected standings article, you may recall that we projected many close races. Our projections are based on the average results from 100 simulated seasons, and it's unlikely that any one season would feature so much competition. Still, it does indicate a level of parity that we haven't seen in a while. (By the way, it came as a pleasant surprise when Tom Verducci mentioned our work while writing about parity in the April 11th edition of Sports Illustrated.)
As has become our custom, we'll come back to this topic after the season. When the final standings are known, we'll assign accuracy scores to all of the predictions in our database and rank them.
For 2005, however, that exercise may be less meaningful than usual. For example, our simulations had (a) New York and Boston within one game of each other, (b) a three-way tie for second in the AL Central, (c) all four AL West teams within five games of each other, and (d) four NL East teams within eight games.
Because so much can happen between now and October, we cannot say with confidence that New York will finish ahead of Boston, that Cleveland is the best bet to finish second in its division, that Oakland will win the West or that the Mets will finish fourth. All of these results are well within the margin of error for this type of exercise.
For example, if the real-life AL West finishes as follows ...
Los Angeles 87 75 .537 -
Seattle 83 79 .512 4
Oakland 81 81 .500 6
Texas 77 85 .475 10
... our accuracy score for that division won't be very good. And yet those standings would confirm much of what our simulations told us -- that the Angels aren't head and shoulders better than the others, that Seattle should bounce back in a big way, and that the Rangers probably can't replicate their success of a year ago.
Nevertheless, we've been using the same method to assess the accuracy of predictions since 1998, and we're not going to change just because we're projecting a lot of close races.
Getting back to the purpose of this article, it's always fun to see how our projections differ from others you might have seen, so let's take a stroll through the divisions and see how the baseball world looks to these experts.
For this discussion, I'll leave out the over/under lines, past standings, and salary ranks, focusing instead on the 56 predictions that represent the views of a publication, an individual, or the consensus of a group of individuals.
The vast majority see the Yankees finishing ahead of the Red Sox. All six of the Boston sportswriters picked New York. And of the 16 who put Boston first, 14 are from Baseball Prospectus. Other than BP, only Baseball America and the Dallas Morning News picked the Red Sox.
(By the way, the BP site has lots of predictions to choose from. One is from their PECOTA projection system. One is from Joe Sheehan's BP Today column. The others are from a poll of BP staffers. We included each staffer individually plus the group average.)
Everyone seems to think there's a great divide between the top and bottom of the AL East. Nobody had any of the remaining three teams cracking the top two. All but 8 had Baltimore third, with 7 of the others picking Toronto for that spot, and one lone voice (BP's James Click) going for Tampa Bay. Click was the only person to pick Baltimore for last place, but 11 expect Toronto to repeat in the cellar.
In the AL Central, only two forecasters have the current leader, the White Sox, winning the division. Six picked Cleveland, with the other 58 giving the nod to the Twins. Kansas City was a unanimous pick to finish last. Overall, these 56 predictions portend a Min-Cle-Chi-Det-KC finish. That's consistent with our simulations, though our results had the middle three bunched so closely together that the order of finish cannot be predicted with a high degree of confidence.
The AL West is the first division where the consensus differs from our simulation results. Just about everyone other than Diamond Mind and Baseball Prospectus picked the Angels to finish first. Keith Woolner of BP picked the Rangers (their only vote), while 13 went for Oakland. BP accounts for 10 of the 13 Oakland picks.
The collective wisdom of this group says the finish will be LA-Oak-Tex-Sea, which mirrors the 2004 finish. If we were being totally scientific, we'd have to say it's too close to call. But that wouldn't be any fun, so we'll stick with the simulation results, which were Oak-LA-Sea-Tex.
The NL East is like the AL West in three ways -- it was tightly bunched in our simulations, our simulations disagree with the consensus, and the consensus matches the 2004 final standings.
Of the 56 predictions, 29 picked Atlanta, 10 picked Florida, 4 took the Mets, and 13 the Phillies. There's a strong sabermetric bias here, as most of the Philly votes are from BP and Diamond Mind. Everyone has Washington in the basement. Even though Philly got more first-place votes than Florida, the group thinks Florida will finish second, with the Phillies third and the Mets fourth.
It's worth noting that although we're among those projecting a last-place finish for the Nationals, we have them being more competitive than most. They averaged 79 wins in our simulations. Among the other eight projections that included wins, the range was 66 to 74.
It's also worth noting that Florida is one of my sleeper picks. Although they finished third in our simulations, they were only seven games off the pace, and it's not hard to imagine them having a breakout season if their young pitchers can add consistency to the flashes of brilliance they've shown over the past two years.
The NL Central was the only division with a runaway winner in our simulations, with the Cardinals averaging 103 wins, the Cubs 83, and the other four clubs under the .500 mark. Others agree that it's a two-team race, as all 56 picked St. Louis or Chicago to win the division, and only a few intrepid souls picked Houston to finish second. But 14 picked the Cubs to win the division, so it's clear that not everyone see this is a walk in the park for the Cards.
The bottom end of the division is a little more interesting. There appears to be broad agreement on the Pirates, with all but two predictions putting them last or second-last. But the Brewers were picked to finish anywhere from 3rd to 6th. All of the third-place votes came from the BP crew, but 30% picked them fourth, and most of those picks were from outside the BP family.
The consensus nearly matched the Diamond Mind simulations. We agree on the first four places (StL-Chi-Hou-Cin), but we've got Pittsburgh two games ahead of Milwaukee, while the consensus has the Brewers in fifth. I can't say that I have a lot of faith in this aspect of our results, mainly because a two-game spread is too small to be meaningful. I won't be the least bit surprised if Pittsburgh finishes in last place.
The NL West shows the biggest gap between the average prediction and our simulation results. Our results were LA-SF-SD-Col-Ari, while the 56 predictions netted out to SD-SF-LA-Ari-Col.
I'm not entirely sure why, but I have more confidence in our NL West results than for some of the other tightly-contested divisions. I'm not yet sold on the Padres, and we may have given Barry Bonds too much playing time in our simulations, though only time will tell on that front. All I know is that when the Dodgers popped out of our simulations as the front-runner, I didn't break out in a sweat.
The bottom of the NL West is another too-close-to-call situation, with only two games separating Colorado and Arizona. This one does make me nervous. I can't figure out what Colorado's management is trying to do, and I could easily see them finishing with the NL's worst record.
It's too early to know whether we'll continue to see the level of parity we saw in the simulations and the first three weeks of the 2005 season. If that keeps up, it'll make for a fascinating six months of baseball, with several multi-team division races and just about everyone having a shot at the wild card.
In the April 28 game between Oakland and Chicago, the White Sox were forced to start Joe Crede at shortstop and Chris Widger at third because of injuries to three infielders. A customer asked whether this would entitle Crede to be rated at short and, if so, what those ratings would be (assuming this was his only game at the position in 2005). Our general rule is to rate a player at any position where he starts at least one game. In this case, however, we're very likely to make an exception. Crede started at short because everyone else was hurt, not because his manager considered him a viable shortstop. In the nine years since Crede was drafted, he has never played a position other than third base. Not in the majors. Not even in the minors. One emergency start doesn't make him a shortstop.
This was also Widger's first game at third base and Jermaine Dye's first at short. (Dye played short in the bottom of the ninth after Crede was ejected.) Neither Widger nor Dye had previously played a single inning at those positions in the majors or the minors.
Because DMB gamers can use players out of position in an emergency, we already have this situation covered. As a result, we don't feel compelled to rate these three players at these positions. That could change as the season unfolds, so we won't make any final decisions until November.By the way, Crede was ejected for arguing what I felt was a very good call by the home plate umpire. On an inside curve, Crede flinched momentarily and then leaned forward and down to get his shoulder in front of the pitch. The ump ruled that Crede wasn't trying to get out of the way and refused to award him first base. I'd love to see this call made more often.
- Tags: Newsletter 2005