Thursday, March 24, 2005

2005 Predictions Wrap-Up

It was long. It was tedious. And frankly, I'm sure many of you found it too boring to read. But it's done. I've looked at the NL Central, attempted to figure out who's playing for who and where, and have made my fearless predictions for 2005. Here's a bit of a correction to my predictions, along with some basic re-cap.

"Lucky Wins"

As anyone who read any of my articles knows, I used Win Shares to predict the number of wins for each team, but further used the teams 2004 Pythagorean wins to adjust for the 2005 season. For more information on Pythagorean wins, see this article. Basically, it's a simple equation that predicts the number of wins a team should have based upon nothing more than runs scored and runs allowed.

By using a team's 2004 "expected wins" total, I adjusted their 2005 wins by the opposite amount. So, for example, the Cardinals had 5 more wins than expected last year, thus I subtracted 5 wins from my prediction total in 2005. Why? In general, when a team is lucky in one season, they are going to be unlucky in the future, and vice versa. Bill James calls this tendency "the Plexiglas principle" in one of his baseball abstracts. However, it bothered me during my predictions that I was using a 1 to 1 ratio in making my adjustments (especially in the case of the Reds, who had 9 lucky wins last year.) That didn't seem to be the best method of adjusting for luck.

In order to adjust my "luck" corrections, I did a quick and dirty study. I simply looked up all win totals in the NL Central since the 2001 season, including expected wins. That gave me a total of 18 seasons in which to look at. Granted, that's a small sample size, but I only had limited time. Using that data, I simply compared predicted "lucky/unlucky" wins for each of the 18 seasons in question and looked at how close the predictions were.

Using a 1 to 1 ratio, the results weren't too bad. Over the 18 seasons, the median error was 2.5 games. The prediction was exactly right 1 time, and was within 4 games 11 times, or 61% of the time. Unfortunately, there were 4 seasons in which I was off by 10 or more games.

From there, I simply started manipulating the prediction values by simple methods. I tried dividing by various numbers and fractions, raising them to different powers, taking the square root, etc. When the dust cleared, the best correction factor appears to be to simply divide last years "lucky" wins by 3 to adjust for this year.

When dividing by 3, the median error dropped to 2 games. But more importantly, the average error was reduced from 4.7 to 3.4, as the huge errors were reduced by quite a bit. The prediction was exactly right three times, and was within 4 games 13 seasons, or 72% of the time. And most importantly, the prediction was off by 10 or more games only 1 time.

Naturally, there could be much more work done on this topic. For one thing, 18 seasons is not much to go on. Ideally, you'd go back at least 10 seasons, and use every team in baseball. And even if you find a great relationship - well, luck is called luck for a reason. You never know when a team is going to have things fall their way one year, after nothing went right the year before. Bullpens are going to improve, thus helping a team win more one run games, offenses will get more consistent, new coaching staffs will teach new things, etc. That's why they play the games, as they say.

Anyway, my predictions can now be modified slightly, by taking the luck factors I used and dividing them by three. When you make those final adjustments, my modified predictions look like this.

NL Central

Cardinals 98-64
Cubs 92-70 (6 Games Back)
Astros 89-73 (9 Games Back)
Reds 80-82 (18 Games Back)
Pirates 79-83 (19 Games Back)
Brewers 68-94 (30 Games Back)

A couple of notes. I think the Cardinals win total is a bit high using this adjustment, but it makes the win totals for the Cubs and the Reds look more in line with reality. I also think the fact that this prediction shows 5 teams within 19 games of one another shows the reality of the division, which is that is should be better than last year.

Here is the breakdown of the division based upon various aspects of the teams. Listed below will be the Win Share predictions that I made in my NL Central reports, broken down into Starting Pitching, Bullpen, Starting Lineup, and Bench.

Starting Rotation

Cubs – 79
Astros – 66
Cardinals - 55
Pirates – 50
Brewers – 48
Reds – 34

This makes sense to me. The Cubs have what is far and away the best rotation in the NL Central. What's more, that will likely hold true even without Wood or Prior for a chunk of the season. (If both of them miss time, it's another story.) The Astros in 2nd also seems sound, with a nice 1-2-3 punch at the top of the rotation.


Pirates – 41
Reds – 37
Astros – 37
Cardinals - 36
Cubs – 27
Brewers – 20

This one is a bit surprising to me. While I like the Pirates bullpen, I'm not sold that it's the best one in the NL Central this season. I'm also not sold that the Astros 'pen is good enough to be near the top, but Brad Lidge is going to make it very formidable. The Cubs and the Brewers being back from the pack is very realistic.

Starting Lineup

Cardinals - 179
Reds – 151
Cubs – 137
Astros – 128
Brewers – 115
Pirates – 113

I'm completely fine with this. The Cardinals lineup is the class of the NL, let alone the Central. The Reds have a very strong lineup as well, which could challenge the Cardinals if Griffey and Kearns stay healthy, and if Casey can repeat last year. And the Cubs have a great infield, but their outfield is going to drag them down by a bit.


Astros – 36
Cardinals - 36
Pirates – 32
Reds – 27
Cubs – 27
Brewers – 20

I'm not at all convinced that the Cardinals have the best bench in the division. I'm not certain that this is an endorsement of the Cardinals role players, so much as it's an indictment on how bad the rest of the league is in depth.

All in all, my predictions were done in fun, and to give me (and hopefully some of you) a better understanding of the strengths and weaknesses of the division this year. I do, however, find it kind of amusing that it took me six weeks and several hours of work to come to this conclusion: The Cardinals should win the NL Central, with the Cubs having the best shot at catching them. I could have told you that in January without going through these gyrations!


At 1:52 PM, Anonymous 84 said...

artial Features List to Consider:
easy to use
easily add audio
no need to know html
no experience required
records audio for you
built-in editor
scrolls text for you
user-adjustable font size
brand any element (text, graphics, URLs, banners)
easy to learn
video tutorials show you how
easily combine text, audio and graphic all-in-one
powerful WYSIWYG Text Editor
import pre-recorded voice or music file

At 7:55 PM, Anonymous 6 said...


Post a Comment

<< Home