Predicting HOF election - Smith-Downey model

[Paul Wendt]

See today's Wall Street Journal, "A Computer Cracks the Cooperstown Code"
http://online.wsj.com/article/SB124864577360682129.html

It appears that the model does not concern election outcomes directly. Clearly it is designed to fit current HOF membership status.

>>
Of the 1,592 players considered by their study -- anyone who retired
between 1950 and 2002 and met several other criteria such as having
played 10 years in the majors -- the model was able to accurately
identify whether they had been elected 98.7% of the time.
<<

Is that good? Evidently their model or rating misclassifies 20 or 21 players. Before yesterday's inductions there were 83 Hall of Fame players with major league finales 1950 to 2002 inclusive. (That counts Roy Campanella, not Monte Irvin or Negro Leagues stars with less major league playing time.)

Perhaps this note will attract attention from authors of relatively simple ratings intended to identify Hall of Fame players, Ron Henrich for example. How many recent players do they misclassify on both sides of doorway?

>>
Before the 2009 class was voted in, the model gave Mr. Henderson a
97.2% chance of being elected. On the flip side, Mo Vaughn had a 1.3%
chance.
<<

Those would have been good predictions of their shares of Yes votes (Henderson 94.8%, Vaughn 1.1%). I doubt there would have been many bettors on the conservative sides of odds that correspond to 99.72% and 0.13%, about 1-400 and 800-1.

This work evidently models the HOF membership status of players from a timespan that runs up to the present; that is, including the latest players who have been eligible for a single election. (I doubt that the model concerns election outcomes directly.)

Hall of Fame induction may happen 50 or 100 years after a player's major league finale, so there is a more than ordinary challenge fitting the membership status of the mid-century and the recent players at once. Two recent developments must help: during the last few decades it has been more common to elect eligible players soon; during this decade almost no one has been elected very late, by the veterans committees rather than the baseball writers. Fifty years ago, we know, there were many eligible players in the process of being elected, so to speak. Today that may not be true.

[Brad Harris]

Perhaps this note will attract attention from authors of relatively simple ratings intended to identify Hall of Fame players, Ron Henrich for example.
You know, this seems like just the thing he'd be interested in, too! Where is Ron? Haven't seen him for a while. :(

[Jsquared83]

14% for Kent? 47% for Mussina? 54% for Edgar!? Idk about that.. Chipper's # seems a little low as well..

[Freakshow]

14% for Kent? 47% for Mussina? 54% for Edgar!? Idk about that.. Chipper's # seems a little low as well..The system's predictive accuracy hinges on whether or not the factors that have worked for past marginal selections will continue for future marginal selections. These factors do, indeed, evolve over time.

Having said that, you never can tell quite where a player will end up. Twenty-five years ago, "nobody" doubted that Jim Rice, Steve Garvey and Dale Murphy would zoom right into the HOF.

[jjpm74]

The system's predictive accuracy hinges on whether or not the factors that have worked for past marginal selections will continue for future marginal selections. These factors do, indeed, evolve over time.

Having said that, you never can tell quite where a player will end up. Twenty-five years ago, "nobody" doubted that Jim Rice, Steve Garvey and Dale Murphy would zoom right into the HOF.

Maybe so, but Jeff Kent as a 14% HOFer seems out of place. I don't expect him to get in his first year, but I'd be surprised if he doesn't eventually get elected during his 15 year run.

[Paul Wendt]

You know, this seems like just the thing he'd be interested in, too! Where is Ron? Haven't seen him for a while. :(
You don't cross paths at the Cincinnati public library?

"henrich" frequently promoted CWA as a representation of rankings implicit in the roll call of HOF members --20th century major league players, iirc. Now I wonder, how many players with 1950 to 2002 finales does it mis-classify? --the false positives plus the false negatives. I wonder that also about the HOF Monitor by Bill James.

Perhaps some others have tried to explain (model, fit, retrodict) Hall of Fame membership more systematically than henrich or BJ, still using elementary methods rather in contrast to the avant-garde reported today by WSJ. Knowing how many players some other ratings mis-classify is really part of understanding the 20 or 21 misses (1.3%).


The system's predictive accuracy hinges on whether or not the factors that have worked for past marginal selections will continue for future marginal selections. These factors do, indeed, evolve over time.

Yes the factors do evolve. Because the authors limit their scope to 1950-2002 finales, it is generous to understand the yield of their advanced methods by comparing error rates with more ambitious simple rankings. Part of any gain inaccuracy must be due to the narrow scope rather than to the advanced methods.

The marginalist point is a good one. By design such a model should be good at identifying the close calls but it may be no good at distinguishing 90% from 99.9% or 10% from 0.1%.

[Jsquared83]

The system's predictive accuracy hinges on whether or not the factors that have worked for past marginal selections will continue for future marginal selections. These factors do, indeed, evolve over time.

Having said that, you never can tell quite where a player will end up. Twenty-five years ago, "nobody" doubted that Jim Rice, Steve Garvey and Dale Murphy would zoom right into the HOF.

GaIn the midst of their peaks, both Rice and Murphy looked like no brainers Im sure but Kent and Mussina are both retired and Chipper is a first ballot guy if he never plays another game. They weren't trying to project guys mid-career. As of 1983 Dale Murphy was one of the best players in the game. By 1993, one wondered if he'd even break 50% on the HOF ballots.

[Jsquared83]

Maybe so, but Jeff Kent as a 14% HOFer seems out of place. I don't expect him to get in his first year, but I'd be surprised if he doesn't eventually get elected during his 15 year run.

I'd be surprised if it takes him more than 3 years. Those numbers are hard to ignore, especially from second base. Yes, below average in the field but no one's penalizing Mike Piazza for his futility behind the plate.

[NJRob65]

I'd be surprised if it takes him more than 3 years. Those numbers are hard to ignore, especially from second base. Yes, below average in the field but no one's penalizing Mike Piazza for his futility behind the plate.


Although i do think Kent will be elected by the writers, keep in mind that it took Ryne Sandberg 3 years to be elected, and most observers would consider him to be a superior all round to Jeff Kent.

I'm not sure Piazza was as bad behind the plate as most of his detractors think.

[jjpm74]

I'm not sure Piazza was as bad behind the plate as most of his detractors think.

Maybe Piazza could call a good game; I honestly don't know, but it was painful to watch him try to throw out base runners attempting to steal. It felt like half the time the ball would end up in the outfield during his NY Mets years. Fortunately, his offense more than made up for it.

[NJRob65]

Kent 's 14 % chance is way too low, 50% would be on the low side. I would expect him to be elected., although until his fifth or sixth time on the ballot. I think Edgar Martinez's stated chances are pretty accurate, perhaps a bit too generous, if he put up the same numbers while primarily playing third base, he'd have a much greater chance of being elected, if he played first base in most of his games, he'd have a better chance than he does as a candidate who appeared in most of his games as a DH. Paul Molitor appeared as a DH in a plurality of his games, 1174. He played the field, at several positions, in 1415 games. No predominate designated hitter has ever been elected to the Hall of Fame.

I would think both Mussina and Schilling would have about a 50/50 chance of being elected by the writers.

Guerrero, Jones and Hoffman are all locks. Helton's 22% chance, seems a bit high too me, but they do acknowledge him as a long shot.

[Jsquared83]

Kent 's 14 % chance is way too low, 50% would be on the low side. I would expect him to be elected., although until his fifth or sixth time on the ballot. I think Edgar Martinez's stated chances are pretty accurate, perhaps a bit too generous, if he put up the same numbers while primarily playing third base, he'd have a much greater chance of being elected, if he played first base in most of his games, he'd have a better chance than he does as a candidate who appeared in most of his games as a DH. Paul Molitor appeared as a DH in a plurality of his games, 1174. He played the field, at several positions, in 1415 games. No predominate designated hitter has ever been elected to the Hall of Fame.

I would think both Mussina and Schilling would have about a 50/50 chance of being elected by the writers.

Guerrero, Jones and Hoffman are all locks. Helton's 22% chance, seems a bit high too me, but they do acknowledge him as a long shot.

I'd rank Mussina higher than Schilling. Schilling's career is way more inconsistent, albeit a higher, scattered peak. I'd put him around 50/50 and Mussina around 70/30. 270 wins with his outstanding winning % in this era is hard to ignore. Retiring on top of his game will also help him out.

Im not sure Hoffman's a first ballot guy. Def deserving though.