Just a thought on adjusting for 162 game schedules:

This doesn't actually apply much to the 19th century, but moreso to after the inception of the 154 game schedule,

I was reading the above article on time lost due to labor stoppages, and was noticing how lots of players lost about 50-100 games due to the two big strikes in 1981 and 1994-5. It seems that most long-time players whose career began after 1961 would be effected by one or both of them.

Just picking some names from different eras, here are the top 6 all time in stolen bases, counting only seasons in which they were at least semi-regulars (min. 20 games):

Post 1961:

MattB, I think it's a good thought. Maybe what we should do, is make adjustments for 1981, 1994 and 1995 like we do for the 154-game seasons, to bring those seasons up to 162?

I used 100 games as a threshhold for "regular" status. I think that is more reasonable than 20. Following are the average number of games played in seasons in which there were at least 100. They're a lot closer than using 20. That's because the three contemporaries played a total of 12 seasons at less than 100 games, the old timers just 9 (one fewer per player).

Henderson 2674 games in 19 seasons thru 2000 (140.7)

Matt, Craig and Mark:

I've been using 162-game seasons for everything I've done, as far as my offensive W-L records you've seen scattered on Primer over the last year and a half (like Belle/Kiner, Trammell/Ozzie, etc.).

I think that's the way to go, because over the long haul, that will be the standard, once we've 'caught up', we won't have to adjust every year going forward.

I've adjusted for the strikes in everything I've done, and I plan to do that here as well. Everyone should be on a level playing field, at least as far as the numbers go.

If you then want to decide that this era is tougher, etc., you can do that from there. But I think first we have to set up 'common denominator' which should be IMO, 162-game seasons. We adjust for things like the league and the park, it seems obvious to me that the schedule should be a adjusted for as well. Whether it's 118 games in 1918, 108 in 1981, 154 in 1954, or 162 in 2002 (thank God!), 1 pennant, 1 World Series is the end result (except for 1994).

Because of the 'pennant' as the cutoff, I very strongly believe that the accomplishments of each player should be weighted equally towards each pennant that was pursued. If not, you introduce a bias towards player's whose careers began before the schedules grew. You have a bias towards players who had their best years in shortened seasons (or for players who had their worst years during those years). It's pretty simple to fix this (adjust the numbers) so I think it's a no brainer to do it.

Marc, I don't think it's fair to say they could have chosen not to strike. It was a condition of the time, like the wars, or the shorter schedules of the 1800's. It was not a reflection of a player's ability or skills, like an injury or early death would be.

It's an oversimplification, but the early strikes were for a pretty good cause (owners were making money hand over fist and players weren't seeing too much of it, I think Mattingly made $35K as a rookie in 1983, for example). Pretty much every strike since the players have leveled the field has been forced by the owners not agreeing to not lock the players out if they didn't strike.

I realize this leveling of the playing field only applies to the 'number crunchers' out there, obviously those that don't crunch won't be affected by this either way.

"I realize this leveling of the playing field only applies to the 'number crunchers' out there, obviously those that don't crunch won't be affected by this either way."

I, for one, have been known to get soggy in milk.

First, I guess I'm the last to discover that scruff and joe are one and the same? Is this accurate?

Second, I really agree with the pennant as the fundamental unit of measure here. I really love Bill James but the one thing I don't agree with is the combination of the short seasons and the timeline adjustment creating a tremendous bias against 19th century players, and especially the earlier ones.

Which leads to my third question. Is anybody doing WS or any other further analysis on the old NA? Say what you will about the NA or even fair-foul hit, no second sacker except Hornsby and maybe Morgan ever had the offensive impact of Ross Barnes. If the fair-foul hit disqualifies him from HOF consideration then how do you justify Burleigh Grimes? And there's no SS through Honus Wagner as dominant as George Wright. I'd like to see some further analysis of these guys. Of course I could do it myself but I too get a little soggy when the milk is down.

Marc -- NA Win Shares on my to do list. I've developed a spreadsheet that calculates WS (it's pretty nasty, like 6 worksheets, about 2 MB w/out any data) it's just a question of finding some time to go back and fill in the blanks.

Actually, there is one thing that's holding me up: A decent NA Runs Created/XRuns formula. Once I find one that I'm comfortable with (if anyone out there has one, please share it!) it's just data entry.

Matt/Marc -- I guess the 'number crunchers' comment wasn't necessarily for those who actually bust out the calculators, it was intended more for those people that don't even like to use the numbers crunched by others. Sorry for the confusion.

Almost forgot, you are correct Marc, I am JoeDimino as well.

Eventually I'll grow up, but I'm trying to stave that eventuality off as long as possible. When I first started posting on Sean's Outside the Box (back when Primer was still in the womb) I wasn't sure if I'd post something dumb, so I wanted to remain annonymous until I got comfortable, being new to the whole weblog thing. So I used my college nickname. I just never bothered changing it after that.

Joe -- you can use the 1876 RC formula for the NA. The various correctors make sure the number of runs a player creates adds to the team total of runs scored. In this case, it just adjusts everything upwards.

Thanks Charles! What would your confidence in the accuracy be?

Now that I remember it, you had mentioned this to me in an email a few months back, it had slipped my mind. Thanks again.

This is somewhat unrelated, but more and more I am coming to the conclusion that Loss Shares are doable. I even have an idea of how they would work.

*** The following is not tested in any way. Many levels are theoretical. ***

Loss Shares

* Loss Shares, for a team, are equal to 3x Losses.

* The basic idea of Loss Shares is "Win Shares in reverse."

* We shall handle the issue of negative claim points different. When first figuring Win Shares and Loss Shares, figure Win Shares with the negative total, thus giving out negative Win Shares and Loss Shares. Once we have figured both, we then subtract negative Win Shares from Loss Shares and subtract negative Loss Shares from Win Shares. We readd the two columns, and, using the refigured Win Shares/Loss Shares as claim points, reapportion the team total of Win Shares.

* Proportioning Loss Shares between the offense and the defense is the reverse Win Shares's method. Claim points for the offense is equal to LgR*Park*1.48-R, and for the defense it is OR-LgR*Park*0.48. Loss Shares are distributed proportional to claim points.

* Proportioning Loss Shares between the pitchers and the fielders is mostly the reverse of Win Shares's method. (I am keeping my musings in Bill James's context, even where I do not like the context and even have proposed fixes.) All adjustments apply in reverse -- if home runs allowed reduced the fielders' Win Shares claim points, it will increase the fielders' Loss Shares claim points.

The exception is the strikeout adjustment. This is the same adjustment as in Win Shares. Teams with many strikeouts will have fewer Win Shares and fewer Loss Shares going to the fielders.

* When figuring claim points for fielders:

<pre>

I have no idea on the accuracy. You would be best to use generate uncorrected Runs Created values for each NA team, then run a correlation between that and their real Runs Scored values. If the correlation is high (and we're talking r higher than 0.80, preferably higher than 0.90), then this would work fine, as I suspect it would.

Speaking of pennants as the relevant unit . . .

Take a look at today's Baseball Prospectus.

Wolverton uses his "pennants added" metric from BP 2002 to look at the top "pure hitters" not in the Hall.

Nice to see Ken Singleton so high on the list.

Is this article on the web anywhere? I was thinking of flipping it to a friend, but can't find it.

Not on the web. Sometimes you just gotta buy the book.

I'll let him know, thanks, Matt.

Wolverton's pennant-impact values are very similar to values I derived awhile before his research was published. The idea is to reflect the belief (based upon pennants) that a player with consecutive seasons of 12 & 8 WAA is more valuable than a player with two 10 WAA seasons.

Straight adding of seasonal values ignores the pennant-value issue. So I suggested that we take every seasonal value to the 1.25 exponent and then add across seasons. I experimented with different exponents and I honestly cannot remember what exponent I settled on (I think it was 1.25).

For what it is worth, I plan to utilize this approach in evaluating HoM candidates, and encourage others to consider it as well.

Sorry, WAA is Wins Above Average.