My All-time A's project brought up an interesting aspect to the eternal question, how do you compare stats across the eras?

Plenty of stats, like OPS+, use league average (or current replacement level) as a baseline and describe a player's accomplishments in terms of that. But the question came up - has it been easier to dominate the game in certain eras, especially before integration - than after?

In an attempt to shed some light on that question, I've come up with what I call OPS++. It's a pretty simple stat, to derive it all you do is take the player in question's OPS+, subtract 100 and divide it by the rolling 5 year average of the average of the OPS+ for the top 10 in the majors over those years, minus 100 and multiply by 100.

Take Eric Chavez, circa 2003, for example.

In 2003, his OPS+ was 132.

The rolling 5 year average top-10 OPS+ was 173.

So Chavy's OPS++ was 44.

For another example, let's take Reggie Jackson, circa 1968.

In 1968, his OPS+ was 137.

The rolling 5 year average top-10 OPS+ was 168.

So Reggie's OPS++ was 54.

For the all-time record in OPS++:
Bonds '02 275/176
Bonds '01 262/174
Bonds '04 260/171
Ruth 1920 255/163
Ruth 1921 239/167
Ruth 1923 239/166
Williams 1941 235/161
Williams 1957 233/161

OPS++
Bonds '02 230
Bonds '01 219
Bonds '03 225
Ruth 1920 246
Ruth 1921 207
Ruth 1923 210
Williams 1941 221
Williams 1957 218

So, in order, OPS++'s top 8: (I think
Ruth 1920 246
Bonds '02 230
Bonds '03 225
Williams 1941 221
Bonds '01 219
Williams 1957 218
Ruth 1923 210
Ruth 1921 207

So, while Bonds dominates OPS+, when his the ease of domination in his era is factored in, Ruth takes the top spot and the top 8 is jumbled a bit.

So why a rolling 5 year average, you ask?

Well, mostly to get a reasonable sample size. Comparing an OPS to 50 datapoints is a lot more relevent than 10. Also, any real trends (such as integration, a live ball, etc) that would affect this would last for more than one years and this helps smooth out annomalies that can occur from year to year.

Top 10 avg OPS+ by decade
'01-'10 160
'11-'20 163
'21-'30 166
'31-'40 163
'41-'50 158
'51-'60 160
'61-'70 164
'71-'80 159
'81-'90 156
'91-'00 169
'01-'05 173

OPS++ for various OPS+ values over the course of history (1901-2005).

And, of course, I always appreciate a recommendation.