>> I actually just ran a correlation for Opening Day Patroll and
>> percentage for 2000. It clocked in at a whopping .327.
>> If we did forecasts from it, the Yankees projected to have a
> .535 win %
>> and the Twins projected to have a .463 win% with the standard
> error being
>> Here's a scatter graph of Opening Day Payroll against eventual
>> season winning percentage:
> Can you create another one using payroll rank, rather than salary
> dollars? That's the other way people sometimes try to make the
> case for a causal relationship.
My next maneuver will be to calculate the change in ODP (opening day
payroll from here on out) from 1999 to 2000. Then I'll test the
correlation between change in payroll and change in win%.
There will still be correlation/causation questions there, an increase in
winning will cause an increase in payroll the next season (arbitration,
extra revenue), but it should filter out some of the problem with
causation and correlation.
The only other problems I can think of with that is that there would be a
bias for good teams in 1999 to decline and poor teams in 1999 to improve
so that would somewhat confound any relationship that might exist between
increased payroll and increased winning.
As far as correlation by rank in 2000 goes the coeff. is basically the
same at .326 (.327 was the other).
An any event, there is a graph of the ranks at:
It's got a little more color than the last to differntiate it, and also
since they are ranks the points are backward with the high payroll high
win teams in the southwest corner and the low payroll low win teams in the