from Amazon
Tuesday, March 03, 2009
Farm Value Rankings
Victor compiles the list from 4 different sources and gives us his overall rankings.
For each list, I took the No. 1-rated prospect and gave him 100 points, 99 points to No. 2, 98 points to No. 3 etc.
Little food for thought: a better way is to have it follow a more logarithmic scale. After all, the difference between the #1 and #11 is not the same as the difference between the #71 and #81.
For example, if you use this equation:
4.65-LN(Rank+3))*56.63
Will give you the following for the top 10 and bottom 10 players:
Rank Tango Victor
1 185 100
2 172 99
3 162 98
4 153 97
5 146 96
6 139 95
7 133 94
8 128 93
9 123 92
10 118 91
...
91 6 10
92 5 9
93 5 8
94 4 7
95 4 6
96 3 5
97 3 4
98 2 3
99 1 2
100 1 1
The average for both of us is exactly 50.5. It’s just that in my case, they are more spread out: standard deviation of 43 for me and 29 for Victor.
Now, whether you want to do “Rank + 3” or something else all depends on how much spread you “should” have. And of course the slope and intercept values would change so that you still get a mean of 50.5.
Just something to think about…
The only down side to that is that players whom the lists don’t agree on, will be overrated. Under Victor’s numbering, getting a 1,2,3 is the same as 2,2,2. Under yours, the 1,2,3 player is rated higher. I’m not sure whether reliability or upside is more important in a prospect. Probably upside, so maybe your way is the best.