math wizards: fairest way to score poker players ranks

I play Texas Hold 'em in a weekly on-line tournament (play money). I am in charge of keeping score of players ranks. This is the system I've been using:

If there are ten players in a tournament, everyones score will look like this:

player 1- 1pt. player 2- 2pts. etc. If there are only eight players, the scores look like this: player 1- 1.25pts player 2- 2.5pts. etc.

Then every week, I average everyones scores and see who's doing the best (the lower the score, the better). There have been five games this quarter. I came in first in this quarter's first game, and John came in second. In three of the four following games John has done slightly better than me. These are our rankings so far:
me: 5.72pts.
John: 5.36pts.

Now I thought this was a pretty good way of keeping score, and then I found this article on how the North Jersey Poker League keeps score. They use what they call a Natural "Logarithm Rankings" system to keep score.

So, I tried using their system to see if the results would be different, and they were. I got:
me: 57.12%
John: 54.77%

It doesn't seem right that I should have a higher score than John when I only placed one level above him in two tournaments and he's done better than me in three others. Is the NJPL calculating using the wrong formula? If they aren't, is my scoring system better?