Currently I am keeping points for my home tourney by assigning the following points:
So a 6 player, 2 payout game would be: 9, 7, 4, 3, 2, 1
The problem in the long run is that there is clear separation between my good players and "Weaker" players.
Is there a better system to score from game-to-game and is there a good system of establishing a handicap overtime?
I'll give a known ranking procedure and a handicapping example further down.
Pokerstars awards the points to the top 15% of players in a tournament based on this calculation:
Points = 10 * [sqrt(n)/sqrt(k)] * [1+log(b+0.25)]
Where:
Here's an example of theirs:
A player takes third place in a $20 tournament with a field of 150 players.
Here:
n = 150, k=3, b=20
The total points awarded to this player are:
= 10 * [sqrt(n)/sqrt(k)] * [1+log(b+0.25)]
= 10 * [sqrt(150)/sqrt(3)] * [1+log(20+0.25)]
= 10 * sqrt(50) * (1+1.31)
= 10 * 7.071 * 2.31
= 163.09 (rounded up)
You could expand the awarded points to cover whatever percentage of the field you like.
If you wanted to you could use this scoring system to establish a handicap by judging how players Actually perform vs how you Expect them to perform at random, and then adjust the points they receive post-tournament taking into account their handicap. For example:
A player takes second place in your 0 Dollar/Points tournament with a field of 10 players. You award the top 30% of players some points. Here (aside from the answer, rounding all calcs to 3 decimals):
Actual:
n=10, k=2, b=0
The total points awarded to this player are:
= 10 * [sqrt(10)/sqrt(2)] * [1+log(0+0.25)]
= 10 * 1.147 * 0.398
= 4.56 (rounded up)
Expected (the same except for k, which is the average random finishing place):
n=10, k=5.5, b=0
The total points awarded to this player are:
= 10 * [sqrt(10)/sqrt(5.5)] * [1+log(0+0.25)]
= 10 * 1.348 * 0.398
= 5.36 (rounded up)
You then take his Actual points, divide them by his Expected points (random) and you can see how much better (or worse) the player is against random chance. Here:
Handicap:
= Actual/Expected
= 4.56/5.36
= 0.85 (rounded up)
or the player is about 0.15 (15%) better than random. You could adjust his score after the game based on this, giving or taking some points away.
That was much longer than I anticipated :D I hope it gives you some good ideas.
I think your system is good when is separating good players from weaker players, then in other scale separate good players from totally weak players.
From comments i see you have payouts and motivation issues to good weaker players, is good to see that in ranking i think you should change your payout system (after 6 months) because in that is problem not in ranking system.
Other problem you have when you paying relative to points, for example 1 point = X USD. Then generally you merge ranking system with payout structure, this is poor idea, you can separate them back to easier manipulate payout system without changing ranking system.
In other way, you have complicated algorithm, and for win you have same amount of points than for all players in table when someone will be kicked, this is not 1:1, from that you have high points variance + exploiting system, next for what you have point for cashin? Many poker leagues makes ranking based on what place finished only, and from for example 100 players only 25 getting any points rest 0, and from 10 to 25 is very low amount of points.
It seems like your current system is working as a point system to define and reward consistent skill. As far as a way to handicap, couldn't you use the reverse of the point scale to determine handicap? I'm not sure what you would do with that, but it seems it would work fairly well.
If you are seeking to use a system that lets players start out as poor players but finish in the top, consider only counting some number of top scoring tournaments. Consistent players will finish at the top still, but players who had several bad games when they were first learning could still easily reach the top of the field overall, if they have improved to consistently finish strong in enough events.
On Grinderschool, we count the top finishes from what would be about 1/3 of our tournaments to determine the player of the year in our tournament series. That has worked pretty well.
