Wednesday, April 21, 2010

Thoughts on Alternative Systems

"I suggest we stick with what we have for the rest of the year. However, I will implement Klokkhammer's point about showing the number of games played since this only effects the way the results are presented and not the way they are calculated."

I've done some looking around at various systems and here is what I have found.

1. Many poker leagues use the points system that we currently have:
Points = p - (f-1)
p is the number of players
f is your finish place

This gives a linear result that looks like this for a 9 man sng:

This accounts for 3 elements:
  • Number of players since more players means more points to the top spots. This reflects the increased difficulty in placing higher as the number of players increases.
  • Your finish place. Obviously the winner needs to get the most points.
  • Number of tourneys played. You get at least 1 point just for showing up.
Now, all this seems like a waste of blog space until we look some other systems. The most popular league system seems to be DrNeau's system which looks like this:
Points = (sqrt[(a * b) * (b / c)] )/ (d+1)
a is the total number of buyins in the tourney
b is the buyin price
c is the total price for the player (2*buyin if you had to rebuy for example)
d is your finish place

This system is great if you are running rebuy tourneys like we do in our home games. It also takes into account the buyin amount. If we get a regular home game league going, I'll probably use something similar to DrNeau's system. Since our private games don't include rebuys we can move on to another well know system.

Pokerstars tournament leader board uses the following fomula:
Points = 10* [sqrt(n)/sqrt(k)]*[1+log(b+0,25)]
n is the number of players
k is your finish place
b is the buyin amount

The pokerstars system results in a non-linear graph that looks like this:

As you can see, the top places get more points. I've actually adjusted their formula in the graph above. The last part of the folmula [1+log(b+0,25)] is only needed to account for different buyin levels. We have a set $5.50 buyin so I have dropped this from the eqation in the graph above. So, quit Googling what a "log" is:)

Notice that the minimum number of points you can earn is 10. the 10* at the start of the formula simply controls the baseline number of points. . This is important for the "no show" issue that we have been talking about. Missing a tourney means losing out on at least 10 points. This seems a bit harsh so we can take it out of the equation as well. It doen't effect the actual curve, just the points on the y axis. 30 becomes 3 as the top score and 10 becomes 1 as the bottom.

Now we are left with a curve that only accounts for the number of players and your finish. Just like a payout structure, the top spots earn earn the most. So, moving up from 2nd to 1st results in significantly more points than moving up from 9th to 8th. This would put me ahead of MrSmith and would put seabreeze ahead of Klokkhammer in the standings.

You may have noticed that these other systems all use square roots in their equations. This is good for narrowing the gap between 1st and last, as well as, for weighting the curve. We could adapt our current system to something similar like this:
Points = sqrt[p - (f-1)]
p is the number of players
f is your finish place

This results in an interesting curve.

We can now see that the gap between 1st and last has narrowed and that the lower places are weighted more than the top places. This would not change the standings. However, it would put even more value on playing in tourneys since last place gives 1 point which is huge on this scale.

DrNeau and Pokerstars systems would be great if we were running multi-table tourneys with rebuys and varied buyins. but they don't work that well for our current situation. Furthermore, these systems as used today only give points to ITM players. By including all players, we create a very negative situation for missing one of our private games.

Furthermore, I really don't want to change our current system mid-season without a clear reason. After reviewing the diffent options, I suggest we stick with what we have for the rest of the year. However, I will implement Klokkhammer's point about showing the number of games played since this only effects the way the results are presented and not the way they are calculated.

It is still quite early in the season. We all have some skill. You have beaten us (me) enough times live Klokkhammer to prove that you are still in a serious threat. Not to mention benkogambit whose sng skillz are light years ahead of the rest of us. I have gotten lucky in 3 games thus far, but that can change fast. The title is very much still wide open. There may even be new unknowns from the blog showing up. if we assume that MrSmith, MrEMC2 and myself are better players than the rest of you, we could consider a bounty system where whoever knocks us out of the tourney gets an extra point (or fraction of a point). Although I don't care too much for this route.

MrSmith suggested that we drop a low score or no show. Let's look more closely at this. Assume we play 11 games this year, 1 each month with a break in July. We can then take the top 10 scores to determine the champion. If everyone has at least 1 no show, then this is good. It helps balance the scores and creates no problems. If however several players have made every game, then it can have a serious impact on the standings.

For example, let's assume that the season ended now. MrSmith, MrE and I have played all games thus far. currently MrSmith is leading over me by 1 point (bastard lol). if we drop our lowest score I drop my last place finish from February (1 point) and MrSmith drops his 2nd place finish from April (4 points). Thus, I would win the championship by two points! MrE would also lose a 3rd place itm finish for 3 points. How this will play out at the end of the season is impossible to predict at this point. However, if we want to make this change we ought to agree on in now rather than later. I don't have a strong preferance regarding this point. Let me know if this is something you want to implement.

Finally, since we have used so much space on the leaderboard, perhaps we should discuss what we are going to do with the year-end results. I can arrange some sort of award and immortal fame on the blog. However there are plenty of other options. Come with ideas if you have them, but this doesn't need to be decided any time soon.

I need to get back to work! Let me know you thoughts...


  1. I can see from our new flag counter (left side panel) that you have seen my post regarding our leader board system. Perhaps I'm the only one who finds math fun - LOL!

    I apologize for the rambling post. Anyway, just to sum things up. Here is what I will be doing for the rest of the season.

    1. No changes to the way we calculate points.

    2. I will show number of games played when posting the leader board each month.

    3. I will incorporate MrSmith's suggestion. I intend on running 11 private games and at the end of the season we will say that the top 10 finishes count toward the final ranking. That gives everyone 1 freebie. Most of you have already used it up already though...


  2. Wow - Roland!

    I saw the post, but I felt I had to digest it before commenting.

    I agree 100% with your proposal.

    Sorry for raising the issue, but as a consequence we have now developed an even more comprehensive understanding of another important component of the game.

    Look forward to our next games :)

  3. I'm glad you agree with my final proposal.

    I like messing around with problems like this. It is exactly the type of thing I do all day at work :)

  4. Roland, I have now had time to give this some serious thought and here are my conclusions.

    First, it goes without saying that there must be something seriously wrong with any system which has me in 7th place with only 5 points after 4 months. After wracking my brain, I believe I have identified the problem. As you yourself pointed out, the current top three on the Leader Board are the only players to have played in all four tourneys. But what they also have in common is geographical location. The starting time of the tourneys is obviously designed for the convenience of the players flying the Norwegian flag. Being across the ocean, I missed the 2nd and 4th games which took place at 3:00 p.m. on a Sunday, my local time. I've spoken with the poker gods and they have confirmed that I was in fact scheduled to win those two tourneys, which translates into a total of 12 points (or 3 points per month) which have somehow gone missing.

    I therefore propose that anyone playing outside of Norway be awarded an additional 3 points per tourney, whether or not they actually show up. This would encourage global participation and turn what is currently a rather parochial affair into a world class event. If applied retroactively, it would give Seabreeze and Klokkhammer 20 points each and me 17 points which is clearly more in line with reality, with the top 6 now tightly bunched within a range of 17-22 points. Please note also that this proposal is completely non-discriminatory, as any local players are free to hop on a train or plane for Stockholm or Gothenburg or elsewhere in order to receive the 3 point inconvenience bonus.

    I am hesitant but in the interest of full disclosure feel compelled to also point out that the DATES of the tourneys are chosen for the convenience of one particular player (who shall remain nameless). Since this player is guaranteed to play in every tourney, there's no reason to reward him with a point just for showing up so I propose that he be docked a point every month. But I don't feel strongly about this.

  5. No extra points for +/- 1 GMT. That means UL for Klokkhammer + my Swedish and Roland's Danish co-workers. Agree on Benko's US timezone/extra points.

  6. And yes Benko! dock "1 point pr. month" off the one player who we don't need to name............

  7. LOL! Sound logic as always benkogambit. You should have been a lawyer :) I'll adjust the leaderboard asap.

  8. BTW, I'm both a Swedish and Norwegian flag on the counter.....

  9. I saw a Swedish flag yesterday after you had been on the site and assumed it was yours. I have no idea how they determine country, but it seems to be more than a simple IP address.

    The more the merrier as far as I'm concerned:)