I'm trying to work out the best way to calculate tournament rankings and am interested to hear feeedback on the best one (or if you have a better idea I'm all ears). Here's the options I have so far:


1. All players start with 1000 ranking points. Entering a tournament costs you 50 points. 1st placed player wins 50% of the points pool, 2nd place 25%, third place 15% and fourth place 10% (or similar ratios, perhaps extended to the top ten players?)

 + Simple, easy to understand and easy to apply to all tournament types
 + Gives nice separation between winners and losers
 + Zero sum system (except if players get to 0 points!)
 - Doesn't reward you for winning several tournament games if you still don't get to the bottom paid out place
 

2. All players start with 1000 ranking points. Calculate ranking point changes the same way that they are calculated for regular games. Essentially the tournament ranking points is exactly the same as a tournament specific global ranking score.

 + Same as game points system which is well understood and well proven
 + Rewards you for winning a single first round game even if you don't progress any further
 + Zero sum system (except if players get to 0 points!)
 - Not much separation in points between the winner and the second placed person (will be only around 20 points)

That's all I've up with, I'm tending towards (1) because of the way it rewards the first placed player highly which is essential IMHO. Bear in mind that any ranking system needs to take into account the different types of tournament (round robin / knockout).

What happens when you run out of Tournament "Ranking Points" with Option 1?
That's 20 tournaments without a place, which probably isn't going to be unheard of.

I think this was Yertle's question, but I'm not sure... if someone runs out of Tournament points (i.e. they join 20 tournaments and don't place in any of them), do they just not get to play in tournaments any more?

And I assume the overall point of the tournament points is just as a ranking thing, and they have no benefit other than being used to join more tournaments?

Yes it's just a ranking thing. If you get to 0 points you'd just be gifted with the points to start the tournament. Or it could go negative I suppose :)

i guess i kinda like option 2 better. that gives someone who wins their way the benefits.placing 2nd in a big tourney is darn near as good an achievement as 1st so 'only' being behind in points really doesn't seem like a good argument against it. the winner is the winner, gets the badge/trophy wife/whatever plus a slight edge in points. the way it should be IMO.

Thinking about it more I also like #2. There doesn't need to be a big disparity between 1st and 2nd place in a tournament, because as Weathertop said even 2nd place is a big accomplishment. Plus, over the course of 5-10 tournaments if 1 person continually places up there and another doesn't, then their scores will begin to reflect that. Or, if a couple people continually battle for 1st and 2nd, then they both deserve to be close in points. As you said, it's a proven system and it does accurately reflect someone's general skill level when many games are taken into account.

I personally like option 1 but what i would do is have the winner of each round get a small prize 1-2% then the rest be divided between the final table.
if there is such a small margin between 1st and 2nd place then it is a little pointless to play to win.

KrocK wrote:
if there is such a small margin between 1st and 2nd place then it is a little pointless to play to win.

You still have to win to get points.  It's not 1st and 2nd of each game, it's just that at the end of a tournament 1st and 2nd may not be all that much different, actually it could come out to where 2nd nets more during the tournament if they started low and faced some high tournament ranking opponents, where as 1st had the easy track... at least I think.

I think only 1st should get the "Trophy" which should be a Ranking all itself. 

If went with option 2, perhaps there could be a "bonus" for 1st, 10 points per 20 players in the tournament (or some better numbers).

I imagine option 2 will result in less 0 rankings (lowest global ranking on the site is 408), since it's on a game by game score.  Where as option 1 I could see more potential for hitting 0, but probably an easy rebound with 1 tournament win or a couple good placements.

Yertle wrote:

I think only 1st should get the "Trophy" which should be a Ranking all itself.

That's a good point really - # Trophies is a measure of how good you are at winning tournaments.

And your tournament ranking score shows your overall game winning performance within tournaments.

That's closer but I still don't like it. My first thought is to add a multiplier for later-round games. But that would penalize players for losing against tough competition. So maybe just multiply the score gain but not the score loss. But then you lose the zero-sum beauty of the system. So that wouldn't work.

My next thought is some sort of complicated system where you take extra points away from early-round losers and award them to late-round winners. Let's see if I can work this out at all.

Let's stipulate a 3-round tourney of 4-player games, single elim. So 64 players. Using the existing Global Ranking Score system (and assuming everybody starting at 1000 for simplicity's sake)...

Round 1, status quo:
1000 -> 980
1000 -> 980
1000 -> 980
1000 -> 1060

Round 2, status quo:
1060 -> 1040
1060 -> 1040
1060 -> 1040
1060 -> 1120

Round 3, status quo:
1120 -> 1100
1120 -> 1100
1120 -> 1100
1120 -> 1180

final results, status quo:
48 players knocked out first round, rating 980
12 players knocked out second round, rating 1040
3 players knocked out third round, rating 1100
1 winner, rating 1180

I think that could be improved by a flat rule saying anyone getting knocked out in the first round gives up 1 extra rating point that goes into a holding pool to be distributed to the final table. Maybe just to the winner.

So under this system, using the above example:
48 players knocked out first round, rating 979 (taking & saving 48 total points for later)
12 players knocked out second round, rating 1040
3 players knocked out third round, rating 1100
1 winner, rating 1228 (gets the 48 points)

If 2nd/3rd place get a few of the extra points:
48 players knocked out first round, rating 979 (taking & saving 48 total points for later)
12 players knocked out second round, rating 1040
4 final table players:
4th place, rating 1100
3rd place, rating 1108 (gets 8 of the saved points)
2nd place, rating 1112 (gets 12 of the saved points)
1 winner, rating 1208 (gets 28 of the saved points)

I think something like this more properly rewards the extra difficulty of winning late-round tournament games without overly penalizing early losers. Is it too complicated? Maybe. Not as bad as I expected when I was first thinking about it, though. I'd think you can sell people on giving up an extra point for losing immediately in order to more richly reward the eventual winner(s).

I'd still vote for the poker-style entry fee proposal over a straight-up application of the Global Ranking Score system. Anything that draws no distinction between a first-round and fifth-round tournament game is useless in my eyes. Over time your score would just come down to volume of games won and in that respect wouldn't add anything to a players stats or rankings above what we've already got IMO.

thats exactly what i was thinking. i'm not necessarily playing for ranking scores as i am for that trophy. but if i fail to get the trophies as i keep coming in second, then at least i have a high ranking to fall back on.

Do tournament games count towards regular rankings on those boards?

I like how option 1 hands out points.  Maybe create a distribution calculation that divides the points pool up amongst all players making it to the final table. 

That being said, the 'entry' fee should be calculated differently.  I think it should be calculated based off of  current tourney ranking / 1000 * X, where X is the base entry fee. That way it stays a zero sum system, and those that win a tourney are risking more to try to increase their score.  

IRoll11s wrote: Do tournament games count towards regular rankings on those boards?

No... the overall opinion when this was discussed is that it should be kept separate.

I like your idea asm, sounds like a hybrid of 1 & 2 overall.

I do think that the whole concept of a tournament is that it's a winner takes all type situation, or at least winner takes most - after all you don't see the winners of the US Open splitting the winnings 51:49.

AntiScourge - that's a great way of solving the zero sum problem, I think that's down the right lines. I think there also needs to be a formula for the allocation of points - top 10 works fine for a 60 player knockout but not for a 10 player round robin game.

Something like the top 1/3 of the players gain ranking points?

Top 15-20% sounds better imo. i.e. in a 64 player map, if you want 10 people to get points, that's just over 15%.

I'm not sure how you can determine the 'ordering' of anyone other then the people that make it to the final table. Lets say for example you have 64 player tournament with 4 player games.

Round 1:
16 Games - 64 Players
Round 2:
4 Games - 16 Players
Final Round:
1 Game - 4 Players

So if you wanted to award the top 10 players, you'd have the top 4 easily, but how do you determine who gets 5th? In risk you can't really say who got 2nd, 3rd, or 4th. If you go by the order each player is eliminated then usually the only difference between the player finishing 2nd and the player finishing 3rd is whether the winner of the game decided to go left or right. That's an unfair distinction in my books. Granted in some games it might be 'clearer' then that, but it's a judgement call that no current computer system can make. Then the next problem comes in if you did determine who got '2nd' in each of the round 2 games, how do you order the people who got 2nd into 5th-8th places?

If you want to award people who win in the first round, the only other system I can see doing is awarding maybe 1/2 of your points back if you made it to Round 2, and then distributing the remaining points amongst the players at the final table.. which unfortunately again would have to be in the order of elimination.

Scourge is obviously correct. There's no way to award ordered points for finishes outside the final table.

I don't think it has to get nearly that complicated though. I don't even agree that only the top x% of finishers should get points. If you win a first-round tourney game, you should get a few points for that. To me the only problem with directly applying the Global Ranking score system to the tourney setup was that it didn't properly scale, not that it rewarded early wins at ALL.

There aren't really that many possible setups, so creating a bare-bones framework with some numbers set in stone wouldn't be all that hard.

Assuming single elimination:
- if tourney is 3 rounds or fewer, take 1 extra ranking point from every first-round losing player and distribute those extra points to the final table players.
- 4 rounds: take 2 extra ranking points from every first-round losing player and distribute them to the final table players.
- 5 rounds: take 2 extra ranking points from every first-round losing player and 1 point away from every first-round winning player and distribute them to the final table (5th round, so 4th round winners) players and every winner of the 3rd round

To ward off confusion this can be a black box system with the help pages just saying something generic about how losers of first-round games lose just slightly more points than typical in order to reward the eventual final table players more proportionately.

Similar systems can easily be worked out for swiss and round-robin formats.

* * * * * ** * * * *

Alternatively, if none of you like that, it would be fairly easy to answer the question posed by AntiScourge using a similar mindset. Since you clearly cannot rank player finishes outside the final table, just make a base award for making it past a certain point. In a 5-round 4-player single elimination tournament you have 256 players entering, 16 of whom will win their 3rd-round games, of which 4 will go on to the final game. Paying those 16 would be a little over 6%. 64 players will win their 2nd-round games. Paying those 64 would be 25%. 25% is closer to 15% (our target proportion) than 6% is, so that would be the cutoff point. So the final table players would get paid according to a predetermined distribution, then every 3rd-round winner would get paid, then every 2nd-round winner. You could design a sliding scale that would do the calculations for how to divide the prize pool in all the different cases (or just have Hugh do it).

One other thought: would it be possible to have the system "know" whether eliminations occurred on the same player-turn or whether a turn elapsed between eliminations, and call single-turn eliminations a tie?

For example, Andernut eliminates Yertle easily, taking his cards and using them to knock out Weathertop. But in my desperation I've spread myself out so that I can't be reached in one turn. It takes Ander another turn or two to wipe me out. In this scenario, Yertle and 'Top would "tie" for 3rd/4th and split the allocated prize for those places 50/50 while I take 2nd.

On the other hand, if Yertle has played so poorly that he set himself up to be eliminated easily and was holding 5 cards, allowing Ander to turn in 2 sets at once, and he gets all three of us in the same turn, the three of us woule "tie" for 2nd/3rd/4th and evenly split the allocated prizes for all three spots.

That would get around the problem of giving out rankings based on random elimination order, if it were possible. Another possibility would be to do it the way poker tournaments do and base place finished on how many armies each player had at the start of the turn in which they were all eliminated. Of course this would also involve pulling data out of the Player into the tournament calculations.

What's up with asm getting all Mathy now?

If you read it there's actually almost no math at all. It's just my usual harebrained schemes squeezed into a math-looking context.