Ladder Point Adjustment

[LPhillips]

Hi there. How is the ever-growing list coming along?

I did a search, but all of my inquiries brought up too many extraneous topics, so I thought I should start a new one. Can we discuss options for changing the ladder point system?

Let's use Hammer Mountain for our example. Games there end in a matter of 6 turns or so. It's rather silly to get the same points for, say, Burning Ice, which takes around 20 turns to play (if your opponent is garbage), or around 30+ if you face someone good. Some people rack up dozens and dozens of games that take less than 10 turns to play. In my humble opinion, those games should be worth 5 points or less. They're hardly even games. I realize there is some small strategy and planning involved, but (unless your opponent is no good) it simply comes down to a roll of the dice in one battle for those games. If you play on a really competitive map, and invest the time (Battlefield or Bull Run, to list a couple of specific examples) then you should be rewarded properly. I don't propose rating the maps based on difficulty, because that seems too arbitrary. I merely submit that the time spent to win a game is nearly a direct function of its difficulty and of the reward deserved by the victor.

My proposal is that there is an amount of points available for winning a game on a certain map size, and then the points are awarded based on the number of turns used. Just to give an arbitrary example, let's say 50x50 maps can net up to 20 points. <10 turns is a modifier of 0.25, for 5 points. <15 turns is 10 points. <20 turns is 15 points. The scale should curve up toward the end, so that it takes ~25 turns to get 20 points. Again, this is an arbitrary example and it needs careful study before picking actual numbers.

Any reform probably can't be implemented until next ladder season, for the sake of fairness. But could it be discussed and established now?

[piranha]

Right now its like this:

mapsize < 5000 k = 16;
mapsize >= 5000 && mapsize < 7500 k = 22;
mapsize >= 7500 && mapsize < 10000 k = 28;
mapsize >= 10000 && mapsize < 15000 k = 34;
mapsize >= 15000 && mapsize < 20000 k = 40;
mapsize >= 20000 && mapsize < 25000 k = 46;
mapsize >= 25000 k = 52;

So the constant K is bigger when you play on a bigger map. There are some problems with having this. One is that you can make a big map with water around the center of the map, some popular 1v1 maps could be 50x50 or at least 75x75 (if it existed) but are 100x100.

If you use number of turns too it will affect gameplay as it means you don't want to use a fast attack strategy because it gives you less points, but a fast attack strategy is sacrificing expansion for pressing the opponent so I don't think you should be punished for aggressive gameplay.

I guess an even more advanced formula could be created where we measure the average number of turns it takes to play a map and have a gradual K value depending on that. That way a map would be locked to the number of turns it takes to play it instead of size or the number of turns it took on one game. Still could be abused by playing several games where you just click end turn a bunch of times before surrendering. Still might be the best idea so far.

[Moonknight]

LP, you should not be punished for a quick slaughter of your opponent.

Likewise, if you're a huge favorite to win on a big map, you could just wait for that next set of points before finishing the game to get more points.

I think the size of the map scale is a better route...

Now, what I don't remember is if who you're playing adjusts how many points you win? So if someone experienced gets to play a noob (and I know this is being adjusted in the next version so this will not occur as often), do they get the same amount of points?

[LPhillips]

The adjustment according to ladder placement is a pandora's box which must be tackled eventually.

Moonknight, players aren't punished for beating their opponents quickly in my system. In fact, it rewards harder fights by awarding more points for a protracted game; in other words it accomplishes what you are suggesting. If someone can be beaten very quickly, then you don't deserve many points for beating them. There's no situation in which I stomp a very good player in a few turns (except bad map design). It's not actually ripe for abuse either; the map still has a size-based point limit so sitting on a newbie who doesn't know they have lost for 20 turns isn't going to get you some insane amount of points. It's more to your advantage to end the game quickly and move on to another game where you can gain 10-15 points (in my example) instead of protracting a dead game for an extra 5. It's extremely abuse resistant.

The size of the map is the worst possible option after no scaling at all. We've come a long way just by scaling according to map size; now it needs to become a more exact science to avoid the caveats Piranha has mentioned and the ones I have described.

Piranha, your solution for an adjusting K value sounds like a pretty good one. As long as...
1) Only ladder games are counted (no AI etc)
2) Players do not intentionally abuse a map. This is something that I don't see as a possibility; if you host ladder games and end them quickly you A) wreck your ladder stats, B) lower the map's K value and point rewards. So there is no upside to the abuse you mentioned.

[KGB]

Is the K value a multiplier or a constant that is simply the base game points rewarded to the winner?

If it's just the base game points then a K=40 for 15000 (100x150) vs a K of 16 (for 50x50) still isn't going to help. A game on a 100x150 map can easily take 30-40 turns and if it has multiple players/team play can take 2-3 months. A 50x map that completes in 10 turns between 2 players might be done in 1-3 days. So you are still hugely rewarded for being able to play a TON of games on 50x maps even if your winning percentage isn't that good simply because of the sheer volume of points.

Maybe a better solution is to split the ladder into 2.

1) 50x50, 50x75, 75x50 and 75x75 maps make up one ladder. These are for fast game play.
2) All other sizes for longer games that often involve multiple players/team play.

This way you won't have to try and come up with ways of balancing points between the various map sizes. That's probably going to be impossible to do without harshly penalizing something because I often see friend reports of winning/losing ladder games in under 5 turns. Even with LPhillips 5 points for such a win you can rack up a TON of points if you have 6-10 of those games going at once and win 50% of them generating 35 points every couple of days.

KGB

P.S. Another way to do ladder is to drop the point system entirely and use what Chess does in tournaments. That is winners play winners and losers play losers and best winning percentage wins the ladder. So at the start everyone begins with a 0-0 record and you can play anyone. Once you win and go 1-0 you can only challenge another 1-0 player. If you drop to 0-1 you can only play another 0-1 player. After 2 games the 2-0 players must play, the 1-1 players must play and the 0-2 players must play. And on and on it goes. This forces top players to play each other and not beat up players with inferior rankings for points. Since everyone won't have the same number of games played, you use winning percentage to challenge another player so that two 60% winners can play but a 60% can't play a 40% player.

[Maze]

I agree with Moonknight that a player should not be punished for a quick slaughter of the opponent. Also, the differentiation of points depending on map size does not sound appealing. Map size does not equal speed since a small map with mountains/hills, few passages, slow units, no roads, poor income can be slower than a large and rich map with wizards and flyers as starting units and lots of roads between the cities.

KGB's suggestion to let players with same win/lose percentage fight each other and drop the point system sounds good, I would include a tolerance and a precaution to that: if a player won e.g. 60% of his games and is ranked 15th, he should be able to play all players that won 60% plus or minus x% (where x is a tolerance of e.g. 10%) AND he should be able to play all players with rank 15 plus or minus y (where y is a second tolerance or precaution of e.g. 20). The second tolerance or precaution is necessary to make sure a player can play a minimum number of players. It also makes sure that a player who is ranked N. 1 with e.g. 80% wins is not sky high and lonely with N. 2 only having 60% making N1 unable to play any ladder games for a while.

[LPhillips]

KGB,
How would # of wins be factored in to the system you propose, where players are ranked according to percentages?

Moonknight,
Maze, the system must indeed have some tolerance for players of different rank to play, or one could enter situations in which no opponents were available. This must be taken as a given point in designing the system.

~~~
I think separating the maps by difficulty (rather than by size, as I maintain that maps cannot be equivocated by size) is a good move for the ladder system. It's easy to roll the dice and pull off Hammer Mountain (sorry, but it's the best example of a cheap win). You can't balance it in comparison to a Battle Field win, or even a truly difficult 50x50 map. It's also impossible to use raw map dimensions as a measurement of the playing field. Maybe Piranha's K value (K is generally used as a constant; IE a multiplier) could be modified to simply classify maps.

Finally, it seems to me that the % system as proposed would require the client/website to suggest or even enforce opponents. A matchmaking system is unworkable currently given the community and the maps. Is it possible to simply adjust the # of points awarded according to the winning percentage of the opponent? With 0/0 as your baseline, and then the percentage adjusts it from there? That would give players the freedom to play whomever they choose, but reward them best for taking on strong opponents. Playing someone with a 20% win percentage is not very attractive compared to trying to take Igor for a game, and gaining 4x the points. None of this solves the map size issue, but it does remove the high reward for beating newbies (and, I still maintain, chasing people away from the website by the dozen).

Are newbies restricted from entering ladder games yet? That is the most paramount problem I see with the ladder system: if they are able to enter them at will, then they're the best prey for these one-off, favorite map players who try to cycle a hundred wins in 3 weeks on 5-turn maps. I guarantee that makes people leave in frustration and never return. It's not as if the ladder climbers can stop this on their end.

[KGB]

LPhillips,

Total # of wins doesn't matter. Let me clarify a bit more what I mean.

Lets say the ladder season is 6 months in length. You set a minimum number of games played to qualify for the title (so someone who is 5-0 can't win). Say that's 25 games completed (roughly 4 a month or 1 a week). You also have to have a minimum activity requirement of say at least 2 active games a month in the last 2 months (to prevent someone from quitting the last 2 months). Note: Active means you are playing, not necessarily completing them because at the end of the season Piranha enforces winners in every game anyway.

That's pretty much all you need to do. For new players/players joining in mid season, I'd say they have to complete 3-4 games before they start getting matched up winning % wise. Before that they are free to play anyone so this is one place where you can exploit a newbie but only for a couple of games until they get ranked.

Thus in my example (6 month ladder, min 25 games played) it would be best winning % wins ladder. So someone who is 20-5 is the same as someone who is 40-10. Playing a lot of games is useful if you are winning those games. But winning will get harder and harder because the more you win, the better the players you are forced to face. Right now there is nothing that forces top players to play against each other so we have no idea for example if the top 2 players in last seasons ladder ever played each other or if they did, how many times they did.


Note: if you go to this system 1v1 and Team/FFA games will have to be separate since there is no realistic way to account for 2nd place in an 8 player FFA unless you wanted to award them a 6-1 record and the winner an 7-0 record. But I'd prefer to simply put teams into their own ladder and remove FFA ladder play entirely.

KGB

[Maze]

KGB,
Your reward for an 8 players FFA should read 6-1 for 2nd place and 7-0 for 1st place. (winning 1vs1 = 1-0, 3 players FFA = 2-0, 4 players FFA = 3-0, ..., 8 players FFA = 7-0. Note the 1vs1 = 1-0 is mentioned for completeness but I would keep 1vs1 and FFA ladder split.)

I am not really fond of this system (where the runner-up takes 6-1 and the first to die takes a 0-7 hit as it would change the game from "winner takes it all" to "let me try and survive as long as possible" resulting in a totally different strategy where the player who is extremely strong will get truces from everywhere and all the other players will be trying to fight each other to finish second and leaving the strongest player play and win his game.
Even when only three players are left, player two and three try to kill each other to make sure they won't finish third instead of teaming up against the strongest player.

This said, in a "winner takes it all" mindset, I would "reward" an 8 player FFA game as follows: winner gets 7-0 and the other 7 players get 0-1 (resulting in a balanced aggregate outcome of 7-7 for this game or if you wish "if 8 players who are equally strong play 8 FFA games in an ideal situation they would all have score 7-7 after 8 games, each winning 1 game and losing 7 games).

[KGB]

Maze,

I edited to correct the error you noted.

I agree with your concept of 7-0 for the winner and 0-1 for all the other players making an aggregate 7-7 for the game.

Incidentally it appears there are already separate ladders for 1v1, FFA and Team play which is exactly what we both think is ideal.

KGB