Multiplayer Forums

The system dont use the number of games, becouse some games are worth more than others. For examples, beying the first or the last will decrease your volatility slower than beying, lets say, the third. Also, playing with opponents with ratings much different from yours will make the volatility go down slower too.

I see, i was actually thinking about a very slow erosion anyway (1/4 for 100 games played; 1/2 for 500, 3/4 for 1000, or something like that), so i guess that won't change a lot from your result. A little less Yo-yo effect in best case.


Thank, i'll try to have look soon.


Always good to use what others create for you

I agree with claymore on this point. The more games you play the smaller or narrower the point spread should become as the math is attempting to focus in on your playing level. The more data (games) it has the more accurate the rating should become. The math should be trying to determine your rating based upon a scale. If the scale is ever increasing then the rating becomes meaningless.

I think you got it wrong. The volatility is not the scale of the rating and it will never increase. It will just never reach 0, even when games reach infinity.
The +100 will not increase the volatility, will just make system consider your true strength may change with time! This could be removed if we were rating AI`s for example. The number of games is not used, as some games may decrease your volatility more than others. But yes, playing more games will decrease your volatility

okay, you lost me.
1) Why are you making a distinction between AI and players?
2) You say the number of games aren't being use because some games might decrease your volatility more than other. So - why is that an issue?
3) If playing more games will decrease your volatility but if it does not take into account the number of games. Then how does playing more effect it?

1. I was just trying to explain that the +100 in my formula wont make volatility increase... I was saying that it is needed becouse we are humans and we may learn, so the system may never be completely sure of your rating ( volatility 0 ). AI`s dont learn, so if the system were developed for AI`s, then volatility close to 0 would be possible. But we can use this code for AI`s too . Will work well.

2. An issue? Sorry didnt got it.

3. Every time you play a game the system decrease your volatility. So more games mean your volatility will go down.

1) really clear now

2) neither do i

3) do you mean volatibility always go down, i mean never go up (i know ridiculous to preciseit , but i'm becoming paranoîd ) ?
That would be perfect

#2 - In the quote above you say: "some games may decrease your volatility more than others". Why is this? Why will some game effect it more than others? And why is this an issue?

#3 - Give us an idea of how quickly it decreases from 100 based upon how much more you play. For example, how much has it decreased after 10 games?, 50 games? 100 games? 200 games? And when does it flat line?

You did say it won't drop below 40 right? So the volatility range is 100-40?