| Rune Central http://forums.runequake.com/ |
|
| Statistical Note http://forums.runequake.com/viewtopic.php?f=2&t=213 |
Page 1 of 1 |
| Author: | Baker [ Mon Oct 13, 2003 10:16 pm ] |
| Post subject: | Statistical Note |
Jackhole has changed GameStatus.Net to "experience" based scoring. The concept is that if you get a frag in a "full game", that means more than when less players are around ... and generally speaking this is true based on statistics. For example, if there are 10 players on a server, the odds of all 10 players sucking is poor. Hence, mediocre players rarely win a game with a ton of players, there are usually 2 to 3 decent players present. The less players present, the less likely that a "great" player is on the server. This isn't perfect, but if you take a look at the results in progress it seems to rank players better on ability. It is a significant improvement over frags per minute. |
|
| Author: | toyo_mr2 [ Sat Nov 08, 2003 7:33 pm ] |
| Post subject: | |
Why is ExP cumulative? Currently, it rewards players for playing more games and therefore is somewhat redundant to the "frag" statistic. If you want a stat that ranks players on ability, as opposed to endurance, you should take the percentage of ExP accumulated vs what was available to that player. Of course, under ExP theory, winning a map with more players should be weighted more. Here's an example: "Player A" wins a map with 10 players, and gets 10 points out of 10. wins a map with 5 players, and gets 5 points out of 5. comes in 2nd with 10 players, and gets 9 points out of 10. comes in 2nd with 5 players, and gets 4 points out of 5. After four games, "Player A" has a ExP of 28 and a ExP pct. of 93.3%. "Player B" comes in 7th with 10 players, and gets 4 points out of 10. comes in 9th with 10 players, and gets 2 points out of 10. comes in 4th with 10 players, and gets 7 points out of 10. comes in 2nd with 8 players, and gets 7 points out of 8. comes in 3rd with 8 players, and gets 6 points out of 8. comes in 2nd with 5 players, and gets 4 points out of 5. After six games, "Player B" has a ExP of 30 and a ExP pct. of 58.8%. "Player B" has a higher ExP than "Player A" simply by playing more games. The ExP percentage gives a much more accurate picture of their performances. This type of scoring only works if GameStatus can accurately determine if players actually played the entire map. Of course players shouldn't be penalized for finishing poorly if they entered late while players shouldn't be rewarded for beating late comers. toyo |
|
| Author: | Baker [ Sat Nov 08, 2003 11:49 pm ] |
| Post subject: | |
Toyo, GameStatus uses a frag multiplier. If there are 9 people present, you get 9 points, if only 2 are present, you get 2 points. The theory is that Player X might win a game and get a lot of frags against 3 others, but the more players present, the more likely skilled players are present to suppress said players scoring ability. For more information, check QuakeBoard.Net or ask questions there. I tried to sell Jackhole on a statistically sound method similar to one you would approve of based on statistical theory. The current method is pretty close to what I suggested anyway. |
|
| Author: | Diazoild [ Sun Nov 09, 2003 2:23 am ] |
| Post subject: | |
There will be always contention with the process that assigns rankings. |
|
| Page 1 of 1 | All times are UTC - 5 hours [ DST ] |
| Powered by phpBB® Forum Software © phpBB Group https://www.phpbb.com/ |
|