|
|
| From | Message | ||
|---|---|---|---|
|
 Rating Tactics is not Rating EloAfter having played more then 100 tactics and after having read all what is said about the feature I can only conclude that the rating of the tactics itself is not supposed to match the ELO Rating we have here at Gameknot. Only a quick look at the top-chart for tactics proof that ratings are just not as high for tactics as they are for the general type of playing games here. The same can be said about the rating of the tactic itself which proposed rating is generally lower then what you experience when solving it. Now this would mean that for eg a 1400 tactic is supposed to match for eg a 1700-ELO player. And a 1200 tactic for eg matches a 1400-ELO player (Note that the interval don't have to be same as well.) -> 1200 * 20x for general games might be like 1050 * 12x for tactics where x is a variable. Now why does this difference exists. Probably because all tactics started as 1200 tactics and are only gradually picking up there true rating depending on the rate of success we solve them and the rating of solver. It's kinda the same with what we see at the normal ratings who will slowly increase for the top-players .... here we see it again but it's in its baby-fase at the moment. Another thing that comes to mind is that I think more good players are playing the tactics than there are bad players playing them. Now I can imagine that if a certain tactic is solved 19 times out of 20 but every-time by a 1700+ player ... it will be difficult for the tactic to climb to the real rating because the solving population is not evenly distributed. Things would be more clear if all kind of players tried the tactic and we had as many failure as we had success. Bottom-line is that I think the longer we wait and keep on playing the tactics in the meantime, the more accurate the proposed ratings of a certain tactic will be and the more our tactic-rating will start to match our ELO-rating. |
||
|
|
||
|
maca 03-Sep-10, 01:58 |
You are correct. We observe the exactly same phenomenon in the 'team ratings' of individual team players, that are ratings calculated based on the team games only. My chess rating is over 1800, whereas my team team rating is still below 1500. It is due to the fact that all team ratings were initialized at 1200, and it will take a long time until they will reach the levels of players' ordinary ratings. In order to get there quickly, I'd need to play players with team ratings significantly higher than my own, but there aren't too many players like that at the moment. Similarly, it will take time until sufficiently challenging tactics puzzles (in terms of their challenge rating) will be realized by the system, as the rating system of the puzzles themselves is also dynamic. In team ratings, each player's rating depends on their opponents ratings, and my team rating depends on my opponents' ratings. In tactics puzzles, each puzzle's challenge rating depends on the tactics rating of the players who take a shot at it, and each individual's tactics rating depends on the challenge ratings of the puzzles they attempt to solve. It takes a long time for this type of a fully coupled system to seek into it's 'equilibrium' position, where the tactics ratings of the players reflect their true solving strength, and the challenge ratings of the puzzles reflect their true difficulty. "Another thing that comes to mind is that I think more good players are playing the tactics than there are bad players playing them. Now I can imagine that if a certain tactic is solved 19 times out of 20 but every-time by a 1700+ player ... it will be difficult for the tactic to climb to the real rating because the solving population is not evenly distributed. Things would be more clear if all kind of players tried the tactic and we had as many failure as we had success. " The composition of the solver population (in terms of their true skill) should not be so significant long-term, because a more skilled solver is already likely to have a higher tactics rating, and their influence on to the challenge rating of the puzzle will therefore be more significant if they fail in it. But since all the tactics rating were initialized at 1200, it will take a long time until this effect is felt through the system. "Bottom-line is that I think the longer we wait and keep on playing the tactics in the meantime, the more accurate the proposed ratings of a certain tactic will be and the more our tactic-rating will start to match our ELO-rating." Given that team rating were instituted a very long time ago, it could easily be years with the tactics exercises. Regards, MaCa. |
