Which is a more credible assessment of an opponent's skill? Board Championship or Global Ranking? Can they even be compared versus each other? Discuss.
Overall I would say (Board) Championship Points... shows the range of boards better than Global Ranking. IE I could have a 2500 Global Ranking but that could all be from the same board, whereas if I have 100 Championship Points I have to be the top and above 1500+ on a minimum of 5 boards (which is more realistically going to come from multiple boards).
Not that having a high Global Ranking is not good and cool, I would just say that as for looking at a range across WarGear I would say Championship Points.
in general I agree with Yertle. but BC points have their issues also - they tend to favor long-term/heavy users. For established players I think the H-Rating is just as good also.
Amidon37 wrote:but BC points have their issues also - they tend to favor long-term/heavy users.
Good point.
H-Ratings do it for me. BC means very little if a player has favorites and/or specializes, and Global Rankings are way too volatile. My GR has been above 2000 and right now is below 1000, but my H-Rating is consistently around 57-62. I still would like to see a 50 or 100 game moving average for Global Rankings.
i'm an H-rating guy too.
I've noticed that the players who have the highest global rankings usually have a system by which they achieve their high score. They find a map scenario where they can win most of the time (often involving lighting rounds or non-standard mechanics), and play that the majority of the time.
This working of the system also boosts H rating, although I generally think H rating is less transient than global score.
Take a map you dominate on more than any other. Play only that map. If I played only Octagons, my H-Rating would start to approach 96% and my equilibrium global rating (assuming 1000 rated avg opponents) would be just under 5000. (Not too much worse if I use the more mappy Stargear!) But alas, I like to bowl and arm wrestle.
With such map variation on the site, assessing "overall skill" in a single stat is truly challenging. I like CP's okay, but agree that they are definitely longevity skewed.
I would submit that one system does not necessarily show skill more than the other. Rather a combination of the 3 (championship, H-Rating, Global score) is the real measure of ones skill. Take BlackDog for example: He is almost always in the top 10 in both championship points and global ranking. (hard to look at top 10 in H-Rating since so many haven't played enough games.) Most here would consider BlackDog the best all-around player.
If you compare stats between championship and global scoring you will find the following facts:
Global Score has an average H-Rating of 76.3 and an average games played of 443.8.
Championship Points has an average H-Rating of 71.1 and an average games played of 815.4
What does this show? Championship points showcase players who play lots of different boards. Whereas Global Score leaders play fewer boards/games.
It is impossible to rank high on the championship points if you limit the boards you play. It is also hard to rank high on global scoring if you play many boards and games because the higher the rating the higher the win % is required to stay on top.
For example when I beat a player with a rating of 1000 points I gain 6 points, but if I lose to that same player I lose 60 points. Thus I have to maintain a 90% win ratio to stay the same.
An average player could climb the ranks in either category. (I consider myself one of those players.) If I play every board and had an H-Rating between 50-60% I believe I could climb to the top 10, but my global scoring would be no where close enough to the top 10 global scores. I could do as BlackDog said and limit myself to a few boards and master them to increase my H-Rating and global score, but would be far from close to the top 10 in championship points.
After saying all that, I would throw in one more point that kind of goes against what I am saying...
Some of the toughest opponents I have played against are not even listed in the top 10 of either category. So my final feeling on the matter is... Rankings do not mean much in my book.
falker1976 wrote:I would submit that one system does not necessarily show skill more than the other. Rather a combination of the 3 (championship, H-Rating, Global score) is the real measure of ones skill. Take BlackDog for example: He is almost always in the top 10 in both championship points and global ranking. (hard to look at top 10 in H-Rating since so many haven't played enough games.) Most here would consider BlackDog the best all-around player.
Ok; so there seems to be consensus that the three systems have their "holes". Individually they can be "gamed" to a certain degree (by specialization, longevity, etc). But it has been pointed out that it's mighty difficult to have high scores in all three. So how about we create a composite score and settle the debate once and for all and see how Black Dog fares?
(CP * GR) /100 * HR
HR is a percentage so just convert it to a decimal. Simplified:
CP * GR * HR / 100
Here's a sampling of current top players in a number of categories.
BlackDog 5253
Mad Bomber 3193
Djembe 2539
Hugh 2323
Conan 1869
Seige07 1851
Falkner 1702
Dud 1688
Cona Chris 1643
Yertle 1603
KrocK 1538
Davidny212 1462
Viper 1193
Toto 1149
You definitely need to have a decent amount of CPs to register on this scale.
Alpha 1259
What about tournament points ?
No single score/method is ever going to be the be-all and end-all "credible assessment of an opponent's skill" like Attila started the debate with. For that it would be cool to have more objective type statistics that were easily accessible/sortable. Like in sports no one statistic can define how good a player is overall. For us trying to come up with ones that work good across all boards is not easy though. I think Global Ranking and H-Score fit into this category and both are decent. More would be cool.
The other way to assess your opponent is to see, in general, what awards/trophies/achievements that player has achieved in their career. Board Championship points (and tournament trophies) are more this for us. We have talked on and off about more such things and that would be cool too.
Agree with Amidon37. There are many ways to assess your opponents. I did not include Tournament Points in my composite calculation because there are too many people who do not participate in tournaments. Trophies fall into that category as well.
That said, I see no reason why you couldn't multiply the composite score by TP and divide by an additional 1000 to get an even more inclusive number, making our equation ...
CP * GR * HR * TP / 100,000
The math here isn't very good.. if I have 1.5 times as many championship points as another player, 1.5 times their H rating, and 1.5 times their global score, I would assume my overall score would be about 1.5 times theirs. Using the original formula it would be 230% higher, and adding in the T score, it would be more than 400% higher.
Also, Global ranking and T score are calculated logarithmically, while championship points and H rating are linear. This skews your current number in favor of the latter two values.
Finally, a championship score of 0 would give you a score of 0 using this formula.. I would feel pretty cheated if I had a 2500 global ranking and no championship points.
Stopped in to see how things were going. Good to see after being inactive so long my name is still showing up in the top player discussions. =)
As BD pointed out, the composite M57 is using is very heavily weighted by CP's since they are linear and favor someone who has played longer but I'm wondering if there is some way of scaling that by total number of games played. Equally skilled players who have played 100 games vs 1000 games would be expected to have different CP totals.
Like all of the other variables it is something that could be gamed, but it could be factored in to help balance the CP scaling. I'm not sure exactly how yet and don't feel like making my brain hurt at the moment, but it's a thought.
would it be somewhat better if you did something like:
[(CP/2) + GR] * HR
taking from an earlier post that it took almost 2x as many games for CP to have similar HR...
i agree that tourney's are completely different ballgame than any of these and can't see them being thrown into the mix cus all you would be doing is adding on another term (literally) that would be 0 for those that don't participate.
Instead of trying to come up with one score that attempts to summarize what we have I think it would be more useful/fun to create others - some off the cuff thoughts
1) Somehow tie in our stats with winning seat position %'s for a board and a players winning set position %'s. Give a player more points/higher score for winning games from seat positions that others are less likely to win from.
2) Somehow adjust a players eliminations to the number of players. Earning an elimination in a two-player game is less of a deal (except it means you won) then getting 5 eliminations in a 6 player game.
3) Try to do something universal with luck stats. I know the difficulties of this have been discussed, but I wish they could get solved. How about an ability to win despite poor luck stat?
More fun is the achievement track - which has been discussed in multiple threads, but we certainly can again.
Agree with BlackDog's assessment of my very ad hoc composite score idea in that CP is weighted too heavily (don't know about the logarithmic vs linear part of his argument though). Sorry BD, but think you are the best player on the site ..anyway you slice it.
Weathertop's /2 idea is too arbitrary a solution, but it's in the right direction. the problem is the top CP number will continue to grow, while GR will remain static. Surely there are some ways to mitigate this. For instance, somehow make the range of CP points comparable to the range of GR points.
All of amidon's suggestions are additional assessment/ranking tools, but not really that helpful when it goes to answering the Top Overall Player question. And Amidon, I agree that the achievement track will be a great addition to the site ..if we can ever come to a consensus about what it should look like.
I'm not even sure that we need or should have a composite ranking system, but it is kind of fun to be able to come up with a quick down-and-dirty number based on a range of what are generally generally acknowledged as the top rating categories.
My first point is to come back to ATH's initial question. I give more value to GR than CP as I don't like that RedskinsSuck has no CP for being n°11 on WarGear Warfare despite a score of 1677 while BlackDog is getting 20 points on Koprulu Sector with a score of 1513 (just an example BD ;) ). As more players are around the 2000 score now, may be the scale to get CP should be changed (for example 1500+ for 6 points, 1600+ for 8, 1700+ for 10, 1800+ for 12, 1900+ for 15, 2000+ for 20) ?
In addition to the hoped-for achievemnt system, I agree it would be very cool to have an "overall skill" stat. We all know that such a figure will never be perfect, far from it. But let's try something to improve M57 idea's.
Championship points : I also believe like BD said that that stat is given too much weight in this calculation. Using log(CP) instead of CP (with a minimum of 1) would be my suggestion.
H% : like BD said, it's less transiant then the GR. So I believe it's an interesting figure.
Tournament points : I see no reason at all not to take them into accounts. Players not playing tournaments would be given a 1000 score, so they would not get penalized.