I see, so you're talking about some sort of "activity ladder" ?
Yes that's a good idea, but I dont know how difficult it would be to set up, considering players practicing on unranked servers, aborted matches etc. I dont think they're taken into account.
But maybe some stats from the database as it is today could be enough to do it, hopefully.
- A 3rd graph on each player profile indicating how many duels a player has played in the last 30 days. Could also be how many hours played, but I think how many duels is easier and sufficient.
- Incorporate a separate ranking for #tdmpickup stats in addition to the QLRanks official ELO (?)
If the website had a quote on every players page like the most commonly used sentence or word, I think mine would be "kjhdfjklsjflkvsdjvlk !"#€%&/()" or very close to it.
Player search needs error control and input trimming, as it allows the user to search without typing anything and returns an ugly ASP error. The same happens if you include a space at the end of a player's name and search for it.
I think the possibility to browse all the matches between two specifics players could be really neat, you could probably add a lot of interestings stats to the page too (favorite maps, average score etc).
Also you have a bug where the stats page is displayed on the page of any player with a nickname beginning by stats.
The only reason this doesn't exist already is because neither Szr or Myself understand css / web design. If someone could provide a template for these things, they would be made extremely quickly.
It wasn't "frequently" broken, it was just broken, always showing wrong result.
You will be delighted to know that this problem has been fixed just a week ago.
The historical data wasn't recalculated, so the left part of the graph will stay as it is. But each game you play now inserts correct graph points. In a month we'll all forget about this problem that survived 1,5 year.
Filter to results from the last month. Its not ideal but looking at a few results it looks kind of ok. I don't think you can search for 13** using google anymore. Maybe another engine can.
The site is setup horribly for search however - massive page duplication surely causes problems. For example each gametype has a different url for each player that contains the same information, albeit for a different gametype.
Not a bug, it's by design. It shows games in the order they were added. Easy to get any sequence you like by updating your opponents instead of yourself.
there are some really big mistakes. I joined a server whose elo was so much higher than me in CA, and we lost, after the game I lost 292ish points.
After that I wont a game still with higher elo and won 17 points, then lost again to still giher elo and lost 170 elo.. I was like :D what is this
make it possible for viewers to choose which match they want to see / what players they wanna spec =o
also sometimes makes me feel bad when qlranks joins and im just messing around in warmup with lettu :E make it so that viewers could then choose another match instantly!
Could my top 5 beaten list rather be top 10? Or have a link to a page showing top 100 or whatever
Right side, top in country, could it show a second column showing their world rank too?
The home page is great and shows lots of useful data.. how about this same page, but reflecting per country - ie: the South Africa home page could show top 5 currently playing, last scanned, followed by our top duel/ca/tdm/ctf. The top 100 is really great and is still needed, its just not enough :(
How do I know if a server is ranked or not? I'm sure I've played on non-ranked servers but the results have still been captured in QLRanks...
Otherwise, its the most awesome site in the world, well done team! Wish we had a site like this back in q2/q3 days!
I thought that quitting games now still affects your elo? I swear I saw something about 'quitters won't prosper anymore' in one of the qlranks irc topics, but I could be mistaken
One that looks like the current one, just stylized a bit, and second is long thin stuff that goes on top of player score/ammo/armor area.
Szr should have both.
Can you do something with "page" links in match history?
I use two browsers atm - one is webkit based and the other one is xulrunner (firefox) based. I know they aren't your standard browsers but in both of them these links don't work. There must be something wrong/peculiar in the code.
They are modern, get full scores in generic tests and don't show any problems with other sites (again they are linked against most recent webkit and xulrunner).
Yes this is browser specific bug but as long as you don't use too fancy stuff this shouldn't happen.
__doPostBack is just a JavaScript function defined in a script tag on the page... it is part of Microsoft's ASP.NET, and not QLRanks-specific at all. Might want to double-check things on your side (add-ins, etc.).
I have idea - remove TDM, CTF, CA elo. Current crap realization is only confusing people.
You can better make some formula based on delivered damage, net damage, net frags.
Yes and no. idsoft didnt even manage to implement a decent system to begin with, so they wont be able to do what you suggested. Upside of this if it does get put in, itl probably work quite a bit better if its a smart system. One that can avoid putting 5 railerheavy players on the same side after a shuffle :P
just when i thought it couldn't get dumber (elol in ca) you've now added another team mode elo (ctf), congrats.
why are you doing this? pubs are getting worse day by day, just have a look at how many whores are posting here lately threads like "why did i lose elo" or "fix ca elo", or just the quitters in games
Something I always wondered is why not add a FFA ladder ? I think FFA is still a very active mode and the calculations would be pretty straightforward and accurate, at least way more than CA or anything else.
Plus it seems that the FFA top players had a pretty competitive mindset so they would surely be happy about that.
I agree, plus from what I've experienced FFA players are probably the least faggy of them all, so they will appreciate it's implementation more than everyone else, barely whine at all if there are bugs during the early stages, and the rankings should be pretty accurate and competitive because they are also the least likely to tierslum.
Once you finally addressed some of the DB and security errors, I like to share some ideas with you.
1. I never understood why is the graph with ELO rank change only for last 30 days? Does it really show players progress? I think it doesn't and is of no value atm. It should be like year or even more, and make the graph bigger.
2. I think there might be something like highest ELO achieved: XXXX, date.
3. Also you can make statistics like ranked "fastest improving players" or "the fastest improving players in each country", "most stable player", you can really play with this, you can even do e.g. "the best railgunners, shafters etc." based on acc, use and players ELOs against and stuff like that. Then the best should have it written in profile:e.g. in K1llsen profile: The rail grandmaster you know..
4. Make it more fun, I mean instead of the T.S stat who IMO nobody much cares about anyways, lets make funny sentences: "enjoys raping noobs", "enjoys playing tough competition", "Likes to be spanked by stronger players.. " :)
5. What about giving some special achievment to player reaching 1st place in the ladder and display it in profile forever? Like "reached number one rank on date. Also I'm not sure wether Master, Grandmaster etc. should stay by player forever, once he reaches it, I think it should.
6. I sometimes saw score like -1 vs 30, with the ELO points going for -1 player :) I mean maybe the stronger player had lag or something, my idea would be in those cases make it +-0 ELO for both.
7. OK and there's visibly one issue with players ELOs. You can see it has problems reflecting skills when you comparing players from different locations/continents. Now looking at the ranks I think e.g. the brazilian player Flesh1 or Szarko would be defeated by european high skill players and raped by euro/us top players, despite having such a high ranking. I don't know about ELO starting rules, but my idea would be to base it on player base in a specific location, assuming those players will mostly play each other. SO e.g location with 5000 active players: ELO 1200; with 500 active players ELO 1100, I won't tell you the numbers exactly but you may be now more than capable to collect some data from the history. Players visiting other continents (Fraze, czm..) might be quite good indicator about skill vs ELO.
About your last point, changing the starting Elo from 1200 to 1100 won't do anything because the Elo system doesn't work like that.
What you want is changing the K-Factor making it harder and longer for these players to reach the same numbers as Europeans or North Americans.
Right now I think QLRanks use a K-Factor of 30 or 32 for everyone, putting it at 15 or 16 for South Americans would mean that they earn 50% less points as other continents when they win but also lose 50% less when they are defeated, so really I don't know what the overall impact of this would be.
Basically your problem is something that can't be solved because it isn't one, they are deserving their points as much as everyone else because of how the Elo system works, but of course because both the playerbase and the overall levels are lower, their 2000 score isn't the same as EU/NA 2000 score.
Are you really sure it won't do anything? Just read about ELO and I think it might just work. About Kfactor I understand - Higher K-factor means good players will reach high score and bad players low score faster, is it really what we want? I don't think so. Please if I am missing something, tell me:)
If I understand you correctly, your point is that if the starting score was lowered from 1200 to 1000, everyone in the ladder would have 200 less points ?
I'm afraid that's not exactly how it would work, because when you reach like 2000 Elo you are not playing against starting players anyway, sure it might have an impact on how many people are able to reach 2000 but if the playerbase is big enough I don't see how that could be a problem.
But that's certainly a very interesting point, there are a lot of aspects of the Elo rating system that are critiqued and the starting score is definitely one of them.
And what I was saying about lowering the K-Factor is that it would simply take longer to reach the same scores.
Let's consider this example : Player A = 1800 Elo, Player B = 1500 Elo, K-Factor = 30, they play 3 matches and Player A wins all 3, the final scores are Player A = 1813 Elo and Player B = 1487 Elo.
Now if the K-Factor is lowered to 15, the final scores are Player A = 1806 Elo and Player B = 1494 Elo.
As you can see it would take them twice as much time/twice as much games to reach the same score as their counterparts in other regions.
I don't think so. Apart from the initial re-calculation of all matches to get Elo per map, of course. It could get ugly on disk reads depending on how they would modify their database to integrate that, though. But the main problem is probably the time it would take to develop it (changes in the database, the program calculating Elo and the website's design).
Well, I don't think you need any changes in the DB at all.
Since all the matches are stored, all it takes is a different algorithm to calculate it.
Let's say at the start of QLranks everyone is at 1200 elo, people play and elo changes. That's how it develops over time.
Just do the same thing for every single map. Go back in history and set elo to 1200 for each map, and take the games played against others on this map.
I think all it needs is to copy/paste the algorithm and add just one statement like "AND map = bloodrun" or something like this.
They would have to get additional table to store player/map combinations (and a big-ass index on that). That's a DB change, storage requirements change, with potential performance impact not when updating, but upon displaying.
(I think they still use MySQL and not some key-value store?)
And on top of the small db/code work, they will need a new UI, that may be the worst part :)
(I hope it happens though - can't wait for the whine about ratings for little-played maps :))
If you don't play quake for 30 days, you will not be personally updated/scanned by qlranks until you are found to have played a game against someone who is active.
Can the ELO graph have a smaller range? Right now the graph spans about 500 ELO but if your ELO hovers around the same value then the graph will look relatively flat.
CTF
page: /ctf
page: /ctf/country.aspx
fix: On player's CTF page, the Country's top 20 also does not work.
fix: ctf map thumbnails do not link to match URLs
Player
fix: make country current. I'm german in duel, american in CA, russian in CTF... (donkz)
add: In world rank table only display people from the same region.
Global
fix: drop-down doesn't work 90% of the time(Chrome). First i look up a player under duel. Works ok. Then i switch drop-down to TDM - instead queries duel again.
add: next to player's nick - add buttons to easily switch between pages <mode>/player/nickname
The graphs really need improvement. Even the old qlranks site had better graphs... something that looks/works like the google-finance graphs would be awesome
also how far back is elo data stored ? would be so cool to see historical trends... even if u can only retrieve that data from api call or something, it should be possible - it's silly that we can only see 30 days back and that theres no way to obtain that data otherwise unless we make scripts to mine it ourselves [which i've done already, but only have 1 month history...] xD
- somehow visually emphasize current page number -_-
- Win % per-map in profile
- some global stats, like MAU (monthly active users), or percent of pro\premium players
mmmhh, maybe during american hours with some rule "if elo of two players < 3600, go to highest ranked us CA serv"
just maybe, I'd rather get some re-run's of the highest elo games of the day but the qlr team would have to code something to get the demos and play them, it's a bit much I guess