|
henne posted:Luna can you post your trade link again? I have some cases I can send you to open for tourny winnings. Any preferance on cases? No chroma but I have huntsman, phoenix, vanguard, some other poo poo too probs. doesn't matter what case it is so long as it contains items for which i can add to the pool https://steamcommunity.com/tradeoffer/new/?partner=72957442&token=UgACfGbO
|
# ? Feb 17, 2015 00:38 |
|
|
# ? May 30, 2024 07:40 |
|
How long until I can bet weapons from a Chroma case? If its not for a few weeks I have something nice for the prize pool.
|
# ? Feb 17, 2015 00:40 |
|
Luna Was Here posted:doesn't matter what case it is so long as it contains items for which i can add to the pool Just sent you about a dozen cases
|
# ? Feb 17, 2015 01:22 |
|
Ahdinko posted:Just sent you about a dozen cases As did I.
|
# ? Feb 17, 2015 01:37 |
|
Sent a sexy awp from my an alt account. Can you use it as a prize for the Silvers and Novas? Most kills in a match or something? Best clutch as voted by all players?
|
# ? Feb 17, 2015 01:45 |
|
Just asking to see I guess, minute thing but whatever; is there any way to set your server browser to default to the favorites tab when it opens? Clicking to change tabs is hard.
|
# ? Feb 17, 2015 03:23 |
|
Ahdinko posted:Just sent you about a dozen cases Fauxshiz posted:As did I. bpower posted:Sent a sexy awp from my an alt account. Can you use it as a prize for the Silvers and Novas? Most kills in a match or something? Best clutch as voted by all players? i now have exactly 40 cases, which is the max i can crack open with the $100 on wednesday. and thanks to bpower we even have a guaranteed good item!
|
# ? Feb 17, 2015 05:16 |
|
https://www.youtube.com/watch?v=vxoQmwf87LU
|
# ? Feb 17, 2015 07:10 |
|
Skuto posted:If you're asking for a rating system to try to incorporate individual performance metrics, while the actual game is team based with a single outcome, you should be aware that you're asking for faster rating adaption at the cost of robustness and most likely accuracy too. afaik it actually does this though, people have tested leveling new accounts together in the exact same games, and their ranks will differ based on performance..
|
# ? Feb 17, 2015 08:00 |
|
This makes sense because they have to show a rating after only 10 games, so grabbing some data is more useful than not having any. Once you played more games there's little reason to take the individual metrics into account in any significant measure; you have the game outcome data which is way more reliable. For what it's worth Dota2 is documented by Valve to behave exactly in this way: influence of personal performance tapers off as uncertainty of the rating drops (i.e. You've played lots of games recently). You can calculate accurate ratings based on game outcomes only if you have a sufficent amount of them, this is really not even arguable. If you have this data, why would you add gameable metrics back in? That makes no sense. If you play regularly, your rank reflects your skill. Your individual performance is taken into account by making your team more likely to win if you're above average. Edit: Someone with a shitton of player/match data should be "easily" able to verify that you can predict the match winner much more accurately via an outcome based Elo-like system rather than looking at past ADR or RWS. Hiowf fucked around with this message at 09:03 on Feb 17, 2015 |
# ? Feb 17, 2015 08:55 |
|
cs is a game about winning rounds not about getting kills. getting kills usually wins you rounds though.
|
# ? Feb 17, 2015 09:01 |
|
Slaapaav posted:cs is a game about winning rounds not about getting kills. getting kills usually wins you rounds though. Deep
|
# ? Feb 17, 2015 10:26 |
|
its actually possible for a team to get more kills than the other team and still lose the game. why should your sick kda have any effect on your mmr?
|
# ? Feb 17, 2015 12:31 |
|
Slaapaav posted:its actually possible for a team to get more kills than the other team and still lose the game. why should your sick kda have any effect on your mmr? That's a bit too black and white. The goal of a rating is to predict game outcomes. You could have lost that game because the rest of your team is terrible. Does this mean the likelihood of you (in isolation without those teammates) winning games has now necessarily dropped? It's a balance of information. Some of the information is solid and unbiased (match won or not) and some isn't (kda/mvp/whatever). If the only information we have about a player is that he played 1 game in a team with an average MMR of 1500 against opponents with an average MMR of 1500 and scored 60/10/20, yet tied the match, then the expected ratings distribution for him is very different from a player that has played 100 games with average team MMR of 1500 against opponents with average team MMR of 1500 and winning half of them. Note also that in the latter example knowing the score of this player in those games shouldn't change your opinion much about what the rating would be.
|
# ? Feb 17, 2015 12:49 |
|
Skuto posted:The goal of a rating is to predict game outcomes. Is it?
|
# ? Feb 17, 2015 12:56 |
|
thats a lot of words to say the same thing i just said
|
# ? Feb 17, 2015 12:57 |
|
kcer posted:Is it? Yes. (In the context of CS:GO, you use that prediction to try to ensure a "close" game) quote:thats a lot of words to say the same thing i just said I'm answering your rhetorical question to show the cases where it does make sense to account for kda.
|
# ? Feb 17, 2015 13:03 |
|
Skuto posted:Yes. Is there a "prediction" other than a draw when mmr works?
|
# ? Feb 17, 2015 13:16 |
|
kcer posted:Is there a "prediction" other than a draw when mmr works? With perfect MMR over infinite games all matches would be draws. The closer all matches are in general to a draw, the better the algorithm works.
|
# ? Feb 17, 2015 13:23 |
|
kcer posted:Is there a "prediction" other than a draw when mmr works? Of course MMR will always aim towards a draw, or as close to a draw as it can. The inherent variance and the inaccuracy of MMR means it will rarely succeed in causing a draw. If you think about it for a moment there is no other way it could operate, unless you think it operates on a system to try and create unequal games.
|
# ? Feb 17, 2015 13:27 |
|
Jeza posted:Of course MMR will always aim towards a draw, or as close to a draw as it can. The inherent variance and the inaccuracy of MMR means it will rarely succeed in causing a draw. If you think about it for a moment there is no other way it could operate, unless you think it operates on a system to try and create unequal games. b-b-but ENFORCED 50% WINRATE
|
# ? Feb 17, 2015 13:28 |
|
Just click heads, simple.
|
# ? Feb 17, 2015 13:32 |
|
I mean, a coin flip is a perfect MMR simulation of two completely evenly matched teams. If you flip a coin a billion billion times, the overall outcomes of heads/tails is 50%. Flip a coin 30 times and although 15 heads and 15 tails is certainly possible, it is not more likely than every other possible outcome, not by a long shot.
Jeza fucked around with this message at 13:36 on Feb 17, 2015 |
# ? Feb 17, 2015 13:34 |
|
And then someone on your team abandons after 4 rounds, BOT Jack keeps buying random guns and running up mid and you lose 16-8. And then you get downranked.
|
# ? Feb 17, 2015 13:43 |
|
Katana Gomai posted:And then someone on your team abandons after 4 rounds, BOT Jack keeps buying random guns and running up mid and you lose 16-8. And then you get downranked. Admittedly it's a mystery to me how you would deal with abandons in a graceful manner. To prevent exploiting the system you probably have little choice but to act as if nothing has happened. If not people could disconnect when they're about to lose to avoid down ranking.
|
# ? Feb 17, 2015 14:07 |
|
Luigi Thirty posted:With perfect MMR over infinite games all matches would be draws. The closer all matches are in general to a draw, the better the algorithm works. You also can't guarantee the teams have identical MMR unless you want to queue forever. To get some idea of what matchmaking without MMR looks like pick random captains in the next goonman and let them pick their teams randomly. Hilarity will ensue and goon tears will be shed.
|
# ? Feb 17, 2015 14:12 |
|
Skuto posted:Admittedly it's a mystery to me how you would deal with abandons in a graceful manner. To prevent exploiting the system you probably have little choice but to act as if nothing has happened. If not people could disconnect when they're about to lose to avoid down ranking. They could introduce a low priority queue system like the one in DotA, and on top of that decrease the MMR loss/gain from the match that had an abandon (if the team with an abandon loses.) Still, current system isn't too bad. Only had one abandon in like 20 games, and he got a 7 day cooldown for it.
|
# ? Feb 17, 2015 14:32 |
|
Jeza posted:They could introduce a low priority queue system like the one in DotA, and on top of that decrease the MMR loss/gain from the match that had an abandon (if the team with an abandon loses.) Still, current system isn't too bad. Only had one abandon in like 20 games, and he got a 7 day cooldown for it. I'm still in silver and from my last 10 games, 8 had an abandon. I don't doubt that it becomes less of an issue the higher you go but to most people who just started playing CS:GO, a match which finishes 5v5 is the exception, not the norm.
|
# ? Feb 17, 2015 14:37 |
|
Skuto posted:Admittedly it's a mystery to me how you would deal with abandons in a graceful manner. To prevent exploiting the system you probably have little choice but to act as if nothing has happened. If not people could disconnect when they're about to lose to avoid down ranking. I think the solution to this is easy. If someone abandons before, say, the 5th or 6th round, it won't affect your MMR. After that and it will, but maybe at a slightly reduced rate depending on the score (If you're getting 14-0'd and you get abandons, it should go through as normal, etc.) Katana Gomai posted:I'm still in silver and from my last 10 games, 8 had an abandon. I don't doubt that it becomes less of an issue the higher you go but to most people who just started playing CS:GO, a match which finishes 5v5 is the exception, not the norm. Add me on Steam, I'll play Matchmaking with you: http://steamcommunity.com/id/fortheemperor/
|
# ? Feb 17, 2015 14:40 |
|
The Jumpoff posted:I think the solution to this is easy. If someone abandons before, say, the 5th or 6th round, it won't affect your MMR. After that and it will, but maybe at a slightly reduced rate depending on the score (If you're getting 14-0'd and you get abandons, it should go through as normal, etc.) Not sure. You could still bail if it goes 5-0 and then there's the issue of maps like nuke etc.
|
# ? Feb 17, 2015 14:42 |
|
Jeza posted:They could introduce a low priority queue system like the one in DotA, quote:I'm still in silver and from my last 10 games, 8 had an abandon. FWIW you get exactly the same experience playing DotA with a new account. I mean it's hard to get people who have nothing invested in the game to stick with it, regardless of what mechanics you try to invent.
|
# ? Feb 17, 2015 14:45 |
|
Skuto posted:Not sure. You could still bail if it goes 5-0 and then there's the issue of maps like nuke etc. Well the person who abandons obviously gets a loss, it's the other 4 we're talking about. For groups, if one person abandons, everyone who queued with them gets the loss.
|
# ? Feb 17, 2015 14:46 |
|
Jeza posted:Of course MMR will always aim towards a draw, or as close to a draw as it can. The inherent variance and the inaccuracy of MMR means it will rarely succeed in causing a draw. If you think about it for a moment there is no other way it could operate, unless you think it operates on a system to try and create unequal games. Don't even remember what I was getting at. I know how mmr works, I just thought the wording of it predicting outcomes was a bit weird, but is, in a roundabout way, correct.
|
# ? Feb 17, 2015 14:52 |
|
Katana Gomai posted:For groups, if one person abandons, everyone who queued with them gets the loss. This is probably not even necessary - they're already playing with a guy less :-) Reducing the MMR adjustment depending on how far along the match the abandon happens sounds like it'd work, though realistically that's such a minor tweak at that point that you're not likely to feel much difference in practice. You could still downrank after losing a game with a quitter etc.
|
# ? Feb 17, 2015 14:54 |
|
In the short run all sorts of poo poo can happen with quitters, cheaters, smurfs, and so on. But that poo poo happens to your opponents too. In the long run it will even out. Just focus on making good decisions and maintaining good vibes in every game regardless and your rank will follow.
|
# ? Feb 17, 2015 14:59 |
|
I've had the craziest gambling luck this past week. Decided that I didn't care about skins anymore so I placed my most expensive skin (AWP Graphite worth about 14€) on Property against Penta and won a FN Deagle Blaze and some other smaller stuff. From there I somehow snowballed my winnings into about 150€ worth of skins on a series of impossibly lucky bets, like VP winning against nV in a bo3. I can finally buy myself a knife!
|
# ? Feb 17, 2015 15:07 |
|
Skuto posted:To get some idea of what matchmaking without MMR looks like pick random captains in the next goonman and let them pick their teams randomly. Hilarity will ensue and goon tears will be shed. goonmans didn't have player ratings for the first year or so and it was actually fine Also, trueskill accounts for abandons by essentially changing the expected outcome to a loss for the team with an abandon. So if you play 4v5 and lose it does not have a significant effect on your rating, but if you win you get a larger increase than usual.
|
# ? Feb 17, 2015 15:17 |
|
I'm trying to get my rank back and shake off the competitive rust. Considering I've lost 3 in a row I'm probably down to silvers, but I'm not bad, honest. If anyone wants to matchmake tonight, add me and when I get home from college we'll shoot some nerds. (5 cst) http://steamcommunity.com/id/Jassassin
|
# ? Feb 17, 2015 15:32 |
|
J posted:In the short run all sorts of poo poo can happen with quitters, cheaters, smurfs, and so on. But that poo poo happens to your opponents too. In the long run it will even out. Just focus on making good decisions and maintaining good vibes in every game regardless and your rank will follow. EXACTLY. Blaming your rank on anything besides "I'm not winning enough" is seeking out excuses and is the opposite mentality needed to get better. If you're good enough to rank up being handed the occasional shithead teammate will not slow you down.
|
# ? Feb 17, 2015 15:47 |
|
|
# ? May 30, 2024 07:40 |
|
astr0man posted:goonmans didn't have player ratings for the first year or so and it was actually fine Yeah but did you have team picks like you do now? That naturally keeps the teams balanced provided the captains have *some* idea who's how strong. quote:Also, trueskill accounts for abandons by essentially changing the expected outcome to a loss for the team with an abandon. So if you play 4v5 and lose it does not have a significant effect on your rating, but if you win you get a larger increase than usual. Have to put a strong punishment on the abandon though, else this is gameable in the manner described a few posts ago. Hiowf fucked around with this message at 15:51 on Feb 17, 2015 |
# ? Feb 17, 2015 15:48 |