| Forum Index > LoL General |
|
Sonnington
United States1107 Posts
On September 04 2015 05:59 Ethelis wrote: It's rare in my games. Funnily enough I've typed "from?" and I've gotten a handful of people who picked up on that, probably more replies to that than any glhf. I've seen people say that before. What's that in reference to? | ||
|
Fildun
Netherlands4123 Posts
On September 04 2015 06:50 Sonnington wrote: I've seen people say that before. What's that in reference to? I believe it's just asking from which country/state you are. | ||
GrandInquisitor
New York City13113 Posts
On September 04 2015 06:41 Sufficiency wrote: I looked at this before and did some scoping. I ran some machine learning algorithms and I was able to predict the outcomes of matches (ranked) with roughly 55% accuracy using only the champions picked and a rough elo bracket estimate (e.g. gold). Comparably I can do so for ARAM with around 65% accuracy. This by itself is only somewhat useful, but it can be leveraged to figure out the ideal bans and picks in a given situation using the predicted winrates. Let's talk about this. I envision a site that: 1) takes in input, like op.gg/multi, a list of summoner names for your team; 2) responds with your team's best win rate + play rate champions; 3) uses that to inform pick/bans. Specifically, it produces this table: Your team high win rate / high play rate + general high win rate / high play rate = priority pick/ban Your team low win rate / low play rate + general high win rate / high play rate = priority ban Your team high win rate / high play rate + general low play rate = priority pick but not ban It then assembles a list of the priority bans, which is then adjusted live by the other team's bans as well as what champions are remaining. (Such as prioritizing Jinx ban if Vayne is banned, per GMD's blog.) | ||
|
Gahlo
United States35172 Posts
On September 04 2015 05:37 GrandInquisitor wrote: Isn't reasoning like this exactly why relying on intuition is dangerous? We are primed to remember all those 1v5 games by BoxBox smurf, and not the games where Riven was useless. Meanwhile we are NOT primed to remember those far more frequent games where having a Janna on the other team meant their backline was literally unkillable. Worrying about the black swan is not a positive EV play. Sure, once in a while you'll run into a challenger smurf on Riven/Rengar. And once in a while your bot lane won't ever connect to the game. That sucks, but you can't plan your strategy around that. Losses to Riven smurfs may feel bad, but they're not any different than the far more frequent losses to Janna that you don't remember. Also, the whole "Riven just has lots of people play her who aren't good at her dragging down her skill rate" is a fallacy. The average Riven player has 144 Riven games played this season; the average Janna player has 82. Janna's win rate for players that have played fewer than 5 games on Janna this season is higher than Riven's overall win rate (and for Riven players with 50-125 games played). I have about 6 games on Riven this season. I'm 1-5 because I was stupid rusty and my mechanics are shot. There are plenty more people that play Riven sparingly than there are people who one trick pony her. On September 04 2015 04:09 Ansibled wrote: Random question, does anyone ever type glhf in your games? Nobody ever seems to reply to me when I do it lately. I do, sometimes, if I remember. Generally when there's playful pre-minion banter. | ||
|
GrandInquisitor
New York City13113 Posts
On September 04 2015 06:53 Gahlo wrote: I have about 6 games on Riven this season. I'm 1-5 because I was stupid rusty and my mechanics are shot. There are plenty more people that play Riven sparingly than there are people who one trick pony her. I mean, you can try to generalize from your own experience, but this is a demonstrably false statement, both objectively and when compared to other champions. http://champion.gg/champion/Riven http://champion.gg/champion/Janna The first thing to notice is that more than half of all Riven ranked games are played by Riven players that have played 50+ ranked games with Riven this season. Moreover, not only does the average Riven player have nearly 2x as many games on Riven as the average Janna player, only 7% of Riven players have < 5 games on Riven (compared to 12% for Janna); 11% have 5-15 (compared to 17% for Janna); 24% have 15-50 (compared to 29% for Janna). But there are far more Riven players with 50+ games than Janna players with 50+ games. This is exactly the reason why we have data, so we can call out people that just pull anecdotes from their ass and assume it to be true generally. | ||
|
killerdog
Denmark6522 Posts
There was an analogy for elo that I find fitting. Your elo is just currency, you pay elo to learn from better players and worse players pay elo to learn from you. I'd prefer to maximise relevancy of each game, rather then throw away 5 games just to get easy wins in 6 other ones. And i'm sorry, you're right I was joking. Janna has literally zero unique play patterns, strengths or weaknesses you can learn, if she's in the game you just arbitrarily have 5% less chance of winning completely unrelated to how you play the game -.- Also where does champion.gg get it's stats from, because if you're including gold or plat rivens as part of your argument that riven isn't ban worthy, then this whole conversation is pointless as she sucks at those elos. Not all data is magically relevant just because it has numbers in... | ||
|
Caiada
United States3052 Posts
I think banning to have a more enjoyable game is a more legit strategy than a marginally higher win chance from banning whatever's high winrate, but if I had to guess, that's too intangible. | ||
|
GrandInquisitor
New York City13113 Posts
On September 04 2015 07:13 killerdog wrote: And i'm sorry, you're right I was joking. Janna has literally zero unique play patterns, strengths or weaknesses you can learn, if she's in the game you just arbitrarily have 5% less chance of winning completely unrelated to how you play the game -.- I mean you literally said Riven has no counterplay. You put words in my mouth and strawman me about Janna, but I am literally quoting you: "Losing to janna is something you can outplay, she's stronger then other supports, but there's counterplay etc, you just have to play slightly better/differently. Losing because of a carrying or feeding riven there's nothing you can do." If you have such issues with Riven that you think there's "nothing you can do" against her, you are just banning Riven too much and need to learn to play against her. See what I just did there? And don't give me this elitism MMR shit. For starters, champion.gg is Plat+. Second, the gap between Janna and Riven is only bigger at Diamond and Master. This in fact suggests that Riven is more deserving of a ban than Janna only at Bronze to Gold. I bet you aren't Bronze to Gold, right? (See what I did there? Insinuate that I'm right because you're just low elo? Although I'm not the one who wrote there's "nothing you can do" in the face of a carrying or feeding Riven ...) | ||
|
Goumindong
United States3529 Posts
And i'm sorry, you're right I was joking. Janna has literally zero unique play patterns, strengths or weaknesses you can learn, if she's in the game you just arbitrarily have 5% less chance of winning completely unrelated to how you play the game -.- :roll: Yea and so does any other champion. Your supposition here is that champions win rates do not inform as to how strong they are. Or that their play rate data doesn't inform as to the likelihood of finding one. Or that, for some reason special to janna, her win rate and play rate don't matter. None of these things make any sense. Its entirely possible for you to be better or worse against particular champions. And if you have your data then you can do that. But most people don't have enough data to do that kind of analysis. And they certainly don't have enough data on the current patch to do that kind of analysis. And so the next best thing you have is the general data. Because at the very least you should be centered around it. | ||
|
JimmiC
Canada22817 Posts
| ||
|
Sufficiency
Canada23833 Posts
On September 04 2015 06:52 GrandInquisitor wrote: Let's talk about this. I envision a site that: 1) takes in input, like op.gg/multi, a list of summoner names for your team; 2) responds with your team's best win rate + play rate champions; 3) uses that to inform pick/bans. Specifically, it produces this table: Your team high win rate / high play rate + general high win rate / high play rate = priority pick/ban Your team low win rate / low play rate + general high win rate / high play rate = priority ban Your team high win rate / high play rate + general low play rate = priority pick but not ban It then assembles a list of the priority bans, which is then adjusted live by the other team's bans as well as what champions are remaining. (Such as prioritizing Jinx ban if Vayne is banned, per GMD's blog.) OK that's not actually what I have in mind. My system was more like: suppose for now that you are last pick. Which champion will give you the highest likelihood of winning, without any historical informations about the player except a rough elo estimate? Clearly you should pick a champion that synergizes with your own team and/or counters the enemy team, and the statistical model takes care of that. Anyway your idea is interesting too, but I am on my phone so I won't elaborate on it at this point. | ||
|
Ethelis
United States2397 Posts
On September 04 2015 06:51 Fildun wrote: I believe it's just asking from which country/state you are. + Show Spoiler [Particularly at 2:28] + | ||
|
Gahlo
United States35172 Posts
On September 04 2015 07:10 GrandInquisitor wrote: I mean, you can try to generalize from your own experience, but this is a demonstrably false statement, both objectively and when compared to other champions. http://champion.gg/champion/Riven http://champion.gg/champion/Janna The first thing to notice is that more than half of all Riven ranked games are played by Riven players that have played 50+ ranked games with Riven this season. Moreover, not only does the average Riven player have nearly 2x as many games on Riven as the average Janna player, only 7% of Riven players have < 5 games on Riven (compared to 12% for Janna); 11% have 5-15 (compared to 17% for Janna); 24% have 15-50 (compared to 29% for Janna). But there are far more Riven players with 50+ games than Janna players with 50+ games. This is exactly the reason why we have data, so we can call out people that just pull anecdotes from their ass and assume it to be true generally. No, that's the average Plat+ Riven player. Most Riven players at that level are Riven mains. There aren't a bunch of heavy Janna players because Janna is a support, a position that the vast majority of people try to avoid playing. Meanwhile Riven is traditionally a top laner, a position that(at least at my low level) is craved only behind mid. People that get "forced into" top won't play Riven because it's far easier to play much simpler, consistent champions. It's fine if you want to draw conclusions based upon data, but make sure you're applying it to the small subsection of the ladder that it actually belongs to. | ||
|
JonnyLaw
United States3482 Posts
On that note, I also just ban champs I don't want to play against or find annoying. Janna is one of those champs if I'm playing someone who wants to dive the backline and not eat Janna q to the face. I could play around it better, but then there's still ult and move speed. Also ban obvious op of course. edit - someone asked about Voli top earlier. It's being played in Shiphtur's game that's about to start. | ||
|
killerdog
Denmark6522 Posts
On September 04 2015 07:22 GrandInquisitor wrote: I mean you literally said Riven has no counterplay. You put words in my mouth and strawman me about Janna, but I am literally quoting you: "Losing to janna is something you can outplay, she's stronger then other supports, but there's counterplay etc, you just have to play slightly better/differently. Losing because of a carrying or feeding riven there's nothing you can do." If you have such issues with Riven that you think there's "nothing you can do" against her, you are just banning Riven too much and need to learn to play against her. See what I just did there? And don't give me this elitism MMR shit. For starters, champion.gg is Plat+. Second, the gap between Janna and Riven is only bigger at Diamond and Master. This in fact suggests that Riven is more deserving of a ban than Janna only at Bronze to Gold. I bet you aren't Bronze to Gold, right? (See what I did there? Insinuate that I'm right because you're just low elo? Although I'm not the one who wrote there's "nothing you can do" in the face of a carrying or feeding Riven ...) Do I really need to explain how champions like riven and rengar can snowball harder then other champions, like janna... If you want an example check out gross gores stream right now, enemy kog is 8-1 but can't do anything at all because gross gore is getting hard carried by his rengar. There aren't many champions that can shut the entire map like that from level 6. If you won't listen to anything else i'll put it in terms of statistics. Banning champions like rengar reduce variance in the rng of soloq, If you think you actually deserve to win games then that should be a good thing. | ||
|
Gahlo
United States35172 Posts
On September 04 2015 07:25 JimmiC wrote: This seams like a super interesting idea. I would be very into if they did something like this! They did something similar with Nunu getting different buffs depending upon what types of monster he's smote. | ||
|
killerdog
Denmark6522 Posts
On September 04 2015 07:55 Gahlo wrote: They did something similar with Nunu getting different buffs depending upon what types of monster he's smote. Is it smited or smote? Like, given that we're using smite meaning to use your smite, (where smite is a noun,) is the conjugation still the same as with the original verb smite? | ||
|
Sonnington
United States1107 Posts
On September 04 2015 06:51 Fildun wrote: I believe it's just asking from which country/state you are. I thought so too. I would see it from time to time on EUW, but whenever I'd say it people would act confused. | ||
|
Frolossus
United States4779 Posts
On September 04 2015 08:09 killerdog wrote: Is it smited or smote? Like, given that we're using smite meaning to use your smite, (where smite is a noun,) is the conjugation still the same as with the original verb smite? act of smiting is a verb so smote | ||
|
Goumindong
United States3529 Posts
On September 04 2015 07:54 killerdog wrote: Do I really need to explain how champions like riven and rengar can snowball harder then other champions, like janna... If you want an example check out gross gores stream right now, enemy kog is 8-1 but can't do anything at all because gross gore is getting hard carried by his rengar. There aren't many champions that can shut the entire map like that from level 6. If you won't listen to anything else i'll put it in terms of statistics. Banning champions like rengar reduce variance in the rng of soloq, If you think you actually deserve to win games then that should be a good thing. It doesn't matter how hard they snowball. It only matters if they win the game. ' A Soloqueue game is a Bernoulli random variable. It has an expected value of p = the probability of winning and a variance of (1-p)p Snowballing? Doesn't matter, doesn't effect the variance. Only the effect of the champion on the win rate affects the variance. Win rates which are closer to 0 or 1 will reduce variance.[I.E. very high win rate or very low win rate champions]. edit: So in a way yes, banning riven reduces variance. But that isn't reducing variance in a way you want if they're picking stronger champions as a result of it. | ||
| ||
|
|
BSL
Replay Cast
RSL Revival
herO vs TriGGeR
NightMare vs Solar
uThermal 2v2 Circuit
BSL
IPSL
eOnzErG vs TBD
G5 vs Nesh
Patches Events
Replay Cast
Wardi Open
Afreeca Starleague
Jaedong vs Light
[ Show More ] |
|
|
EPT
.
