EPTI'm not commenting on whether people are applying the bare minimum of critical thinking to understand the game or not. That's a separate issue.
[Patch 4.5] Rune Rework General Discussion - Page 50
| Forum Index > LoL General |
|
BlasiuS
United States2405 Posts
I'm not commenting on whether people are applying the bare minimum of critical thinking to understand the game or not. That's a separate issue. | ||
|
Ketara
United States15065 Posts
| ||
GrandInquisitor
New York City13113 Posts
On April 10 2014 07:38 TheYango wrote: @GrInq, how well does GPM correlate if you break it down by role or even further by champion (not sure if your data set is large enough for the latter)? GPM is awkward to use across many champions and roles because your expected income tends to vary a lot across roles, but within a particular role/champion, it's probably a much more accurate performance metric. Good point. I only have meaningful samples for Top and Support, and definitely not for champion: Top: GPM = .49 TKDA = .71 DF = .54 KDA = .50 Side = .05 Support: GPM = .73 TKDA = .70 DF = .66 KDA = .50 Side = .18 I'm as surprised as you are, probably. | ||
TheYango
United States47024 Posts
For top, what champions do you play the most? I'm interested in how wide the expected farm distribution of your top laners of choice is. | ||
|
GrandInquisitor
New York City13113 Posts
List of champs i've played 4 or more times: Champion / Games / Wins / Winrate Jax 48 29 60.42% Thresh 35 18 51.43% Blitzcrank 15 11 73.33% Nasus 15 9 60.00% Kayle 15 7 46.67% Cho'Gath 12 7 58.33% Wukong 10 5 50.00% Morgana 8 2 25.00% Leona 6 3 50.00% Fizz 6 2 33.33% Vi 5 3 60.00% Singed 5 2 40.00% Ryze 5 2 40.00% Jarvan IV 5 1 20.00% Kassadin 4 3 75.00% | ||
|
GrandInquisitor
New York City13113 Posts
On April 10 2014 07:45 TheYango wrote: Support's makes sense because virtually all your gold is from objectives and kills, which we already established as having/likely to have good correlation with winning. This is a good point. Next season, perhaps, I should try to track CS just so I can approximate how my gold breaks down (kills vs CS vs objectives). | ||
|
AlterKot
Poland7525 Posts
| ||
|
GrandInquisitor
New York City13113 Posts
| ||
|
dae
Canada1600 Posts
On April 10 2014 07:52 AlterKot wrote: I heard that all good players use attack move to autoattack exclusively and I can see why (to avoid walking into enemies if you miss the enemy etc.). However, pressing shift+rmb is uncomfortable and I found pressing A and then LMB quickly prettty mundane, so I rebinded shift + rmb to A, so I effectively have A on smartcast and it seems to work well. Is there anything wrong with this idea that I should consider? I'm diamond 1 and I just right click on people to attack them, I might just be bad though, I don't play adc. | ||
|
AlterKot
Poland7525 Posts
| ||
|
Nos-
Canada12016 Posts
On April 10 2014 08:09 dae wrote: I'm diamond 1 and I just right click on people to attack them, I might just be bad though, I don't play adc. i'm just not confident enough in my mouse accuracy in the heat of the moment. Better to be attacking something instead of just accidentally walking into them. | ||
|
ShaLLoW[baY]
Canada12499 Posts
| ||
|
GolemMadness
Canada11044 Posts
| ||
|
Sufficiency
Canada23833 Posts
Essentially, the issue is that whatever this contribution factor is (be it GPM, KDA, kill contribution, or a mixture of something in between), it depends on which champion you play. Even champions in the same lane occupying the same role can have different average number of kills/deaths per game won, simply due to playstyle differences. Ideally, if we have the MMR of each player (which I don't) and if we assume MMR is an accurate estimate of the player's "skills", we can take a look how MMR correlates to our hypothesized "contribution factors". Even so, it will need to be calculated separately for each champion. | ||
|
GrandInquisitor
New York City13113 Posts
For example, for the age-old question: do we take objectives or kills -- winning teams are associated with both kills and objectives. Sometimes taking a kill means taking objectives later, and sometimes the opposite is true. What you really care about is, at that very moment, is it better to try to take down mid tower or roam bot? Similarly, if you want to know whether dragon or bot tower is more important, you can't just look at overall number of dragons taken vs overall number of bot towers taken, because even losing teams usually take down a dragon and the outer bot tower at some point. What we want to know is, how does your win probability shift by calling for a dragon, instead of finishing off the tower. | ||
|
KawaiiRice
United States2914 Posts
On April 10 2014 08:47 GolemMadness wrote: What kind of approach do you guys take to picking champions and roles? Do you main champions, main roles or just pick random stuff? I always call jungle/thresh in champ select because those are the two roles I know I can do better than other people in my elo but I fill if people say they literally can't play a role (cough support). If I fill outside of those roles I just try to counterpick my lane or get a safe pick (i.e.: ryze/renekton/leblanc/ziggs/cait), try not to die and take very low % risks. A lot of people climb elo with an incredibly small champ pool so that's not uncommon. Not really possible for meta junglers though since every game is 4~6 jungle bans now (at d3 anyway). | ||
|
Sufficiency
Canada23833 Posts
On April 10 2014 08:57 GrandInquisitor wrote: The problem is that we only have post-game data. If the point of the exercise is to identify how to change our play, any meaningful data must be gathered during the game. For example, for the age-old question: do we take objectives or kills -- winning teams are associated with both kills and objectives. Sometimes taking a kill means taking objectives later, and sometimes the opposite is true. What you really care about is, at that very moment, is it better to try to take down mid tower or roam bot? Similarly, if you want to know whether dragon or bot tower is more important, you can't just look at overall number of dragons taken vs overall number of bot towers taken, because even losing teams usually take down a dragon and the outer bot tower at some point. What we want to know is, how does your win probability shift by calling for a dragon, instead of finishing off the tower. I actually think looking at end-game data is sufficient. Ultimately, we want to know about what "contribution factor" contributes to "winning", not "I-am-pwning-everyone-as-Shaco-at-10th-minute-but-cannot-convert-my-advantage-to-a-win". Don't get me wrong, more data is always better, but even if we have it, it needs to be analyzed with a grain of salt. | ||
|
GolemMadness
Canada11044 Posts
On April 10 2014 08:58 KawaiiRice wrote: I always call jungle/thresh in champ select because those are the two roles I know I can do better than other people in my elo but I fill if people say they literally can't play a role (cough support). If I fill outside of those roles I just try to counterpick my lane or get a safe pick (i.e.: ryze/renekton/leblanc/ziggs/cait), try not to die and take very low % risks. A lot of people climb elo with an incredibly small champ pool so that's not uncommon. Not really possible for meta junglers though since every game is 4~6 jungle bans now (at d3 anyway). Yeah, I'm interested in how many people these days main a role and actually play several different champions in that role. Like I have a bunch of games this season on jungle Eve and support Karma, but very few games on other junglers/supports, so I can't say that I actually main any role, because it's really just one champion in that role. | ||
|
Sufficiency
Canada23833 Posts
On April 10 2014 07:52 AlterKot wrote: I heard that all good players use attack move to autoattack exclusively and I can see why (to avoid walking into enemies if you miss the enemy etc.). However, pressing shift+rmb is uncomfortable and I found pressing A and then LMB quickly prettty mundane, so I rebinded shift + rmb to A, so I effectively have A on smartcast and it seems to work well. Is there anything wrong with this idea that I should consider? I would argue Shift + RMB is the best (and most people use this option), but feel free to disagree. As you are probably aware, using attack move while kiting is very desirable because in case you misclick, you can still hit something instead of walking into your enemies. But one other thing you also should know is to use the `/~ key to only target champions, so you don't end up like turtle here: | ||
|
KawaiiRice
United States2914 Posts
On April 10 2014 09:02 GolemMadness wrote: Yeah, I'm interested in how many people these days main a role and actually play several different champions in that role. Like I have a bunch of games this season on jungle Eve and support Karma, but very few games on other junglers/supports, so I can't say that I actually main any role, because it's really just one champion in that role. If you are a solo queue only player you can get away with maining very few champions. I personally played a lot of ranked 5s so I chose to learn a large variety of junglers to pick based on my team's comp or what my teammates wanted. I approach solo queue jungling as needing to make the most early impact possible to let my teammates carry me later on so I pick a lot of strong early game junglers (lee, old elise, pantheon, eve, vi). I highly dislike low impact early game junglers outside of 5s because I don't trust random teammates to do well without my help (huge example being wukong). I don't know how this view fits with the new feral flare junglers yet though. | ||
| ||
