Can you use smite over the forest walls around blue buff golems?
|
League of Legends General Discussion Thread
|
VoiceOfUnreason Wrote:Search let me down, so: what key binding recommendations do people have for low elo? v8mark Wrote:I use smart cast for everything, and it's awesome. But I don't recommend it for new characters, as you won't easily be able to see the range of your spells.Yeah smart cast is very useful, self cast less so - suggest you get used to hitting alt + Q/W/E/R fast. Since Q is the universal poke for most champions, I usually assign A as smart cast Q, and depending on the situation will sometimes reassign S from "stop action" to another skill. When playing Kassadin you should always change R to smart cast imo, the difference between 2 clicks and 1 can mean survival or death, other skills to use for smart cast are Ashe volley, Cho scream + rupture (easier to hit that fleeing-for-survival-rupture when it's on smart cast) etc. Someone mentioned not putting Trist's E on smart, agree, but only because the range on that skill is lower than her auto attack range. Better just to assign her Rocket jump W so you can escape faster. Speaker Wrote:There is no such thing as ELO hell. It's just a stupid concoction by bad players to try to make excuses for why their ELO is bad.Disagree. It's undeniable that what you say happens to some extent: often there will be a player on the losing team moaning about elo hell, who themselves were partly responsible for the loss. Either because they failed to tank/engage, failed to stay with the team (I've lost so many games where, despite repeated urgings, a member of the team has ignored baron to prioritise creeps. A game where an Ashe went off alone to farm creeps in no-mans land 55 minutes into the game still sticks in my mind - up til that point we looked sure winners, the enemy won 2 minutes later). Sometimes even, the player moaning about elo hell was in fact the best player on the team, but was responsible for the loss because whenever anyone else died, he would spend the next minute telling them how bad they were. Team morale would go down, player performance would decrease because the bashed player was busy replying to the insults, game lost. Sullla Wrote:I also don't believe that ELO hell exists. I actually think the ranking system does a pretty good job of establishing general skill level, with the caveat that of course some people will do much better/worse in an organized team environment than others compared to solo queue. Just look at the YouTube games, where the most highly rated player (Plasma Puffball = 1500 ELO) has thoroughly stomped all over those of us with lower ratings. It's not a coincidence.I think the problem stems from Riot's method of assigning elo at the start. When you reach level 30 and first click ranked, you're given an effective ranked elo of ~1200 and the first 5 games count for more (~45ish elo). Those figures are rough approximations off the top of my head, so correct me if wrong please. That means if you are a player with an actual elo of 1150-1250, you could find yourself in a game with 4 players who have never played ranked, up against 5 players who all have 200+ wins under their belt and have worked their way to that position the hard way. It doesn't take a genius to figure out which team is going to win 9 times out of 10. That was an extreme example, but it could easily happen. A more common example is that you will have only one "soft" 1200 player on your team vs a team of "hard" 1200s, but we all know that it only takes one player to gift a win. You can also extend this example up to 1300 elo or more, where the soft player on your team has only 2 games total and got carried/lucky in both those games. The point I'm trying to make is that the system in place creates a constant stream of inexperienced players being thrown into the middle of the ranking. I believe I read somewhere (Stonewall's blog on Elementz site IIRC) that if you're 1300 elo or above, you're in the top 10% of the playerbase? Why does Riot place these new players in at ~1200? Why not lower and make us all work our way up the ladder honestly? (Starcrafters: Does Blizzard do this with the Starcraft ladder?) Sullla Wrote:After roughly 50-100 games, you should be in more or less the correct place.I think you're being a bit glib here. IIRC you don't play solo queue ranked, so where does this 50-100 games figure come from? (Do you have another account where you do play solo queue ranked?) To take an example from among us: Noobstar/Metallian is ~1400 elo in 5v5 ranked, ~ the same elo in 3v3 ranked, but is unranked in solo queue with around 850:830 wins:losses. After playing with him a bit, do I think that unranked designation is correct? No. I do note your caveat from the previous quote, but can't seem to make it apply for this example. So unless there's a reason I'm not aware of (maybe he spent 200 games trolling with AP Ashe?), I don't think the ranked experience is so easily wrapped up. Long post sorry, but these are my thoughts on the subject. I do think players will eventually reach their elo, it's the law of averages. But the road leading to there can be very bumpy, in part due to Riot's matchmaking system, not necessarily the player's fault. pocketbeetle Wrote:The point I'm trying to make is that the system in place creates a constant stream of inexperienced players being thrown into the middle of the ranking.It might have alot to do with people sitting very close to 1200 where they started. I have experience from an online hockey game that uses the ELO rating as well. Over there 1000 is the ranking every team starts with. If you go a bit under 1k, you'll be around #800 in position, but if you get a bit over 1k, you'll be somewhere around #250. Neither LoL or the hockey game rank inactive teams or players. I think the reason for all this is that alot of people play enough to be ranked, but not enough to significantly drop or rise in ranks. To answer your questions, I think the starting point is relative. If they made us start at 2000 ELO, then the 10% top would probably be the players at 2300+. And to the second one, it wouldn't be fair to the new players if you lower the starting rank now. The whole point of ELO is to try to create a ranking system that ranks players as fair and as accurate as possible. Here's more about ELO: http://en.wikipedia.org/wiki/Elo_rating_system I haven't actually read it myself, but I'm pretty sure it's relevant 
Some thoughts about ELO as I understand Riot's system:
The Law of Large Numbers isn't exactly right here. Your ELO is a weighted average of your performance, where more weight is placed on recent games. This clearly must be so because if ELO were just an average of ALL ranked games, then the magnitude of changes would decrease as more games are played. People who have played 1000 games would see their ELO change by 0.5 or 1 point while people who have played 100 games see their ELO change by 10 or 20 points if ELO were simply an average. My understanding is that this is not the case (from watching Guardsman Bob's stream ).Instead, Riot's ELO lets recent games count for more in the average. The 100th game and the 1000th game and 10000th game all add the same number of points for a win. (If I'm wrong about this please let me know). This is good because it allows people who get better at the game to change their ELO by discounting the earlier games. On the flip side, this is bad because it violates the standard assumptions of the Law of Large Numbers. This also ensures that people's ELOs will never be constant, but instead will consistently be in flux within a band around their true ELO. Modified versions of the LLN still work in this system, but without knowing the explicit details of Riot's system it's hard to say exactly how. In particular, the convergence rate may be very bad and so it takes a long time for one person's ELO to converge to the truth. Or the 'wiggle band' may be quite large and so a 1200 ELO player may oscillate between 900 and 1500. According to posts by Riot's devs, they believe that the above concerns are non-issues. ELO stability within 50-100 points is achieved around 100 games in, if I recall their posts correctly. There are some other issues with team composition and production complementarities. In short, this is the complaint of low-ELO support players. It very well may be the case that they can support a 1500 ELO effectively but not a 500 effectively. Their 'effective ELO' depends a lot on their teammates and so they are a 500 ELO player or a 1500 ELO player depending. Non-monotonicities of this form violate the ELO assumptions and we would not expect any convergence result to hold if this were the case. I don't really believe that this is worth worrying about though. For all this digression on the technicalities, it's worth reiterating: The intuition and logic of the Law of Large Numbers still applies! If you are better than your ELO, then you are more likely to win your game than lose it! Quit whining and carry harder. 
How ELO works is that the amount of rating you gain depends on the outcome and the average ELO ratings of both teams. Let's say you have two teams with exactly the same average ELO, the winners gains 30 ratings and the loser loses 30 ratings (We don't really know if the number is 30, but I'm gonna assume it is just from my own experience). But if one team has larger average ELO than the other, then they will get 0 < X < 30 rating for winning, and lose 30 < X < 60 for losing. The other team will gain 30 < X < 60 for winning and lose 0 < X < 30 for losing. The X depends on how much of a difference there is between the two team's average ELO's. The more extreme the difference is, the more extreme your rating gain or loss is. Basically, if a team of avg 2200 ELO loses to a team of avg 100 ELO, they would most likely lose around 59,9 rating. If they won, they'd gain around 0,01 rating. Of course, the matchmaking isn't THAT bad in LoL.
pocketbeetle Wrote:Why does Riot place these new players in at ~1200? In StarCraft 2 you first play a series of unranked seed games where they pit you against varying skill levels. Then based on how well you did they dump you into a league, either Bronze, Silver, Gold, Platinum, Diamond, or Masters (unlikely). You typically only get matched with someone from that league unless you're about to break out into the one above or below. It's the initial setup that really helps avoid the influx of new material at a single point, meaning you'll still move for several games, but the injection is spread out so you're not completely outclassed. And frankly they have the data from your 200+ other games, which also used a soft matchmaking ELO system, why aren't they using that to help place you? Heroes of Newerth suffers from a similar problem of ELO Hell. Except even more fun is they weight your ELO gain/loss based on your current ELO. Higher ELO players gain/lose less points to protect them from feeding low-ELO players handing the other team the game. This makes ELO Hell even worse since at lower scores your ELO score fluctuates so heavily and is really based on if the higher ELO player actually cares that day to help the team (or just farms the low ELO player on the opposite team).
I generally think that ELO in a team game is always problematic, because as opposed to a one-on-one game like chess your rating will be too heavily influenced by other players.
If I understand correctly, as a rule of thumb you can calculate the probability of winning by the following formula: (ELO difference / 8 ) +50 = expected win percentage. So if a team with rating 1600 plays against a team consisting of level 1400 players the expected result would be: (200 / 8) +50 = 75% win chance for the higher rated team (they would be expected to win 75 out of 100). At the same time once you hit 400 rating difference, your chances of winning go against 0 and I believe many rating systems only use a max of 400 points difference for calculation even though the actual difference may be greater.
I'm a firm believer that if you're at a certain ELO that very likely where you deserve to be. That guy who runs off solo half the time and goes 2/10/0 might have worse teamwork than you but he might be far better at using his skills in a fight. My game awareness isn't great and neither is my 'mechanical' skill or maximising last hits, flashing then immediately hitting someone with a skill is something I seem to struggle with especially. Whatever the case I try to play well, help people out when I can and have fun, maybe there's a major flaw in that that stops me playing like a 1400+ player instead of a 1100 one but I'm not too worried about it
"We are open to all opinions as long as they are the same as ours."
|
