|
Look at the TSM/CLG.eu Game 3. TSM had a 10% gold lead at the 12 min mark and won the game. Did that game feel snowbally? Did that game feel like a foregone conclusion? Did that game feel like CLG.eu had no chance to make a "comeback"? Of course not.
These numbers are impressive in that this has been a tournament of front runners, apparently. They could be completely flukey. They could be incredibly telling. It's hard to know without both more numbers and more analysis on the numbers to filter out the noise.
It's weird though to say things like there is an "issue if 2/3 of the average game time is spent only determining 10% of the final outcome". I don't think these numbers come close to really saying that. LoL has been criticized more in the past for allowing too many comebacks rather than too few. Certainly snowballs happen. But there are plenty of 10% gold games that are not snowballed.
I don't know. I'm having trouble saying things how I want to, but basically I think people are taking these numbers too far if from this they conclude that LoL games are all decided by 12 mins and that therefore it is a terrible spectator sport or in danger. I know that's not what Kronen was going for, and people should understand that these numbers are not sufficient to support that position.
|
United States47024 Posts
On June 10 2012 08:54 Takkara wrote:
It's weird though to say things like there is an "issue if 2/3 of the average game time is spent only determining 10% of the final outcome". I don't think these numbers come close to really saying that. LoL has been criticized more in the past for allowing too many comebacks rather than too few. Certainly snowballs happen. But there are plenty of 10% gold games that are not snowballed.
It has?
I don't remember that ever being the case.
|
On June 10 2012 08:35 TheYango wrote:Show nested quote +On June 10 2012 08:29 little fancy wrote:
I don't think that's a bad thing, because playing at the very top level is a pro players profession so I don't really see a problem with a small kill lead or dragon advantage influencing the winning chances so early. By design there's an issue if 2/3 of the average game time is spent only determining 10% of the final outcome. If 90% of games take 12 minutes to decide, then why should games take longer than 20 minutes? It kills the tension for the spectator when in 90% of games, the latter 2/3 of the game aren't contributing to the final outcome.
Don't get me wrong, I don't think you're wrong and numbers don't lie obviously. Objectively judging, I fully agree with you.
What I tried to say is that this statement is only really true if all players play perfectly which is not always the case (but they come close since this is LoL at its very best [at least it should be  ]). This opens rooms for comebacks and interesting drawn out games.
And there is always a subjective individual perspective of each viewer. Some people find a long diversified 45 minute game to be enjoyable to watch when others prefer a 25 minute minute game that ends with a 5:0 score and is decided by many small details. For me it doesn't matter if statistic proves that the winner is often decided by 12 minutes as long as I am entertained.
The design does have its flaws, but there are many more reasons to watch LoL as an e-sport (and if it only is that you want to see your favourite team winning something) which is why I think that this deficit is not necessarily a 'bad' thing regarding the entertaining aspect I get out of watching this.
Hope that made it clear, it's not that easy to express when you're not a native speaker. Cheers!
|
On June 10 2012 08:54 Takkara wrote:
Look at the TSM/CLG.eu Game 3. TSM had a 10% gold lead at the 12 min mark and won the game. Did that game feel snowbally? Did that game feel like a foregone conclusion? Did that game feel like CLG.eu had no chance to make a "comeback"? Of course not.
It only "feels" close when you don't know any better.
Once you've been following the game for a while and know this information, you see through the fake feeling of "closeness" very easily.
|
On June 10 2012 08:54 Takkara wrote:
Look at the TSM/CLG.eu Game 3. TSM had a 10% gold lead at the 12 min mark and won the game. Did that game feel snowbally? Did that game feel like a foregone conclusion? Did that game feel like CLG.eu had no chance to make a "comeback"? Of course not.
These numbers are impressive in that this has been a tournament of front runners, apparently. They could be completely flukey. They could be incredibly telling. It's hard to know without both more numbers and more analysis on the numbers to filter out the noise.
It's weird though to say things like there is an "issue if 2/3 of the average game time is spent only determining 10% of the final outcome". I don't think these numbers come close to really saying that. LoL has been criticized more in the past for allowing too many comebacks rather than too few. Certainly snowballs happen. But there are plenty of 10% gold games that are not snowballed.
I don't know. I'm having trouble saying things how I want to, but basically I think people are taking these numbers too far if from this they conclude that LoL games are all decided by 12 mins and that therefore it is a terrible spectator sport or in danger. I know that's not what Kronen was going for, and people should understand that these numbers are not sufficient to support that position.
Since when??? We're talking about pro games here, not solo q where people do stupid shit 90% of the time every time.
The fact that 90% of games are essentially decided by 12 minutes is a problem. Sure, there's other variables such as skill differentials between the teams, but the fact still remains that 90% of LoL games are extremely predictable and have foregone conclusions when the lead is as small as 10%. 10% in 12 minutes is absolutely nothing. That's 1-2 kills and something like 100 team-wide cs. When you know with 90% certainty that a particular team is going to win within 12 minutes based off of such a small lead that's just ridiculous. This being a problem has nothing to do with whether or not the games themselves are enjoyable imo.
LoL is far far too snowbally when compared to other competitive games. You could probably extend this research and I will bet that the team that gets fb wins a vast majority of the times.
Take the TSM v. CLG.eu series for example. Game 1 was over by 5 minutes with utter domination of every lane and jungle off of those early kills. Game 2 was essentially decided as soon as CLG fb over TSM. Game 3 was only as close as it was because Froggen is easily one of the best LoL players in the world; if it was any other pro in that position, game 3 very likely would've gone the same way as 1 and 2.
|
OP: Statistically speaking, the only fault I find with your data is that you can't just assume equal skill levels. These lopsided stats could just as well (theoretically) be explained by vast differences in skill, where the more skilled team pretty much always wins the game and 90% of the time has a 10% (or more) lead in gold by minute 12. If you could somehow get past this obstacle and show that you're comparing teams which are more or less even in skill, then I'd have no objections.
e.g. if you had stats for a best-of-49 series of LoL games between two teams, where one team wins 25 games and the other team wins 22 games, and in 90% of all of these games the team that was 10% ahead in gold by minute 12 wins, then I'd agree that snowballing is a problem at the skill level of these teams.
|
On June 10 2012 08:54 Takkara wrote:
Look at the TSM/CLG.eu Game 3. TSM had a 10% gold lead at the 12 min mark and won the game. Did that game feel snowbally? Did that game feel like a foregone conclusion? Did that game feel like CLG.eu had no chance to make a "comeback"? Of course not.
For the record, if you check the reddit post that game is counted as "tie" or otherwise an even game. TSM did have the 14min lead, but they lost it at 33min due to being outfarmed, and got i back at the 43min teamfight.
|
United States47024 Posts
On June 10 2012 09:22 Zato-1 wrote:
OP: Statistically speaking, the only fault I find with your data is that you can't just assume equal skill levels. These lopsided stats could just as well (theoretically) be explained by vast differences in skill, where the more skilled team pretty much always wins the game and 90% of the time has a 10% (or more) lead in gold by minute 12. If you could somehow get past this obstacle and show that you're comparing teams which are more or less even in skill, then I'd have no objections.
e.g. if you had stats for a best-of-49 series of LoL games between two teams, where one team wins 25 games and the other team wins 22 games, and in 90% of all of these games the team that was 10% ahead in gold by minute 12 wins, then I'd agree that snowballing is a problem at the skill level of these teams.
Even accounting for that, the difference between predicted/actual result should be more than 10%.
Consider this: in a "healthy" game, you want to see diverse and innovative play. Particularly in lopsided matchups, you should expect to see weaker teams prepare innovative strategies and drafts against the stronger teams. Inherently this "cheese factor" should create a divergence of greater than 10%, even if EVERY game is lopsided, because you would expect cheesy/innovative strategies to have significantly better than a 10% winrate.
The fact that there's such a narrow divergence means one of two things:
- "Cheese" games result in lopsided results anyway (either they're too successful and the cheesing team wins too often, or they're not successful enough, and the stronger team wins anyway)
- Not enough "cheese' games are being played--people are, for whatever reason, always opting for the safe strategies, even in super-lopsided games where cheesing should give them a better shot at winning
Neither of these necessarily signify a problem with the game's design, but they are issues that should be investigated--if people are opting not to play cheesy/innovative strategies, or they're not being successful enough, we should take care to try and understand why this is the case.
|
All I know is that it's far too easy to think these types of "cool" statistics are causal to the result, when it's just an interesting statistic. For example, the NBA playoffs are about to go to the NBA Finals. Did you know that the team that wins game 1 of the NBA finals has won the title over 75% of the time? So, you could watch 1 game out of a potentially 7 game series and know with ~80% certainty who the Champion is.
Does that mean that winning game 1 is actually the most important thing towards winning the NBA Championship? No. Does it mean that having a one-game advantage is insurmountable for the other team? No. It means that good teams usually win the first game, and having to win only 3 more games when your opponent has to win 4 is actually pretty sizable advantage. It's not often that the worse team wins the first game of the NBA Finals, but it doesn't mean that winning game 1 of the NBA Finals is the reason these teams are winning.
So while it's really interesting that the team that was up at the 12 min mark has won 90% of the time, that's really all it is for now, interesting. It's one of those popcorn facts that announcers use to fill time or stat geeks throw around to try to divine the outcomes of games.
|
On June 10 2012 09:22 Zato-1 wrote:
OP: Statistically speaking, the only fault I find with your data is that you can't just assume equal skill levels. These lopsided stats could just as well (theoretically) be explained by vast differences in skill, where the more skilled team pretty much always wins the game and 90% of the time has a 10% (or more) lead in gold by minute 12. If you could somehow get past this obstacle and show that you're comparing teams which are more or less even in skill, then I'd have no objections.
e.g. if you had stats for a best-of-49 series of LoL games between two teams, where one team wins 25 games and the other team wins 22 games, and in 90% of all of these games the team that was 10% ahead in gold by minute 12 wins, then I'd agree that snowballing is a problem at the skill level of these teams.
Ultimately you're right. I can't account for skill... yet. We'll see as the tourney progresses though because people eventually will be paired against similar skill levels. Already the games have gotten better, but they're still very lopsided generally.
But to play devil's advocate: Would you say that teams that split a series are evenly matched? What of the snowbally aspects of split series?
|
On June 10 2012 09:48 Takkara wrote:
All I know is that it's far too easy to think these types of "cool" statistics are causal to the result, when it's just an interesting statistic. For example, the NBA playoffs are about to go to the NBA Finals. Did you know that the team that wins game 1 of the NBA finals has won the title over 75% of the time? So, you could watch 1 game out of a potentially 7 game series and know with ~80% certainty who the Champion is.
Does that mean that winning game 1 is actually the most important thing towards winning the NBA Championship? No. Does it mean that having a one-game advantage is insurmountable for the other team? No. It means that good teams usually win the first game, and having to win only 3 more games when your opponent has to win 4 is actually pretty sizable advantage. It's not often that the worse team wins the first game of the NBA Finals, but it doesn't mean that winning game 1 of the NBA Finals is the reason these teams are winning.
So while it's really interesting that the team that was up at the 12 min mark has won 90% of the time, that's really all it is for now, interesting. It's one of those popcorn facts that announcers use to fill time or stat geeks throw around to try to divine the outcomes of games.
Interesting... it could also mean that the team that did better during the regular season will win more often than not. Home court advantage is granted to the team that does better during the regular season. You could make the argument then that the inherently better team will have the advantage in Game 1. The 70% mark doesn't surprise nearly as much with that info.
|
For speculation, the snowbally aspect of the game on Day 2 has lowered from deciding 87% of the games down to around 80%. Still very very high, but interesting to note.
|
On June 10 2012 09:56 Kronen wrote:
For speculation, the snowbally aspect of the game on Day 2 has lowered from deciding 87% of the games down to around 80%. Still very very high, but interesting to note.
All the things I've said aside, thank you for collecting this information. I am a total stat and factoid geek and love to pour over these things. I appreciate you doing this so we have geeky stats to argue over. 
|
On June 10 2012 09:57 Takkara wrote:Show nested quote +On June 10 2012 09:56 Kronen wrote:
For speculation, the snowbally aspect of the game on Day 2 has lowered from deciding 87% of the games down to around 80%. Still very very high, but interesting to note. All the things I've said aside, thank you for collecting this information. I am a total stat and factoid geek and love to pour over these things. I appreciate you doing this so we have geeky stats to argue over.
yw sir! This has been fun. And it actually makes me more interested in watching LoL too!
|
This actually makes me less interested in watching LoL, all I have to do is tune in to the first 12 minutes of the game and will have a pretty good idea who's going to win. Kinda like catching the last 5 minutes of the playoffs to see if Shaq could make his free throws. Saves time I know
|
On June 10 2012 09:37 TheYango wrote:Show nested quote +On June 10 2012 09:22 Zato-1 wrote:
OP: Statistically speaking, the only fault I find with your data is that you can't just assume equal skill levels. These lopsided stats could just as well (theoretically) be explained by vast differences in skill, where the more skilled team pretty much always wins the game and 90% of the time has a 10% (or more) lead in gold by minute 12. If you could somehow get past this obstacle and show that you're comparing teams which are more or less even in skill, then I'd have no objections.
e.g. if you had stats for a best-of-49 series of LoL games between two teams, where one team wins 25 games and the other team wins 22 games, and in 90% of all of these games the team that was 10% ahead in gold by minute 12 wins, then I'd agree that snowballing is a problem at the skill level of these teams. Even accounting for that, the difference between predicted/actual result should be more than 10%. Consider this: in a "healthy" game, you want to see diverse and innovative play. Particularly in lopsided matchups, you should expect to see weaker teams prepare innovative strategies and drafts against the stronger teams. Inherently this "cheese factor" should create a divergence of greater than 10%, even if EVERY game is lopsided, because you would expect cheesy/innovative strategies to have significantly better than a 10% winrate. The fact that there's such a narrow divergence means one of two things: - "Cheese" games result in lopsided results anyway (either they're too successful and the cheesing team wins too often, or they're not successful enough, and the stronger team wins anyway) - Not enough "cheese' games are being played--people are, for whatever reason, always opting for the safe strategies, even in super-lopsided games where cheesing should give them a better shot at winning Neither of these necessarily signify a problem with the game's design, but they are issues that should be investigated--if people are opting not to play cheesy/innovative strategies, or they're not being successful enough, we should take care to try and understand why this is the case.
Imagine that you see a lot of games being played between the best korean SC:BW pros vs. the best foreigners at SC:BW. The koreans play economical builds, because they know they'll win the late game; the foreigners play cheesy builds for the same reason. Because they expect the cheesing, koreans manage to win 90% of the games anyway. Then an observation is made: Economical builds win 90% of the games! Clearly economy-focused builds are OP.
^ Wrong conclusion, because there is an alternative explanation for this lopsided statistic (Koreans won 90% of the games because they're better -> skill difference).
On June 10 2012 09:49 Kronen wrote:
But to play devil's advocate: Would you say that teams that split a series are evenly matched? What of the snowbally aspects of split series?
If the winning team in a bo3 series doesn't win 2-0, then I daresay that the skill levels are definitely comparable, yes. If your data was composed of all the games from all the series that weren't won 2-0, and the lopsided snowballing winrates persist, then your argument about snowballing in LoL would be bulletproof for the skill level being examined.
|
On June 10 2012 09:57 Takkara wrote:Show nested quote +On June 10 2012 09:56 Kronen wrote:
For speculation, the snowbally aspect of the game on Day 2 has lowered from deciding 87% of the games down to around 80%. Still very very high, but interesting to note. All the things I've said aside, thank you for collecting this information. I am a total stat and factoid geek and love to pour over these things. I appreciate you doing this so we have geeky stats to argue over.
Got a simpler version. Here's what I PM'd:
It's t-test. Think of it as a more robust z-test http://en.wikipedia.org/wiki/Student's_t-test. Basically, I would do it right now if I wasn't busy doing the same thing (research in parallel computing for statistical modeling). But it should be pretty simple. Here is a much simpler version:
1) Take one team at random from each game. . Make a binary variable w=1 if the team won and w=0 if they did not. Make a binary variable l to see whether it had early lead or not (with your criteria) where 1 is the early lead etc.
2) Run the following regression on the data. The equations would just be to regress the model:
w = A + Br + Cl + e
Where we a binary variable of whether the game was won early and r is the round (round 1, round 2, etc.). Whatever program you use should tell you whether B is significant. If it's not, get rid of it. A should be somewhere around .5 since it is the mean chance of winning from your sample (about half the teams should have won).
3) Good, this model controls for round. If you believe that round is a good proxy for skill differential, then this will control for skill differential. Thus now we clearly see the change in the chance of winning given a lead as the coefficient C! Notice the regression already gives your answer for you if you are simply testing for whether a lead gives a win. The normal test on C would be to test if it's 0, that would just be to test if the chance of winning is independent of which team started with a lead (given the skill differential is controlled for). Lets do something stronger here. Instead, do a t-test testing to see if C is statistically greater than .1. That is just saying that yes, we would expect a team with equal skill to have a .1 better chance of winning (you get an early lead of 10% gold but are roughly the same skill, you should have around a 60% chance of winning). To get the t-score, use a program or simply take (C-.1)/(sd / sqrt(n)) where n is the number of data points (this is usually how the equation is shown). Now that is distributed as a T with degrees of freedom n-1, and check a t-table for its probability.
The question is, is that significant? If it is, then it says that roughly equally skilled teams have better than a 60% chance of winning given an early lead of 10% at 12 minutes. Wow, that would shock me.
What one could dispute though is the use of round as a proxy for skill differential, though I'll take it. In most cases, the higher the skill category, the lower the skill variance (this has been tested a lot with Olympic athlete data). So it should net out a good amount of the effect. Another thing one could dispute is that the model testing will be a little off due to the use of the Linear Probability Model instead of going to a Logistic model here. I think that would be overkill since we are just playing with LOL data but if someone wants to use a Logistic, go ahead. If it's highly significant or highly not, it doesn't matter anyways.
|
I thought it was well known that LoL doesn't allow comeback unless major errors.
Wasnt there a discussion about dragon on that topic ?
|
Where can I find the results of picks/bans for all games from MLG? Like who was picked, who was banned, and how many times.
|
This would be even more interesting if data was gathered for each individual minute, not just the 12th minute (why was 12 chosen? seems arbitrary).
|
|
|
|
|
|
EPT
