4thstringer Posted July 17, 2017 Report Share Posted July 17, 2017 So, I have a question for you all. How does Lucius have only 38 gg2017 Neverborn games? I know I have 12 of them. And are you all really struggling with him that hard? That is a really low win percentage. What have you found to be his problem? Quote Link to comment Share on other sites More sharing options...
Ergonomic Cat Posted July 17, 2017 Report Share Posted July 17, 2017 I have played precisely 0 of them, I know. I don't have a specific issue with him, I just love too many other masters to figure him out. 1 Quote Link to comment Share on other sites More sharing options...
santaclaws01 Posted July 17, 2017 Report Share Posted July 17, 2017 You also assume that a significant portion of people playing him are reporting their games on logfaux. Quote Link to comment Share on other sites More sharing options...
4thstringer Posted July 17, 2017 Author Report Share Posted July 17, 2017 Ok side question. If not, why not? Quote Link to comment Share on other sites More sharing options...
emiba Posted July 17, 2017 Report Share Posted July 17, 2017 2 hours ago, 4thstringer said: Ok side question. If not, why not? I can only speak for myself of course, but I only found out about it's existence a couple of days ago. And even if I haven't been playing for long, I've been reading about the game for a long time. Quote Link to comment Share on other sites More sharing options...
santaclaws01 Posted July 18, 2017 Report Share Posted July 18, 2017 5 hours ago, 4thstringer said: Ok side question. If not, why not? Because why would most people care about reporting their games? You can't just look at logfaux and say "Oh, Titania has a 54% win rate and Lilith only has a 46%. Obviously Titania is leagues better than Lilith". There's so much missing data for it to provide any real metrics other than something passably interesting to look at. Quote Link to comment Share on other sites More sharing options...
Adran Posted July 18, 2017 Report Share Posted July 18, 2017 10 hours ago, 4thstringer said: Ok side question. If not, why not? Time, effort. worth? I know that I don't even think of going to report my games there, and I will try and do things like write battle reports for here and enjoy looking at game stats. Its just not something I've gotten into the habit of. I also dislike reading people badly mis-representing its information as "facts" that "prove" something is rubbish/broken. (Although I've not seen that for a while, but it did seem to crop up a few times earlier this year). Quote Link to comment Share on other sites More sharing options...
Myyrä Posted July 18, 2017 Report Share Posted July 18, 2017 Yeah, the sample sizes in Logfaux are so small that they tell more about what certain good players like to play than how masters actually rate compared to each other. I could probably pick any little played master and get it to the top of the win rate list if I wanted to. Quote Link to comment Share on other sites More sharing options...
4thstringer Posted July 18, 2017 Author Report Share Posted July 18, 2017 4 hours ago, Myyrä said: Yeah, the sample sizes in Logfaux are so small that they tell more about what certain good players like to play than how masters actually rate compared to each other. I could probably pick any little played master and get it to the top of the win rate list if I wanted to. Ah yes, what we know as the "Guild McMourning Effect" It just surprises me that he has so few plays. Lilith has almost 5 times the number of plays that he does. I do suspect that part of it is that people tend to start with Lucius, and then go to other masters. Since new players are more rarely on Logfaux, you don't get many plays, and you also get a low success rate because those are new players whose games were probably recorded by their opponents. Quote Link to comment Share on other sites More sharing options...
4thstringer Posted July 18, 2017 Author Report Share Posted July 18, 2017 5 hours ago, Adran said: Time, effort. worth? I know that I don't even think of going to report my games there, and I will try and do things like write battle reports for here and enjoy looking at game stats. Its just not something I've gotten into the habit of. I also dislike reading people badly mis-representing its information as "facts" that "prove" something is rubbish/broken. (Although I've not seen that for a while, but it did seem to crop up a few times earlier this year). Now, I tend to be skeptical about the win percentages for lesser played masters (which is part of why I started this thread. I don't have that kind of struggle with him), but do you feel like the masters with 75-200 games are inaccurate pictures of that master's capabilities? How do you feel about the faction win percentages? Quote Link to comment Share on other sites More sharing options...
Myyrä Posted July 18, 2017 Report Share Posted July 18, 2017 It is very much possible that there are also some hidden correlations affecting the results. Beginner players might more likely pick certain masters over others because their aesthetics are more pleasing to wider audience, or more experienced or skilled players could be drawn towards more complex masters even if the complexity does not translate to actual game benefits. Same thing could apply to masters that are perceived powerful. Because it is practically impossible to determine if player assignment to different masters or factions is truly random, it will be impossible to draw very far reaching conclusions about the power levels of masters or faction just from statistical data like that. Not to mention there could be other problems with it as well. People could be more likely to report games they have won for example. 1 Quote Link to comment Share on other sites More sharing options...
Adran Posted July 18, 2017 Report Share Posted July 18, 2017 15 minutes ago, 4thstringer said: Now, I tend to be skeptical about the win percentages for lesser played masters (which is part of why I started this thread. I don't have that kind of struggle with him), but do you feel like the masters with 75-200 games are inaccurate pictures of that master's capabilities? How do you feel about the faction win percentages? Having followed tournament results for about 7 years, I have found that the name of the player is a better guide to where they place than the name of the master/faction used. I've not looked at Logfaux for ages, so I haven't a clue what data it shows. but I'm not sure faction win percentage is a very useful stat on its own until you start getting to some very large numbers Does it break down the 10 different strategies? Does it take into account the schemes? Does it tell you that Lucius has fewer neverborn games because his starter set is rubbish for Neverborn? or that I haven't played him in the past 6 months because I haven't gotten round to picking up his erratad cards? Does it tell you crew constructions? After all things like witching thrulls look to make him much better, but aren't out yet. Are people who proxy them (or bought pre-release) getting better results at the moment that people waiting for the release? In short, yes I do feel that just reading the result for 200 games is not really enough to get a picture of a master. Does it say why someone picked that master for that game? And what other options they had when they picked it? 2 Quote Link to comment Share on other sites More sharing options...
Sharp_GT Posted July 18, 2017 Report Share Posted July 18, 2017 I don't think lucius gets much love amongst neverborn players. I for one never touch him, I just think our other masters do the same things better. Quote Link to comment Share on other sites More sharing options...
Ergonomic Cat Posted July 20, 2017 Report Share Posted July 20, 2017 On 7/18/2017 at 8:59 AM, Adran said: Having followed tournament results for about 7 years, I have found that the name of the player is a better guide to where they place than the name of the master/faction used. I've not looked at Logfaux for ages, so I haven't a clue what data it shows. but I'm not sure faction win percentage is a very useful stat on its own until you start getting to some very large numbers Does it break down the 10 different strategies? Does it take into account the schemes? Does it tell you that Lucius has fewer neverborn games because his starter set is rubbish for Neverborn? or that I haven't played him in the past 6 months because I haven't gotten round to picking up his erratad cards? Does it tell you crew constructions? After all things like witching thrulls look to make him much better, but aren't out yet. Are people who proxy them (or bought pre-release) getting better results at the moment that people waiting for the release? In short, yes I do feel that just reading the result for 200 games is not really enough to get a picture of a master. Does it say why someone picked that master for that game? And what other options they had when they picked it? That's somewhat a Catch-22 though. If everyone logged their games in Logfaux then it probably would be possible to do that sort of analysis. If there was enough data to play around with, you could find interesting bits and do stuff. But if no one is logging games, no one will log games because it's not worth it to log games.... I try to log all my games in Logfaux because *somewhere* has to be the source, and it's the best I could find so far.... Quote Link to comment Share on other sites More sharing options...
santaclaws01 Posted July 22, 2017 Report Share Posted July 22, 2017 Unless logfaux also tracks the schemes and strats available and taken in each game and the games played by who in tournament games, it will forever remain a cute thing to look at. There's just too many things that go into the outcome of a game for it to be useful in its current state. Quote Link to comment Share on other sites More sharing options...
Ludvig Posted July 22, 2017 Report Share Posted July 22, 2017 Why I don't use logfaux. Everyone may not agree but hey, it came up! Point one: Statistics require huge numbers of data. Terrain and the mission in the form of strats and schemes would skewer the results as well as which player is playing which master. If there was a tier system for players based on rankings so that you could roughly control for player skill discrepancy it might say something. You'd also need to break down strat and scheme setup as well as which schemes the indvidual players chose since schemes are also of varying difficulty. Then we would need to consider the masters involved so if we had about 200 games of Lucius vs Lilith and so on for each master pairing and the necessary data to control for the other factors above we might get statistics worth basing decisions on. Crew composition would still be a huge confounder but we might see that summoners as a group or arcanists as a faction scored disproportionally high in interference. Point two: why do we need statistics in the game? I just like to enjoy my games and I don't care if my opponent is playing a master with a slightly higher win percentage than me. If something is percieved as unfun or overpowered I think it's more important to correct that than whatever has a slightly higher win percent based on fairly unreliable statistics. I play this game for the enjoyment of the game, I couldn't care less if I'm "handicapped" by the master I play as long as me and my opponent are having an enjoyable game. That goes for tournaments too, I will play the master I think gives a decent chance to enjoy myself and hopefully win but skill discrepancy is a bigger hindrance than model balance to that end. Even with errata the same players keep winning because they work out how to win with a master. When that master gets hit they are the first to move on to another master thatthey have probably had their eye on for a while and they get good and win with that master. In a casual game the current method of hitting NPE:s is probably fine if not better than going on statistics and in a tournament setting you can be expected to invest the time and money to keep competing. 1 Quote Link to comment Share on other sites More sharing options...
Adran Posted July 24, 2017 Report Share Posted July 24, 2017 On 20/07/2017 at 6:59 PM, Ergonomic Cat said: That's somewhat a Catch-22 though. If everyone logged their games in Logfaux then it probably would be possible to do that sort of analysis. If there was enough data to play around with, you could find interesting bits and do stuff. But if no one is logging games, no one will log games because it's not worth it to log games.... I try to log all my games in Logfaux because *somewhere* has to be the source, and it's the best I could find so far.... Yes, and no. If no one logs data because there isn't enough data, then thats true. If the data collected is incomplete, then people can draw incorrect conclusions from the data very easily. I'm not sure it is possible to collect the correct data to actually prove that master A is better than master B in any circumstances. Just because I picked master A in the game against Master C, that doesn't prove in anyway that master A is better than Master B. It doesn't even prove that I think Master A was better than Master B, because we aren't giving any data into why I picked master A. The fact that there are only 38 games logged with Lucius just tells us that not many people who log into Logfaux that use Lucius. Logging games into Logfaux can be really useful for telling some things about the game. I don't believe that most powerful master/faction is really one of the things it is telling you, but that is what most people look to the data to see. 3 Quote Link to comment Share on other sites More sharing options...
emiba Posted July 24, 2017 Report Share Posted July 24, 2017 Disclaimer: I do not have a degree in statistics but have some degree of experience working with data. There's a reason why there are entire academic degrees and university programs dedicated to statistics... Master A vs. Master B -> One dimension Player A with Master A vs. Player B with Master B -> Two Dimensions Player A with Master A and Strategy A vs. Player B with Master B and Strategy A -> Three Dimensions Player A with Master A and Strategy A and Scheme A vs. Player B with Master B and Strategy A and Scheme B -> Four Dimensions Player A with Master A and Strategy A and Scheme A and Scheme C vs. Player B with Master B and Strategy A and Scheme B and Scheme D -> Five Dimensions Etc, etc, so on and so forth. And we haven't even touched on things that are very hard to cater for such as terrain types and coverage, master play style or player skill with that play style... I think there's a bunch of things to be gleaned from asking the right questions from the data but I think the body of data is too small and the model (by necessity) to complex to make any kind of broad assumptions about the state of the global or local meta. For me, I'll use it as a way of tracking my progress with certain masters and strategies schemes but beyond that, I think it's mostly entertainment. 3 Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.