Jump to content


Verified Tanker [EU]
  • Content Count

  • Joined

  • Last visited

Reputation Activity

  1. Downvote
    A_Lethal1 got a reaction from jimjim975 in Do the pre-battle stats you get are the ones you need?   
    The stats we are given at the start of a match is our team & enemy teams
    Wn8,  Number of games played (measured in thousands) & WIN rate.
    Yes Wn8 & Win rate are useful, but they are also 'Historical' and not recent (I understand we can't display them however)  Number of games just tells us how many games they have played, but doesn't break it down by class, tier or tank.
    My argument is that this should be done for that particular tank. itself
    .  1) Win rate with that particular  tank
       2) Number of games with that tank (60% with 10 dames vs 60% of 180 games tells you more)
       3)Mastery & MOE.   Tells you this player knows how to use the tank or has used it at its optimum, though looking at win Rate will tell you if hes consistent with it.
    I have a mate that has 9K games, a meh Wn8 and a Win Rate of 47% my analysis of him for usefulness to the team or opposition would be a plodder and shouldn't be expected to do more then just carry their weight (barely)
    yet my analysis would be different if I knew he had a 73% Win rate with that particular tank, 190 games in it and 2 MOE.
    Which set of information would you want at start of battle?
    This is why we have that battle screen information is to evaluate our and the other teams strengths and weakness.
  2. Upvote
    A_Lethal1 reacted to OneSlopoke in Salvaging the Current Color Scale   
    BTU, I'm glad your post is right here. I have already modified the colors file in XVM to suit my sense of what I think the colors should be and have changed the wording that goes with it. I've been playing this game for 3 years so I think I'm allowed to form my own opinion.
    I was glad to see on the main page that Neverwish removed the 'bad' label and others from his chart there. I just voted here because the poll was already here, not because I care if you change it. I think you guys spend waaaay too much time worrying about this actually. But, I don't care if you do.
    You guys are carving up the numbers to 1% precision in a game where there are artificial means to influence the data (without being any better at driving a tank) that can move the data by 5 or 6% (platooning for example).
    And you really want to show the difference between 99.99 and 99.9 percentile. Really? And lets dump everyone into red/bad from 60% down. Actually, I think it would be more useful to break down the lower skillsets to determine exactly how people can improve up to average levels.
    The average player is probably to not any good to a 70% winner but, compared to a 30%, average has learned a lot. Why not let them have their due?
  3. Upvote
    A_Lethal1 reacted to RichardNixon in Salvaging the Current Color Scale   
    The primary reason that the mean and median account win-rate are lower than (1-draws)/2 is simply that good players play more than bad players. For a trivial drawless example, consider a population where each 60% player with 10k battles is balanced by two 40% players with 5k battles. Total wins/battles would be exactly 50%, but the mean and median account win-rates would be lower.
    Note that this applies even if the skill distribution is symmetrical: If good players play more than bad players, you end up with a right-skew winrate distribution. Average players are more likely to see good players than bad players, and hence their winrate drops below the (1-draws)/2 mark. The effective average skill that players see in games is much higher than the account average.
    You can define "skill" however you like, but I have no idea why you'd define it as a gaussian given that contribution clearly isn't. I assume you're thinking of the central limit theorem rather than the law of large numbers, but that only defines the distribution of the sample mean, not the population.
    WN8 exaggerates the contribution skew because the input data wasn't platoon-filtered. Eureqa was understandably confused that the 65% winrate players weren't doing much more damage than the 60% players.
  4. Upvote
    A_Lethal1 reacted to TheMarine0341 in Salvaging the Current Color Scale   
    We're taking this too personally, instead of a bland objective standpoint. MaxL is being the most objective and focused
    Full Post:
    Average is just that: Average. The last discussion we had as a community on this really opened my eyes as to what percentile Im in (95+) vs. the vast majority of players, which average to around 650 WN8 (and frankly, this helped my rage playing by realizing that those dumbasses were simply doing what the average pubbie would do). And, its bugged me that the color we determined to be *average* was instead assigned to "above average" esq players. Frankly, that really does make *us* look like collective pricks by *personally* deciding that the Average color is assigned to Above, while instead what is indeed average as red, or "Bad." The majority of the users here are WELL above average players, either by recent or overall which skews our view of the scale. 
    I think the current 11 step can be alright with better scaling, but cutting it down to 7 would be better. But with 11 steps, whatever the average WN8 is determined to have been, it should be Yellow regardless of your personal opinions of how bad those players are. 
    The bad players dont realize how bad they are, while the good players dont realize how someone can just be that bad. Its a two way street in this regard and we do need to recognize that. 
  5. Upvote
    A_Lethal1 reacted to Enaris in Salvaging the Current Color Scale   
    Pork, something to remember is the "Compression of Perspective".  A player who is in the top 5% of the server sees little "real" difference between the play of anyone yellow and below.  Yet, think about what Max said before.  While you may consider a 47% to be a "bad" player, the truth is that he's only slightly negative on your team compared to a 44%.  He'll cost his team around 1.5 of 100 games, as opposed to the 4.5 or so (assuming 48.5 average winrate) that the truly bad player costs his team.  That's as much as a difference as the one between a 53% and a 56% on the "positive" side.  That's a real difference, even if the compression of viewpoint makes them all appear "the same".
  6. Upvote
    A_Lethal1 reacted to _Shrew in Salvaging the Current Color Scale   
    Have to agree with this. Even if the average winrate and win8 of the entire player base is considered "average", it makes no sense to have it be red. When we see a sea of red - we think terrible team, when we see a sea of yellows - we think okay, should be a decent game, when we see anything above a sea of greens - we think it's time to put on the serious pants.
    Neverwish wants to create a color scale that is intuitive to the human eye with something that all players can easily relate to based on the social labeling of what is considered a good, bad, or average player.
    It isn't a problem of ratings, that can easily be changed, but when we find ourselves seeing the color red as neither good or bad it distorts our perceptions of colors.
    We've come to a point as community where we can quickly associate certain colors with our own definitions of performance - Unicums will always label greens and yellows as poorly performing players, and reds will always label greens and yellows as high performing players.
  7. Upvote
    A_Lethal1 reacted to Never in Salvaging the Current Color Scale   
    The current scale sacrificed usability for the sake of pragmatism. Average should never be red, regardless of how bad average would be. There are too many steps on the scale, and the naming convention used for some parts of the scale didn't work out well.
    The color scale is supposed to be a quick visual indicator, but it strays too far from the principles of using colors to convey a message. This is why I want to rework it, to be more usable and more human-readable.
    Average is Average. We can think that the average is bad, a lot of people here do. For some people here, anything below blue is bad. Truth of the matter is that what is considered "Bad" is mostly subjective, yet "Average" will always be objectively "Average".
    Edit: Well MaxL explained better than I ever could
  8. Upvote
    A_Lethal1 reacted to MaxL_1023 in Salvaging the Current Color Scale   
    The flaw in your assumption is that the average player is a poor player. What does that actually mean? You are making up a subjective skill-based scale which exists only in the mind of the person who invents it. The fact of the matter is that there are millions of accounts tracked. "Average" is average for a reason - this is the average skill level a player with X battles (I believe 1000 or 2500 for the current system) has attained. People with more battles tend to be slightly better, but this is compensated by newer players. The average WN8 of all players within a set criterion represents (somewhat imperfectly) the average level of skill a player has. If you judge this to be poor, that is your prerogative. However, it is useless for making any kind of metric due to the fundamentally subjective nature of your scale.
    The fact is, average (and bad) players do contribute to a win. The fact that their win rate is higher then an AFKer proves that. Based on a few studies I heard of, the AFK/Bot line (represented by 0 WN8) is somewhere near 40% win rate, as high as 42% in some tanks, possibly 45% in scouts due to their poor MM and low relative weight. Take 42% as an average across tank classes. A 45% player is at the very bottom of the scale, however they contribute enough to their team to win 3 games which a bot would lose for every 100 played. This is statistically significant after a few hundred battles. It is objectively, statistically certain that a 300 WN8, 45% player contributes to their team. They do about 20% of the expected values in their tank, but they still contribute. 
    As you move up, that fraction increases. It is notable that the slope is fairly shallow early on - going from 0.2-0.3-0.45, or roughly 2 of my exponential steps across 3 levels representing the range from about 300 to 650 WN8 only results in a 2-3% increase in win rate. At the bottom, going up one exponential step results in a change of +1% WR, which is still significant after taking into account the huge sample. The 3% from 0 to 300 represents more having your HP in play then anything, but even that is significant. 
    Near the middle, the slope increases, so an exponential step results in a jump of 3-5% in win rate. Dealing ~40% more of everything makes you win another 3-5 games per 100. At the top, the system starts to break down, but you can see the final step of ~35% results in a solid 5% jump. This last step represents by far the largest skill jump based on diminishing returns theory. However, from the viewpoint of how you actually perform in battles, it has a similar effect to jumps lower down.
    We have to change the current scale because it is needlessly complex and in general is insulting to most of the playerbase. We are applying our own subjective scale through the color categories. Replacing it with an easily justifiable mathematical progression which is based purely on relative performance (as measured as fractions of the expected values) produces a system which accurately represents the playerbase and is easily applied at all levels to show both absolute and relative performance.
  9. Downvote
    A_Lethal1 reacted to Angry_Ray in Salvaging the Current Color Scale   
    Therein lies the problem. If you make a bunch of reds/oranges yellows/greens.... where do you put our yellows/greens? Make them teal/blue? People are getting WAY too upset about this whole XVM color shit. Reds deny they're bad so want a different color, oranges deny they're badish and wanna be yellow, yellows deny being called average, greens whine they're not teal, 1900 teals whine they're not blue, 2350 blues whine they're not pruple, purples laugh at all. I think in its current state, the color scale is best as good as it's gonna get without starting the apocalypse.
    As per the whole teal color... I think it's just fine. It's a middle ground between the dark green and blue learning curve.
  10. Upvote
    A_Lethal1 reacted to Pravdy in XVM help please   
    Make sure you are using a 9.4 version. This was just added today 11/6 to the main download page, before today you had to use a nightly build since the patch.
    Edit the "xvm.xc.sample" file in the configs folder. You need to have an "xvm.xc" file pointing to the correct config. If using the default, just rename the sample file.
    edit: you may need to go to the XVM site and Add Client to add the token to your new install.
  11. Upvote
    A_Lethal1 reacted to SFC_Storm in XVM help please   
    Clear your cache folder.
    appdata\wargaming.net search for this in appdata.
    Delete all but Pref.xml. Then go to XVM page and login and press update client and activate stats. It will be fine.
    If its not you need to see what error you have in the xvmlog and post it.
    C:games\wot then open XVM.log and see what errors and post them here. Otherwise we arent telepathic
    I usually tell guys under 1k not to bother with XVM before they learn game mechanics as it really hurts more than helps usually. Both winchance and seeing who is a great player doesnt matter as much. Nor does last seen on minimap as under 1k you just need to focus on angling your tank and act like you could be attacked at every turn.
  • Create New...