Jump to content

Untamed_As_Thee

Members
  • Content Сount

    18
  • Joined

  • Last visited

  • Battles

    11131
  • Clan

    [N0KAP]

Community Reputation

44 Neutral

About Untamed_As_Thee

Recent Profile Visitors

658 profile views
  1. Untamed_As_Thee

    Fix The Game Then Send Me A Message Its Fixed

    Your formulas have absolutely nothing to do with deviation. As said before, standard deviation is a measure derived for usage along with mean, not median. It's getting pretty obvious now that you haven't read what I've wrote. I also didn't set this 35%; all along I've been trying to tell you why your 35%: Means nothing Isn't the 'median' Is wrong to be assumed as the true median. If you are set to read data between 30% and 70%, the median is 50% assuming no skewness. In the case of WoWs W/R, it's between the mode and mean, and those two doesn't shift too far. Choosing active players also doesn't change the requirement that the mean W/R, or the average, to be at slightly below 50%. In the same way that if you can't see the difference between the 3 points I've been trying to deliver above, then it would be too hard for you to understand the difference between M, m and μ.
  2. Untamed_As_Thee

    Fix The Game Then Send Me A Message Its Fixed

    I simply used your logic for calculating that 'median' damage. If I'm wrong, by corollary, you're also wrong. You still don't understand that your measure is simply not the median. And even you acknowledge that the range starts from 30%. The use of the word 'and' simply denotes the upper limit and lower limit that YOU set. I've not told you nor mentioned how many samples there were, and in a continuous approximation, you cannot say how many samples there exists by looking at the approximation. The Euclidean line is complete and dense everywhere; there is no finite number counting all numbers between 30% and 70%. The way you haven't realised that you just halved the upper limit, despite all the explanation I've tried doing, is beyond my belief. By removing whichever outlier you deem unnecessary (which, in this case, there isn't any but I'll play along) you have to find a new median. You can't disuse one outlier, calculate a median, and then cover up the other outlier and call the first median valid within the space where you've deleted two outliers. If it worked that way, the Lehman Brothers would still be in business. You've proven my point for the third time with this copypasta. It's wrong. It makes no sense. It hasn't been any more valid since.
  3. Untamed_As_Thee

    Fix The Game Then Send Me A Message Its Fixed

    Yes? The median is indeed not the average. The section you quoted from the website does NOT tell you that the median is the average. The mean is. Also, that website basically just defined what I described the median as. The centre between 0% and 70% is 35%. The center between 30% and 70% isn't 35%, it's 50%. By your logic, say if you average 100k in Kremlin, and your range of damage is always between 90k and 110k, then the 'median' would be 55k. I ask you again: do you see what's wrong? Why do you keep insisting on this arbitrary 35%? Why don't you actually read what I've written about your flawed construction of your broken formula? To actually answer OP's question: bad luck. I had a 9 game loss streak in the Henri once, even managed to break my cruiser damage record in a 9v9 game, still a loss...
  4. Untamed_As_Thee

    Fix The Game Then Send Me A Message Its Fixed

    For the fifth time, median is not average. I don't know why you keep insisting with this 35% median... it's wrong. If you can't accept the fact that it is wrong, there is no point to continue. Standard deviation and standard error is a measurement by mean, the actual 'average' though no statistician will ever use that term, not median. I have no idea which part of my explanation you're talking about in regards to having a high standard deviation or error. With an active player base of 12,000 players active AT ONE TIME, as high as the standard deviation may be (which is not very high anyway), the standard error is very low. Removing inactive players actually increases this standard error. If you don't know about what standard deviation is, to put it simply one standard deviation away from the mean of a non-skewed distribution covers ~70% of the population. That would be about all the 45% to 51% players. Standard error is standard deviation divided by a constant denoted by sample size. I don't understand what you mean by 'creating a formula' that calculates the median. There's no standard formula but the general concept that I've laid. It's the point at which half the players are below, and half the players are above. And in all honesty, I don't really need to find the median. I know what the mean is, I know what the mode is, and I know the median is bounded between the two because of the Empirical Relationship. What the actual median is bears very little significance. There simply is no reason to create a formula: it's not the median anymore if you change the definition of the median. There's one way to minimize standard deviation: take one data point. One. There, you can't then possibly have a standard deviation. You see how nitpicking data points and eliminating numbers you don't want to see doesn't work? By saying this datapoint and that datapoint is an outlier doesn't make any sense whatsoever. Being active or inactive doesn't change anything; should a player go inactive at any point it also invalidates the active player's previous games. By this rule, the mean holds at slightly below 50%, which is demanded. As additional information, as the sample size gets larger and larger, the asymptotic error (the error you're looking for) is only 25% greater than the standard error; the limit tends to standard error times pi over two to the power of a half. This is the study of sufficiency in statistics, led by Laplace. If you can't get through basic Prussian mathematics, it is futile to even try to understand French mathematics.
  5. Untamed_As_Thee

    Fix The Game Then Send Me A Message Its Fixed

    No such thing as 'deviation of error', but there is something called 'error of deviation'. It's called standard error. Has nothing to do with median, for which it's the fourth time that I have to point out is NOT the average... So glad you posted the formula again to prove my point. Not any more valid than the last time you've posted it. If you're going to quote from me, read the whole thing before you do.
  6. Untamed_As_Thee

    Fix The Game Then Send Me A Message Its Fixed

    Yes, you've quoted 'your' formula many times in this post, and I've proven to you why your formula doesn't work with two counterexamples, and why it means nothing. This is NOT what the median means, or how it's defined. On a continuous approximation of W/R, the median is defined as the W/R at which half of the players are below and the other half of the players are above. It would be easier explained using a cumulative distribution function, but I guess that would be too much. Don't get this confused with the arithmetic mean, otherwise known as the average, of W/R. Since you're really muffled with your definition of mean, mode and median, I'll write them down in an elementary form in hope that you would understand: Mean - summation of all W/R within the population divided by sample size, or average, Mode - highest occurrence of W/R within the population, Median - the W/R entry at which half the population falls below or stands above. In my first reply I've quoted the Empirical Relationship. This is not something made up by me; it's a well documented argument from Pearson. Also, please stop using the median as 'average'. It's wrong. No matter the population is active or not, it does not change from the basic understanding that the mean is just marginally under 50%. Each win and loss is independent from each other. Nothing guarantees you to win again just because you've won your first game. You've also acknowledged that the mode is between 47% and 48%. The median is simply inbetween. Period. Your assumption of trimming off equal amounts of W/R on a skewed distribution (which is already a big no no) doesn't make any sense. Once more, I prove to you why your formula doesn't work: Range is the upper W/R minus the lower W/R. You've chosen them to be 70% and 30%, thus the range is 40%. (Upper limit - lower limit + lower limit) / 2 is simply half of the upper limit. Do you get it now? You don't calculate the middle value of two numbers by halving the larger number. You add the lower limit with half the range. If you're going to continue quoting me your section of wrong formulae, go ahead; all you're doing is making a display of yourself repeating your mistakes. Actually read your sources, namely that link of your first reply, and read my counterarguments would help, as a starter. It is quite obvious that you haven't read any. Enlengthened copypasta doesn't make your argument any more valid; it demonstrates stubbornness.
  7. Untamed_As_Thee

    Fix The Game Then Send Me A Message Its Fixed

    So according to your logic, if one were to take the sample from 99% to 100%, the median would be 50%? You've trimmed the upper edge, but you don't bother trimming the bottom edge? Are you serious? To extend on that, say we only take the data point surrounding 50%. According to your logic, since the range is 0%, and the lowest W/R is 50%, dividing them by 2 gets you 25%. What you then have done is calculated a probability outside of the probable probability space. Stefan Banach and David Hilbert both struggled to grasp this concept - if you think your maths is correct, write a thesis and start submitting it to the Nobel committee. I'm sure they will be impressed! You went wrong on this: you've simply halved the upper limit. Range is the upper limit minus the lower limit. With your calculation, it's the upper limit, minus the lower limit, plus the lower limit, and then dividing the whole thing by two. I mean, it takes less than some intuition to think that in no way would the median be 35%. And just to clarify, the highest occurrence of an event is the mode, not the median. When you're taking a continuous approximation like the Normal distribution we have been discussing, there is only one mode, and it's the turning point, also known as the largest zero. Mean is at 50% and median is inbetween. Read your own link that you've posted. Please.
  8. Untamed_As_Thee

    Fix The Game Then Send Me A Message Its Fixed

    Nowhere does my model include active and inactive players, nor does it require that. The distribution model we are talking about here does not require ruin theory. If you are indeed going to trim, I would suggest from 40% downwards and 70% upwards. As said before, this model contains positive skewness; you can't use a made up deviation and spread it from the mean. Active or inactive has very little effect on how this distribution model works. Each sample is independent and identically distributed. How I play doesn't affect how you nor OP plays.
  9. Untamed_As_Thee

    Fix The Game Then Send Me A Message Its Fixed

    In no statistical measure do people use median as 'average'. The average is always denoted with the mean. The mean of WoWs population is ever so slightly below 50%. Half the players win and lose per match, and draws counts as losses for both parties. The mode is around 47% to 48% as mentioned. That's where you see most of the WoWs players are around. The median is inbetween the mode and mean. This relationship where median is bounded by the mean and median is known as the Empirical Relationship, and holds almost surely for most distribution. The model we are using here to approximate W/R is a Normal distribution model with positive skewness. I think you struggle to understand the fact that 'half of the players are below/above 50%' and 'most players are at 47% to 48%' are two different statements. Let me demonstrate with a simplified model: imagine there's 4 players in a game with the W/R of 25%, 25%, 50% and 100%. The mean is 50%, The mode is 25% since it appeared twice, And the median is 37.5% since it's between 25% and 50%. If you're going to post a web link, at least read it through. There's literally a picture explaining how M, m and μ works.
  10. I want to volunteer to join the Rogues!
  11. Untamed_As_Thee

    On the CCTP&Missouri

    Now that Hapa isn't around to delete my comments and call them 'inflammatory', let's dissect this. Will they be missed? Let's look at what one of your major spokesperson has to say. Yes, you thank them for what they do by ignoring them. Good job. Just so you know, people who suffers from abusive relationships and end up getting divorced usually have a restraining order against their ex-partner. You think too much of yourself if you believe they will talk to you again. There is a difference between 'several' and 'many'. I guess you really do need to hire a 'Teacher of English'. Let me find a few examples: - Yukon was a big PR disaster following the Puerto Rico PR disaster. - Asashio has received none but bad reviews from many CCs in Asia. - CV rework has led to more players gaining an interest in them. In return, they have to be limited to almost constantly 1 CV per side (sometimes 2, rarely 0) because of how popular they have become. One big step backwards. - The direct sale of ARP Yamato and Napoli showed how disgustingly hungry you can be for doubloons, or money. Need I say more? Also, which one of these problems I've mentioned above was 'fixed' beyond a hollow apology, much like this one? Empty apologies, empty promises. I've said this before, I'll say this again. You are about as apologetic as Sub_Octavian is in his Pay 2 Rico statement. Shame on you.
  12. Disclaimer: this is written under confusion, not rage. Again, this does not represent anyone's opinion but ME. The last thread was removed at roughly 80 replies due to the thread offtracking into general WG complaints. So I was told links and 'inflammatory' content cannot be posted on here. No worries, I have already done some calculation based on my experience. I shall use similar/same values as the last thread. 11 successes out of 131 trials. First test: odds of getting a ship is dynamic and not equal to odds of getting flags. X ~ HyGeo (131, 11, 13) Assuming null hypothesis is true, H-nought: Probability of getting a ship is 11/131; H-one: Probability of getting a ship is less than 11/131. Testing at 5% siginficance level, The(my) probability of getting no ships out of 13 trials is 30%, which is more than 5%. Critical value takes 31 trials and no ships. Therefore, null hypothesis is accepted, odds of getting a ship may be the same as the odds of getting flags. Second test: one needs to pay more than he needs to for getting a ship. This test assumes both Georgia and Alaska costing 19500 doubloons. For me, converted currency costs about 19646 doubloons but taken to 3 significant figures. X ~ N (11660, 1652265) Assuming null hypothesis is true H-nought: Mean price for getting a ship is 11660; H-one: Mean price for getting a ship is higher than 11660. Testing at 5% significance level, The(my) price I've paid and still haven't gotten a ship is at 0.0000000397% (you count the zeros, 3.97*10^-10). Critical value is at 13714 doubloons for a ship. Therefore, null hypothesis is rejected, one does need to pay more for a ship than one needs to. As null hypothesis is rejected, Type I error arises, and probability of Type I error is 0.41%. Final comments requires constructive ideas so this doesn't seem to 'unfriendly'. Here's my suggestion: - Remove flags and make the 2021 Independence day bundle only ships with 1500 doubloons. - Make unicorn ships available for purchase with this doubloon bundle i.e. Ohio and Enterprise. of which I expect both of them to be perfectly doable. Perfectly. And at last, I would like to apologise for the last thread. I am as apologetic as Sub_Octavian was over the P2R fiasco. Best regards, JonTheCadet
  13. Untamed_As_Thee

    [ALL] ModStation

    I will do that next time when the error comes up again. For now, I've found a way(?) to somehow fix this with witchcraft. - Remove all mods, delete and reselect directory - Download the latest Aslain modpack, and choose whatever you want - Wait until 'Error 5' occurs, then abort - Restart PC - Launch ModStation and choose whatever mods you desire. Not sure how this works, but it worked for me. Best regards edit: that was luck, it's now broken again
  14. Untamed_As_Thee

    [ALL] ModStation

    All folders are indeed closed. Last time I quasi-fixed it with a restart. Now I must have restarted the computer about 5 times and each time the same error still pops up. Oddly enough, when I launch WoWs through Steam alone, the mods are still there. Best regards
  15. Untamed_As_Thee

    [ALL] ModStation

    Build 1903. Does it make a difference?
×