Jump to content


Mathematics Contributor
  • Content Count

  • Joined

  • Last visited

  • Days Won


Everything posted by MaxL_1023

  1. It doesn't pass the vertical line test = sqrt(4) = +2 AND =2. You just assume only the positive answer.
  2. Technically the square route of x2 is not even a function since you have both the positive and negative solutions everywhere. When you meld the exponents and just get X you basically pick one of them based on what half you are in. x/abs(x) would work - you would get (-x)/x for all negative x and (x/x) for all positive x, and at zero you can define it using limits which converge from both sides. How can it be continuous when at -0.0000000000...1 it is -1 and at 0.00000000...1 it is 1? You have to have a discontinuity there. You could also use the u(t) function - x = u(t) - u(-t). Y
  3. The weakspot aimbot would often hurt at longer range - you are aiming for the edge of the tank and therefore more likely to miss completely. The laser mod would be broken though, especially if there was a HUD popup with a enemy gun view.
  4. If they reconfigured it so that your tournament standing was based on an ELO type system it might work - even matching the best teams together would lead to them having a high rating. Otherwise this is a clusterfuck. WG already made skirmishes shit by making them gimmick contests between undersized teams with random restrictions. Now they make it so that the points standings are compressed and RNG has even more of an effect. Next thing you know the battles will be for fun and every team will get a fixed 500 gold for showing up.
  5. My point is that WN8 as it exists now is not an accurate rating. Wargamings Personality rating was shown to correlate better with WR than WN8.
  6. If you need to use a scale factor like that then it proves that the initial method is inadequate. This would just be a using a patch over a gaping maw.
  7. You would think that when I (Mr. YOLO DERP IN A WAFFLE) tell you to flex to defend base you would flex to defend base, instead of sitting for 2 minutes waiting for a T-54 to peek around a rock 350m away.

    1. MaxL_1023


      Use the 150 cm

    2. MaxL_1023


      Special gun from Warpack+

    3. Masterpupil2


      ayyy there's the old platoon-shit-talking MaxL we all have gotten to know!

    4. Show next comments  15 more
  8. I don't know how to do it in R, but later I will try and figure it out.
  9. Can you take that scatter data and convert it into an interpolated PDF? Basically, calculate the "mean" rwinc for all the accounts at a particular WN8. It will make it easier to see any non-linear trends which pop up. Additionally, that pure correlation might not be what we should be looking for. I think WN9 should predict a PDF of expected win percentage. Match this PDF against real account data. You will always have variation and outliers - the best model does not necessarily have to directly predict them but instead simply predict their frequency.
  10. That graph is nowhere near an accurate model. It shows that T-62A performance is not linear with respect to WN8, which indicates that the model used to determine T-62A WN8 is not adequate.
  11. So what is this trying to accomplish? You can't force every tank's WN8 to the account average - it ignores the fact that players may actually perform better or worse in certain vehicles. It would make more sense to change the WN8 formula so it is not full of ridiculous outliers like artillery and the T-62A.
  12. The final result likely wouldn't be that much different than WN8, the issue is that right now there isn't a lot of statistical justification for what WN8 does. It uses the same forumla for every tank at every tier, bases performance as a ratio of a top 10% player and has obvious errors as a result. In short, you are linearizing one too many times and in the wrong place - the 1565-~54% range is not a fair approximation for the 49% range or the 60% range. As a first step I would like to see the regression which developed the WN8 equation repeated for every tank at tier 10 to see what d
  13. Does capping in a heavy make sense on Mali when they have a scout and arty alive?

    1. Flametz


      Fuck that was my LTTB... And I was wondering why you got off cap initially.

    2. MaxL_1023


      I thought my plat mate would come with me. 2 of us at close range would have wrecked the side.

    3. yolorush


      No you talked so much shit I pretty much told you I was done with you. In the past one could last maybe 10 games with you. Nowadays its down to 2. You need to check your fucking rage. Its getting sad.

    4. Show next comments  15 more
  14. You use the overalls to develop the model and then just use anything else you get as validation data. If mismatches start becoming significant we re-parameterize.
  15. Those just represent an interaction between survival rate and those factors. The final model will certainly involve multiple non-linear terms representing combinations of everything. A 3D scatter plot between damage/battles, frags/battles and victories/battles is quite tightly grouped. This does indicate that damage and frags are correlated but it shows how multiple factors affect the response.
  16. \ This is a WR distribution for all EU server T-34 drivers with between 350 and 360 average damage and more than 50 games. It is roughly normal. Same for 450-460 avg damage - a bit more distorted. 550-560 avg damage. Here sample size is becoming a bit of an issue. My idea is to take these binned normal plots and recalculate one starting at every damage point. Then, we can plot their means and variances and goodness of fit to see what happens. Sadly I am still red@R so it might be slow going to start.
  17. So if I sort by that I get all the data for the same tank?
  18. Why can't R use the same language as MATLAB? FML I also managed to crash it trying to sort that huge dataset. I am assuming that tankid is defined underneath countryid when it comes to tank type?
  19. They can't buff HESH, otherwise it would be game-breaking. If HESH normalized more than 10 degrees and didn't ricochet until 85 (like heat does) it would basically negate armor facing. Anything except for the Maus would have spots which 200 HESH pen could reliably go through frontally. Heck, my 183 even now (after the pen nerf) can go through the LOS front of tier 10 heavies with 230 HESH often enough to make it worth loading when I have a clear shot.
  20. I an going to try and figure out R and see what I can get. I want to interpolate those graphs into a continuous PDF of damage vs win rate for each tank. Then we can see how the mean and distribution of win rate changes with damage. We want to see how accounts at any particular level are distributed to try and filter out platooning or obvious padding. In fact, on some of those graphs as it is I can see a second set of points which appear to be platoon-padded values.
  21. Can you try and fit those to gaussian and gamma distributions?
  22. It forces all the scores towards 50/50, making RNG have even more of an effect on placement. Gotta cater to average teams upset at being wrecked!
  • Create New...