Mathematics Contributor
  • Content count

  • Joined

  • Last visited

About Gryphon_

  • Rank
    Goes Hill on Himmelsdorf in a KV-5

Profile Information

  • Gender
  • Location
    Los Angeles
  • Server

Recent Profile Visitors

2,240 profile views
  1. The strategy for WN8 is that there might be one more update in the usual sense, then the 'final' update will change all the expected values for all tanks to an average by type and tier. So, for example, all tier 8 mediums would then have same values. When the websites have recoded to use the average expected values, no more WN8 updates will be needed. This 'endgame' is necessary as its a lot of work to do updates due to new tanks, etc, and interest in doing that is waning along with interest in WN8. But we dont want to break any websites or mods that use WN8, so this strategy is designed to ensure they will always have valid WN8 data.
  2. T-34-3 needs some love. A third party modder did an HD model for it about a year ago, still nada from WG. Tank is almost identical to the WZ120 so it would be nice if it got the 120 armor model by mistake when updated. It needs some love.
  3. The problem with the tier 7 lights promoted to tier 8 - after all the discussion on VR - is the alpha: 170 / 180 doesnt cut it at tier 8. It just takes too many shots to have an impact on the battle. Fast reload is nice but it means you have to expose when still lit if you want to use the dpm. Overall I think Bulldog and LTTB are both weak now, compared to any t8 medium anyways.
  4. I'm good with the new stun mechanic. Good examples was today, playing Serene Coast, big medium brawl in C1, so was getting artied a lot. Was stunned over and over, taking a few hundred damage most times, but that's better than being one or two shotted.
  5. Get well mate
  6. It always helps to listen to others opinions of tanks you own, as it often opens your eyes to some alternative configuration or playstyle you hadn't considered or had too quickly rejected. For example, with the WZ line the 100 vs 85 debate lives on, rammer vs vents, and whether recon and SA take priority over repairs and camo. And so on. Another good example is how many think the M46 Patton has an awesome gun; I look at the dispersion and AP pen and frown, they look at the soft stats and smile, so I give that some thought., etc The thought I would offer on the WZ-131 is that scouts have to have VR above all else. The 131 was nerfed to 370, whereas the T71 (and Type 62) have 390, so not good there. The 131 lost its 100mm, but the 85 looks decent apart from reload time. The hidden gem with the 85 is the HE round that does 320 dmg against light tanks... if it pens, that hurts. I always carry a few HE rounds with an 85mm Chinese gun. If you can get a Type 62, sell the WZ131, as you can run your 132 crew in the 62.
  7. Its amusing to see the PTA, on paper, can do a better job of scouting than the 9.18 lights. Put PTA side by side with 132a and 54lwt and you'll see what I mean. Acceleration is lower than any light but if you have a long way to go it gets there fast. The only bummer is that recon and SA are both on the commander, so no divide and conquer for 10m more VR. But the speed, the gun, the camo, the low profile, awesome.
  8. Plan is to move to one set of expected values for each type and tier, using a weighted average. So - for example - all tier 8 mediums will have same values. Apart from zero incentive to play bad tanks, it's all good. Best of all, there wont be a need for updates every patch or whenever a new tank comes out. At this point in the lifecycle of WOT, it will be fine.
  9. 132 is working well. Its a lot of fun when top tier. I use the top gun, better dispersion helps. RIP all the lights in lower tiers, though (except the prems)
  10. Base xp would be useful, but the API only releases XP with all the bonuses included - premium account, boosters, etc. It's no use. WG PR is pretty good as-is for an account-level rating, but it doesnt do per-battle or recent.
  11. I was 31-50% and was surprised to miss the tanks entirely. Well at least I only watched the stream for about 20 mins, so I cant complain...
  12. T54lwt feels good to play but the gun isnt great, accuracy is poor and the pen for the premium round is only 225 - which is a tough sell at tier 9. I just lost a battle because I had 3 cappers on my cap, all low hp heavies: Lowe, T10 and E100, and they bounced all my stuff easy, even the gold
  13. Its important to keep using the infrastructure in place around WN8 rather than try to introduce this as a new rating and kill off WN8 as such. Lots of websites pull WN8 values to feed their webpages and the intent here is to keep those pages current and complete. We can achieve that in the short term by issuing a regular expected values update where the values are in fact averages. In future, for every new tank that comes along we just apply the appropriate values for that that type and tier. In time, the sites will rewrite their code to accept the much simpler expected values table we can generate this way (just 50 sets of values - 1 for each type/tier combination), and then life is easy and the good ship 'WN8' will sail into the sunset with no-one at the helm.
  14. I resurrected my old stuff last night and thought more about how this would work. The easiest way to get started is to show the current expected values in a boxplot (for each tier). The boxplot shows how the values change per tier, and the spread of values per type/tier that we aim to consolidate into one (the mean). Once we have that first set of mean values I can run the usual WN8 method with the usual filters (50 battles, upper 50% of tanks by WN8, etc) and use the regression method to fine tune the first set of values to the latest data. I tried this using old data and the changes are small. I established years ago that averaged values do just as good a job at the account level of correlating WN8 to rWINc so we are good in an academic sense. Finally, as the data is historical, we need to make some decisions about how to handle the tier changes and new tanks in the light line. To avoid big changes in player WN8 we should consider uptiering the values we just created for the tier 7 and 8 lights as 75% of them have gone up a tier, and then extrapolating for tier 7 and 10. Its anyone's guess as to whether the lights will be more or less effective after the changes, but we dont want big changes to player WN8 if we want to stay under the radar and avoid a shouting match. We will also need to agree on the 'message' to go out with the change, if we end up doing this (come back Kitten, all is forgiven!). Players will quickly realize that playing bad tanks will hurt them (and already has, as the new values will punish those who played bad tanks). Thoughts?
  15. Here are the update notes: patch hits NA tomorrow morning.