The Performances from TrueSkill2 line up with the kill counts for that match. The 19 and 15 kills players both had more kills than expected so their MMRs went up. The 12-kill player had the highest MMR before the game, and they killed as expected, so didn't change much.NBK DarkWarlock wrote:For old times sake I will give you this game to analyze.QX wrote:Except a moving bar with wins and losses doesn't imply any skill.Cool thanks for the info ; still sticking to my Line of reasoning lol .Player skill swings all the time, both up and down. If you dip from Onyx to Diamond, then either you got worse, a lot of players got better, a lot of players got better faster than you, or a lot of players got worse slower than you. There are a lot of ways that can happen.Thanks for the info and response , but in my opinion I still find the system faulty . Not rankimg up after going 17-3 doesn’t make any sense .In addition, your kpm vs. Onyx players is only 1.06, which is low Diamond.The Ranking system has not changed one bit and is, in fact, more accurate than ever.
lookup my GT on halo tracker for date 11/17/19 : flipflex
as you’ll note, I went 17 wins and only 3 losses. Overall kd and kda well into the positive range .
my rank netted out to basically be the same as when I started the night .
this doesn’t make sense . Such a high win percentage (if as mentioned wins being most heavily weighted should yield a greater CSR gain in total if the system was functioning as intended .
either we are playing under a bugged ranking system or we are all prisoners to a faulty system working within the confines of a depleted halo 5 population where it may be time to apply a new algorithm .
thanks in advance for reviewing and providing feedback...
please return to the ranking system that was present previously it was working well !
Win percentages and KD are ignored and irrelevant. As stated, it matters who you beat, and what your kpm is against the skill of your opponents. If you are winning and not going up, then you are performing as expected. If you beat a team that the system expects you to beat, you will not go up. If your kpm is the same as predicted for that match, you will not go up.
Population is not effecting this. If you play a team that is too easy, you should not only win, but have a high performance. This is not happening in your data. When you play easier teams, your performance is as expected.
I see no evidence in your recent performances that you deserve a higher CSR. You have been playing like a Diamond 2 player (see graph):
If you were Onyx, you would have a kpm of 1.56 against Onyx players, like the average Onyx player does.
Your win% against Onyx is also only 27%
additionally, our team never played “ a team of onyx players.”
we played teams with occasionally an onyx sprinkled in with several diamonds .
i was an onyx about a month ago , and I was playing the exact same way as I am now .
if the system was working correctly I never would have been able to make it to onyx using the logic you mentioned as it would have been mathematically impossible unless I “ beat your system “ but I heard this new algorithm prevents inflation . This isn’t the case if I made it to onyx and my mmr is d2.
You have played 11 matches against teams that had an average MMR between 1500 and 1650. You have won 3 of them.
but anyways back to halo I don’t care about rank that much anyway just fun playing but ya would be nice to see a moving bar with wins and losses simple as that .
Think about it this way. Say you are the best player in the world and play a 1v1 playlist. This would mean that you would never (or very rarely) lose any games at all. That makes sense because you are the best player and therefore would win more than anyone else. That also means that your "moving bar" would never stop moving up and only be limited by the amount of games you can put it. However, player skill doesn't work like that. Even if you were the best player in the world your skill would reach a personal ceiling at some point. So in order to have an accurate measure of skill, your CSR and MMR rank would top out at some point (simply because you cannot perform any better than you are possibly able to do). In that scenario a better measure of skill would have your CSR rank stop increasing and top out at whatever point your skill is. From that point on wins shouldn't net you any CSR since you have already achieved the correct skill level equivalent in CSR. You would be the highest and number 1 but your skill would be shown correctly.
Now apply that to your current rank. You have been perfoming at a low diamond level of play (shown by Josh, the guy who has data on every game ever played in H5 and every player). That doesn't mean you have been losing games. All that means is that in the games you have been played you have performing at a diamond 2 level of skill. Therefore even if you win your CSR shouldn't increase significantly because you aren't at those higher skill levels. The same thing goes if you lose. If you are performing at your current skill level and you lose you shouldn't lose any significant CSR.
What you are asking for is a ladder based system which is misleading in terms of skill. Being onyx in previous seasons means nothing because player skill varies, population skill varies, and they do consistent batch runs to fix misleading CSR ranks.
Tell me the best player on my team and who is the worst.
The 4-kill player did worse than expected, but not much was expected from them in this match so their MMR didn't change much ( Plat played like a Plat).