Now we are just flipping coins to explain the rank system… no wonder it’s so messed up…smh.
4 Likes
Because it’s based on math?
Or because people would rather believe that the system is somehow broken or conspiring against than accepting their rank.
Again, your’e applying a binary example to a non binary situation.
Roll a twenty sided die twenty times, don’t flip a coin. The results will more likely than not appear far more disconnected then or service record data consistently
But it is a very, very binary outcome.
There are only two possibilities;
- Harder match → Probable loss, and lower KPM → lower expected kills.
or
- Softer match → Probable win, and lower KPM → higher expected kills.
And just like flipping a coin it is not uncommon to see runs of harder or softer games. After a win (or run thereof) it’s a 1:2 chance of a loss. 1:4 chance of two losses. And 1:8 chance of three losses. It’s perfectly natural.
So I’m not sure which bit is 343 supposedly manipulating with their 20 sided dice?
I could care less about my ‘rank’…i don’t think my rank should be reflective of my teammates actions in a particular game. If my teammates aren’t addressing the proverbial ball, why are they even playing? For fun or for ego. I’m fine with either I don’t judge, but if it’s an ego thing I’m too old to spend energy on that. The higher ranked player usually thinks they are better than the lower ranked player, and this is a sickening of the mind and soul…the ranked system perpetuates this…As Darwin has said, I would say make it more of a progression system, and let the KPM and KD speak for themselves in halotracker.
Mathematically speaking…the results get skewed because of variables such as jealousy greed ego during the game. Onyx don’t grab the ball or run in hill until fodder is killed first…if the fodder aren’t teamed up with the high player they may give up or quit. Whether true skill admits this as an actual mathematical variable or not i can’t say, but they should.
When performing statistical analysis (sounds like that’s the approach you’re going for, I could be wrong though), it’s generally bad practice to transform values from continuous numbers (integers is probably the best tbh) if you don’t have to.
Taking a data set of values that could be any continual number and transforming them into a binary loses a ton of useful info for your model. Generally this has to be heavily justified if it’s part of a research project with any intent at being published or applied by a business.
A better approach (again, this time slightly modified for you) would be to roll a 20 sided dice 20 times (40 times would be best) . You could mark any value 0-10 as a “low performance” and anything 11-20 as “high performance”.
That will give you a much more synonymous example sample to the real world situation.
That’s a huge part of stats, transforming between distributions.
You can argue that any number of variables that determine skill shouldn’t translate to an even win lose outcome. That’s not math though. That’s based on what you define as a good experience. The fact that they can predict and coordinate our outcomes at all by how they evaluate skill/rank is proof the math works out. That’s not up for debate.
Feel free to argue that this is a bad way to reward people that are more skilled, or whatever your jam is. But please stop pretending any of these opinions are based on math.
2 Likes
Yes, it’s a huge part, but you’re still supposed to minimize it as much as possible. I just think there’s a much more accurate approach to setting up example data for the discussion than coin flips when we’re talking about predicted kills and predicted deaths, which are definitely not represented well by “yes/no”.
This just isn’t true. If your objective is binary that’s what it is. There’s no loss of information if that is your objective.
The term “accurate” is misleading here. One side is arguing that win-lose is a satisfying outcome to attempt to balance out the player experience. The other side seems to be arguing that balancing the KDR is a better way to improve player experience. None of this is about which is a more “accurate” measure of player skill. It’s about what objective measure should be balanced to improve player experience. If you want 1-1 kill-death instead of 1-1 win-lose that’s fine, but none of that has anything to do with accuracy. It’s a matter of opinion which provides a better player experience in the long run. None of that is math. We don’t need to flip coins or argue about link functions into binomial distributions.
The coin flips were just to show the “waves” of 2-4 up and down periods are just due to chance. The match maker finding you games a bit above and below your station (heads or tails) show the same patterns.
It’s not 343 punishing you for doing well.
And I’m still not sure what is so ominous about your “expected kills”.
It is your KPM vs that opponent rank multiplied by the game time. It is not a value 343 are making up on a whim to deliberately manipulate your rank. It’s just a reflection of your past performance.
I see a lot of back and forth between people here about win loss, performances, and the expected results where you know you’re heading into a match you’re expected to lose.
Beat people your level or below you, lose to better players right? Makes sense…but here’s the question. How much better are the players in the lopsided matches? It its too substantial of a gap…then thats bad, definitely not good. At that point it isnt about you reaching your skill ceiling. You can’t make a giant jump like that to “prove” you got better. The gap is too big.
Is that what you feel happens often?go on a win streak and suddenly MM throws a series of lopsided curveballs you feel are deliberately there to claw back?
2 Likes
It’s gatekeeping. People that figured out how to game the system got onyx early and everyone else lingers in lower tiers regardless of their skill levels.
1 Like
Sounds frustrating man. I don’t get a chance to play a lot, but I see what you mean. To have that feeling every session would make playing the game discouraging.
1 Like
Making those rankings public was probably a mistake. It provides little reward to achieve and creates a nagging feeling of incompletion for the rest. Stats are interesting to monitor as a separate indicator for personal growth. I’d love to be able to easily scrape the raw data so I could teach big stats in my lab. No real point outside of that
1 Like
This.
People are losing their minds over a handful of points. Which are pretty meaningless.
I really don’t know why 343 made the CSR the same scale as the MMR. Or why they felt the need to show everyone’s CSR change at the end of each match.
Change the scale. Add a maximum cap. Find something else for people to “grind”.
Anecdotally the gap isn’t huge. The average MMR’s are usually fairly tight. When the level of opponent goes up, so too does your team.
And I haven’t seen anyone ever post numbers that were concerning. You can see the average MMRs of the teams. If there really was a concerning trend it would be easy to show and subsequently plastered all over the interwebs.
Having said that, it may be a little different as the population drops and the ranks widen.
But to be honest… most of the posters who I’ve seen complaining about “unfair” games - when you look the actual games were close enough - it’s just that they get frustrated that they went super positive and still lost the game. No appreciation that there are more ways to carry a side than just a high KD, or that your weaker links can still contribute with a KD less than 1.
When games were evenly matched it feels more consistent. You play well (KD > 1) and win, or you play bad (KD < 1) and lose. It’s harder to feel that balance when you win or lose on a high KD or a less high KD.
It’s frustrating as Hell that the squad manipulators are grinding their CSR up and up.
It does affect everyone as the system forms a normal curve. If you inflate the top end it has to have a skewing effect to the left on everyone else.
But at least we have seasons. Everyone get’s to start again (albeit with your previous MMR as a seeding point). And hopefully heading into Season 3 we have some significant improvement from 343.
Im my discord i have a friend who sometimes hops on and his lobbies are filled with onyx 1900s. He floats 1500s, goes back doen to daimond because of those lobbbies. He gets those matches when he searches alone. He gets those lobbies with a diamond as his team mate. They just inevitably happen after he’s climbed up from a win streak. During his win streaks, he beats challenging players, but they’re pretty much within that range that just seems fair. The losses are extremely lopsided. I mean if you’re being smacked down into daimond when you were say 1550…but you’re being smacked down because the opponents are 1700s 1800s 1900s…while you’re the highest ranked player on your team normally…all the while even get lower ranked players on your team…well…you can see where the frustration comes from right?
He’ll post the scores sometimes and I just tell him man it has to be population, time of day, etc…but…it just doesn’t seem so man. Its just a up down teeter totter. So thats where I’m making that point. You are absolutely not at your skill ceiling if the losing matches are from 1900s players being smacked down to daimond repeatedly after winning. How can that be? I mean some professional tier players are in that range man. How can the player gradually go up with scaling opponents…if the jump goes from a small margin, to a sizeable gap?
I’m assuming others go through it but just don’t make threads. Like I don’t even think that dude uses waypoint. He just plays the game but…I can tell he gets tired of that. It makes the player feel like the session they just started is destined for a downrank session. It’s not that they’re upset at the sheer downrank always, but its more that the scalng of players is all out of wack for some players. We like to try and chalk it up to population…ideally I’d like to believe that. I wish the game had a population counter like the old days.
Though to your point, I get that, and it’s a fair point but I don’t believe that is always the case. Plus the MM has been weird latley. I’ve watched a few streams of pro players getting bottom half of the skill spectrum team mates on their team. Has to be a bug after the update man. Its really hard to believe for example that at the very least a daimond couldn’t have been found in place of a gold you get me? Maybe I sound crazy or it may seem like denial of a dying game…but is this game that far on its death bed already? Holy crap.
1 Like
I’m getting mid tier plat teammates consistently against onyx teams. I even got a gold 6 one game the other day. It’s wildly mismatching parties based on temporary MMR dips and spikes