Ok, we know that “win games go up. lose games go down”
And that you stabilize at a rank, where you will land around 50% win rate.
We can test that it doesn’t take that long to hit a stable area.
We can show that you don’t bounce around ranks all that much once you are there.
We can test that putting someone else playing on an account can push the account towards their rank quickly.
So basically you don’t think that ANY of the above is happening?
You have left rational skepticism and are off in flat earther land. “NASA could be controlled by space aliens and could be hiding the fact that the world is flat and infinite, and gravity doesn’t exist, it is just the aliens are using special beams to push things around in a completely consistent with gravity way”
Which of the above do you not agree with, we can construct a test, OR show from data we do have this is the case.
So you think limited sample size anecdotal evidence is sufficient to prove the algs/code are fair and free of nefarious bugs, errors, flags, or rigging?
You have yet to show quantitatively correct i/o mappings.
None of those things show the MM is specifically unbiased, or that the rewards are specifically sound and complete.
Except your “testing” was disruptive to what you’re actually testing:
Give me a tank that is not a moron charging 1v5 to the point, completely alone, despite using voice chat and command lines, asking for regroup; for then ulting with 20hp
DPS with enough brain cells to understand how important it is to peel.
I know they usually think POTG say they were the goat during match, but they have to stop living by the mastersAim-GoldBrain rule.
If DPS Uninstalls TikTok, they’ll gain 7 divisions before the weekend.
Supports that aren’t one trick on the very same character I use?
Like, is this FFA or Role Q? I mean, are we really having 2 specialists for 1 hero? Hello?
Oh, and pressing tab to see there will be a constant threat in the backline, to know Who, and Where, are the priority targets, so you can aid them the instant they are in danger.
THAT WOULD BE BLESSED
That’s about it
You know, things you gotta put up to against the average deliberate sabotage from the algorithm when assigning bots or players 3 categories bellow your rank
and what’s even worst, teammates above your rank, and with this behaviour xD
I see no math here. Nothing quantitative. No parametric or non-parametric asumptions. You have shown nothing that verifies the internal integrity of the code/algs. There is nothing there that guarantees blackbox correctness.
You have shown nothing to reveal the internals of the system are bug-free and specifically working for all players. You’re operating on tautological, anecdotal observables.
I give you a list of things which would show the matchmaker is working.
I ask you which part you think shows it isn’t working.
You can’t and will not answer.
Here is the list again.
Ok, we know that “win games go up. lose games go down”
And that you stabilize at a rank, where you will land around 50% win rate.
We can test that it doesn’t take that long to hit a stable area.
We can show that you don’t bounce around ranks all that much once you are there.
We can test that putting someone else playing on an account can push the account towards their rank quickly.
As someone who used to be a legit bronze tank player and now a GM player, I can tell you through experience how it works to me personally. The matchmaker in the long run will not stop you from climbing. While this is true it’s 100% designed to make you play as many games as possible before you hit your true rank. I don’t care what anybody says. If you have a good day loaded with wins be ready bc next time you’re on your games will be god awful. Then you’ll start winning again and still be net positive on your climb. It’s just taking more games which is what it wants. I 100% believe this
There are plenty of explanations for what you and other people are seeing. You could be doing particularly well that session and then the next time you get placed at your higher MMR you aren’t playing at quite that level for a number of matches. There is also a lot of inherent randomness involved as anything can happen in a match with all new people.
People also love to find patterns in things when none exist.
Ultimately it comes down to playing harder matches when you win and easier matches when you lose. Play enough matches under this assumption and you will approach 50% win rate unless you are the best player in the world or the world player in the world.
The whole notion that Blizzard is rigging matches so people are certain to lose a particular match because they did too well is absolute bonkers nonsense.
Incorrect and incomplete. You have to show quantitatively that the algs/code are working correctly, not just producing i/o that seems reasonable.
You have not provided the appropriate statistical assumptions (for data set, for sample size sufficiency, and coverage wrt function complexity), and you have not shown what algs/code you have reverse engineered from testing and what the integrity/fidelity is.
Your claim the system is working is based on anecdotal gaslighting and can be dismissed as such.
You still have no specifics for how it works and whether it is correct.
I’m just giving you my opinion as someone who climbed from 1000SR to GM. I had a session today a span of 5 games I lost all 5 and played well in all 5 matches. I lost one match as ram where I went 30-3. My sojourn was 8-19. In all my losses I had a massive dead weight who had no business in a GM lobby. I got back on a few hours later and won my next 7. A net positive for the day. I see people on my team and on the other team who have no business whatsoever in these lobbies
100 matches or even 1000 matches is circumstantial evidence at best to make such a claim. I’ve played over 25,000 matches and have not witnessed any such rigging, but even that is still just circumstantial.
You talk about events of particular matches, 5 matches here, 7 matches, but none of that means anything. It’s like talking about how when you threw a coin against the wall that sometimes it did 10 flips when really you think it should be doing 15 and then 7 times in a row it came up heads and the next 5 times it came up tails.
There is endless randomness involved in a match of Overwatch as there are many players who can do anything they want. But over time you approach 50% win rate because it gives you harder matches when you win and easier matches when you lose. The amount of buffer on each side of your current MMR on average that could be gained or lose on pure random luck would be interesting to know. I’m sure that would come into play when people end up too high or too low from where they probably should be.
Just gonna leave this here again like the other thread.
And this one…
Which is a problem I experience in every game, that mysteriously has disappeared for roughly 1 week when purchasing the last two battlepasses. In fact in those weeks I hit shots while playing Ashe that I swear should have missed. Effortlessly climbed subrank in DPS and Support until the week is over, then games became insane effort again…
The amount of code and data which in involved in it is probably crazy and can likely give hints to the inner working of OW which then can be exploited and manipulated
You need to include MMR and SR calculation, including everything that is tied to it.
For sure you could go through the code, but likely without a degree in the right field you would not understand why
Second: Matchmaker and MMR/SR calculation is worth a ton of money and you don’t want to give out for free
These are NOT built into the matchmaking engine (at least the one I write)
I really don’t give a rat’s booty what MMR calculation the end company wants, or the weightings for ping times, or the pity timer around how long people wait for games, etc.
Not my circus, not my monkeys.
Even if you got my entire matchmaker, it wouldn’t tell you a damn thing about the calculation, it is processed outside of it.
This neatly lets me sidestep that ENTIRE mess.
And make no mistake, it is an unholy mess as I am sure you know.
Yeah. I hit more shots, accuracy went up. But also enemies hit less shots, teammates hit more shots… etc…
Latency reality warping disappears for the week. When it comes back to normal after, nearly every game I have multiple reality warps. Cooldowns go through enemies but don’t register, my last 3 seconds of action doesn’t register in the game server and I see in my deathcam my hero is just walking aimlessly… etc.
as I am sure you know.