he thinks the system is infallible and the most accurate system possible, is the problem
There is no budging of opinion on this
This system completely ignores OBJ skill. Which is why a common complaint you see is “OMG nobody is playing objective! someone grab the ball!”
Those that ignore OBJ and just slay well are rewarded with a bigger improvement to rank per win. So they’ll just continue to ignore OBJ.
Usually it’s 1 or 2 teammates that decide to play obj solely because they want a win.
KPM and DPM. Not KD.
TrueSkill2 looked at hard data for K, D, A, K-D, KD, KDA, damage, objective scores, and medals - to name a few.
It found no supporting data for ranking players up on those metrics.
Essentially what they found is that if you have a Diamond 3 player, for example, that is doing well on objective scores and you ranked them up to Diamond 4 then their team kept losing.
Whether you like it or not - the objective games are still Slayer games at heart. If you can’t match your opposition in a 1v1 you aren’t going to do well at the next level.
It’s not saying those objective metrics aren’t important. They are. Just they are inherently included in the win. It’s a team game and it’s up to the team to delegate the individual jobs.
I’m not a good player, by any means, but I’m sure the good to very good ones will tell you that sacrificing yourself for the object is not helping the team. Especially if you are doing it repeatedly. Your opponents will just use their 4v3 advantage to do their thing.
The faults in the system are 343’s communication.
There are still player’s who think that KD is important. It’s clearly not. That’s the sort of stuff that 343 could easily stop.
And we need support. Being able to look up our own MMR and KPM graphs would go a long way to making the system user friendly.
And I personally think they have made some horrible errors with the presentation of CSR.
It’s not so much adapting to “flaws”. It’s just that the general public don’t know what they are supposed to be adapting to. So they jump at shadows or fall into 343 conspiracy theories. All this could easily be fixed by better communication from 343.
Or give us the tools to see for ourselves.
1 Like
No system is infallible.
But let me know if someone develops a system that is faster and/or more accurate. I’d be on that train in a flash.
And always happy to discuss how things can be improved.
It doesn’t ignore it. It’s part of the result (win or loss).
It’s a team game. Someone has to play the objective. The better teams (and player’s with the higher ranks) share the responsibility.
The data clearly shows that out of all your performance metrics the only ones that showed extra accuracy in rank prediction are KPM and DPM. So if you want to rank up don’t just grab the oddball and die. Kill a couple of players to free up the oddball. Hold it for as long as you can - but then drop it and kill the players coming to get it. Then pick it up again.
Don’t spend your whole game respawning on some insane martyr mission to “save the team”.
If you think you should be ranked up just because you are good at holding the ball - then it’s a false hope. Because those tougher players you will face at the next rank up are just going to kill you more often.
And I get it. When I play with my sons I am the weak link. I can’t go 1v1 vs the harder opponents. So I hold the ball. A lot. The team is better suited by them getting better positions and weapons. But I don’t just give up and die. I try my best to fight back. I get a few kills and a lot of assists. A bucket load of assists.
But I know I don’t deserve to rank up on this basis. Even if we win a few games I don’t want to rank up - I’ll just get slaughtered when I go back to playing without them (my assists turn into opponents who regenerate their shields and take the ball).
1 Like
Oh excellent
For one, take out this KD requirement to rank up in objective, that’s moronic.
Second, have the game match based on rank, not invisible MMR
3rd, remove the insistence on forcing people to lose 50% of the time, its toxic
Right. I don’t think that objective should have strong influence. But I do think that in general, the objective player, when winning is ranked 4th when they win. In general, it appears to me that the highest ranked or MVP of your team sees the most progress in rank. While the 4th place player receives a much smaller gain.
Now if you were doing the lions share of the objective work. (Which we can track with offensive kills, defensive kills, obj specific medals, and zone control time) You are the reason your team won and why everyone is ranking up. So I think we could agree that there should be an objective bonus that you qualify for in the event that you win and did say 50%+ of the objective. It would have to be a small boost, but enough to give the player a reward for outstanding obj work. in ranked, especially diamond+ this would be a rare achievement. but still a nice incentive to strive for.
This is how I play BTB. None of my friends play anymore so I’m always solo. Anyways, I don’t necessarily save the team. But I do have 16 ties lol.
P.S Not sure if the ranking system does this. But on halo tracker it counts tie games as part of your losses when calculating % so my btb win % is 47% even though I have 25 more wins than L
I agree. KD would be moronic. It’s a -yoink- stat.
KPM on the other hand has some compelling evidence. Page 14 of Microsoft’s discussion paper.
If you want to take it out you will need to provide a counter argument against it.
I can’t think of one.
It’s not invisible.
It’s the best way of matching you in appropriate games.
It’s not the best way of describing your rank to the player. Volatile probability curves and such. So we have the CSR.
But having said that I think 343 could do a much better job at the CSR. And I’ve mentioned that many times.
It’s not forced. If you want to continue to believe that - that’s your problem. I’m not sure what 343 can do. Except put out an explicit statement that the algorithm doesn’t force a result.
Oh wait, they did…
Keep in mind the CSR change isn’t influenced by a player’s performance in that game. So you can’t infer who got what from KD, objective time, damage, etc. It’s all about the result, the shape of your MMR curve, the size of the upset, and the relative position of your MMR to your CSR.
The danger, as shown in the TrueSkill2 research, is that you can’t rank a player up on objective score.
They just can’t compete at the higher rank. They lose.
I’m all for rewarding the objective work. Medals are a start. Maybe showing them off in the post game screen. And of course XP bonuses if and when we finally see a proper XP rank.
But the only people who win if they are ranked up in skill is their next lot of opponents.
1 Like
same thing as kd
I’m getting tired of explaining this.
yes it is
Yes it is, they literally say it is
When they say they’re forcing a 50% win loss, they mean wins AND losses. So that means FORCED losses. How is this hard?
Forced unless you’re putting up lebron stats doing all the slaying and obj with less than 5 deaths.
Team slayer you’re especially screwed because it only takes 1 teammate going 1 and 15 to cause the L.
I don’t know how else to say this… and I’m trying to tread lightly here.
But they aren’t.
10 kills and 5 deaths = 2.0 KD.
But KPM depends on the length of the game.
10 kills in 1 minute = 10 kpm.
10 kills in 2 minutes = 5 kpm.
10 kills in 5 minutes = 2 kpm.
10 kills in 10 minutes = 1 kpm.
10 kills in 20 minutes = 0.5 kpm.
All with the same KD.
You can quite clearly see that the same KD can have markedly different KPM depending on the time of the match. And because KPM is the metric that affects your rank it’s important. Whether you want it to or not.
I’m not trying to be difficult here. I’m just trying to explain it so you can make use of the post game stats and see where you need to improve.
It’s pretty much what your CSR is. You can get a very good idea of where it is by looking at it’s behaviour with a win vs a loss. But you need to put a little bit of effort into understanding the system.
They have been quite clear. They are allowing a 50:50 win rate to fall by creating evenly matched games.
This is completely different to your theory that they have decided what they want you to lose before you even press the button to find the next game.
It’s not hard. You are just making it so.
Why do people assume that the winning team hasn’t been faced with the exact same issues? They have just come together and overcome them as a team. Their better players carried better. Their lower ranked players contributed more. Their middle ranked players held their own.
How or why is often a mystery. It’s nigh on impossible to take four random players and predict team chemistry.
And if, for whatever reason (eg. low pop or hard to match squads) that the match maker put you into a game as a genuine underdog - it’s taken into account. Your CSR may fall on the loss (as it has to) but your MMR will resist changing. And if you genuinely carried above and beyond then it may even go up.
So play on.
So what you’re saying is
the best way for me to improve my rank is to completely ignore obj and focus on slaying as many players as quickly as possible?
Definitely not.
If you read back through the thread (and other similar ones) - the primary drive for ranking up is to beat teams ranked above you.
There is a weighting though for KPM and (to a lesser extent) DPM. But it is just that - a weighting.
And it’s not about belting teams lower than you - it’s about matching teams ranked above you. You can check how you fared at the end of the game on Waypoint by comparing your actual vs expected kills and deaths.
Previously Josh would talk about people’s KPM vs levels of opponents. He would look up people’s stats and say you have a 2.2 KPM vs Platinum, 1.5KPM vs Diamond, and 0.8 KPM vs Onyx - and this is what we would expect for a Diamond 4 level player. But keep in mind I’ve made up those figures as an example - don’t take them as gospel.
But it’s pointless putting all your effort into slaying if you ignore the objective and lose. The best that will do is slow how fast your MMR falls.
What I am trying to get across to our friend is that concentrating on KD can lead you down the wrong path. Protecting a KD at the expense of maintaining a healthy (or higher) KPM is going to make it harder to rank up.
And I’ve used that against people in the past. One game was 2v4 in Oddball and the opposition was camping beside the Oddball to farm their KD. So my son and I stood back hiding, occasionally throwing a grenade or shooting in their direction. They farmed a few extra kills but the game went the full distance and their overall KPM was worse than if they had have just picked up the ball and won.
1 Like
10 kills is still 10 kills whether you got it in the first 2 minutes of the match or all through the entire 15 minutes. It’s still 10 kills for the whole match.
Evidently it’s not
Except that’s an outright lie since the matchmaker actively tweaks the quality of players in the match to cause you to lose if you do too well for too long. That 50:50 matchup doesn’t exist. You’re either on the 67% side or the 33% side.
Because most matches are predetermined, with the only variable being luck. The winning team is often playing with a stacked deck.
Cool.
But it’s a much lower KPM.
So your MMR is going to suffer.
But as long as you appreciate that it’s your choice.
Easy peasy.
If your CSR only goes up a bit with each win your MMR is lower.
If your CSR only goes down a bit with each loss your MMR is higher.
If it’s not moving much in either direction you’re probably spot on.
It’s not super hard.
And in the end it’s pointless. The only record you have of your rank to show off is your CSR. So at the end of the day the rank you can print off and stick to the fridge is your CSR.
So go Diamond!
Not true.
But more than happy to examine any evidence you have.
How does HaloTracker calculate this stat? Is it from the team MMRs (which you said where a blatant lie from 343).
Sigh.
Thank the heavens work is about to get super busy… I need the distraction.
Can confirm for a recent ranked game when I was platinum 6 after placements about a week ago.
Cobra had a 67% chance of winning
I was eagle vs a onyx 1669 diamond 4 diamond 3 and a gold 6
our team was p6-d2
we lost by 1 flag capture
Have we worked out how Halo Tracker finds the % chance of winning?
Is it a stat from the data stream?
Is it derived by them? And if so what algorithm do they use? Does it use the average team MMR or the CSR.
If it’s the CSR keep in mind you don’t know the gap to the MMR (mainly post placement when the game could be 2-3 tiers).
And 67:33 isn’t horrendous. The underdog is going to win 3 to 4 times out of 10.
For perspective - a Platinum player will outplay a Diamond player 2 to 3 times out of 10.
Argh. So close!
If you really were the underdog you could have given your MMR a nice leg up with a win! But at the very least your MMR will resist going down on the loss because the result was as expected.
1 Like
Does truskill2 account for friendly fire? For example if I’m in the hill and my teammate blows me and an enemy up, they get the credit and I get told I’m not helping. This is a legitimate question. I’m not referring to betrayals because there is a way for just enough friendly fire and not get a betrayal.
Halo Tracker has a predictive algorithm similar to 343’s one to predict odds. It is really accurate. I don’t think we’re sure if it uses MMR or CSR, but let’s face it, the only people still playing are Halo addicts and have enough games under their belt to have a close CSR/MMR.
67/33 is horrendous, considering the game could have balance teams to get closer to 50/50.
This is my new favorite picture regarding 343’s statement to the forced 50/50 winrate: i.imgur.com/SAZdtU5.jpeg