Some Hard(ish) Data since Season 2 Reset

Ok. I love cold hard data. So much better than the typical observationally biased anecdotal stuff that leads to weird conspiracy theories that 343 hates you (and only you).

So closing in on 150 games in Ranked 4v4 since the Season reset. Not as many as I would have liked - but work, interstate trips, and generally being frustrated by the insane pings I’m getting.

Anyway. For those interested in such things…

CTF = 17.0%, KOTH = 26.7%, OB = 17%, SH =17.8%, and SLY = 21.5%.

So, apart from KOTH, looks pretty even. Probably fairly safe to say it’s random. Or there may still be a bit of a weighting for KOTH given it’s the newest mode.

Aquarius 14.8%, Bazaar 5.2%, Catalyst 8.1%, Live Fire 22%, Recharge 24.4%, Streets 25%.

On first glance that looks to be a pretty fair distribution on the basis of the game modes available for each map; Aquarius 2, Bazaar, 1, Catalyst 2, Live Fire 3, Recharge 4, and Streets 4.

So it looks like the game mode is chosen first, randomly, and then a map. So the frequency of the maps is based on the number of game modes available?

The games so far have been very even based on average team MMR. The average gap is just 1.14 points.

My average ping since the reset has been disastrous. A woeful 163ms. I’ve had days where I’ve averaged well into the 200’s :frowning:

The most interesting observation is based on my expected kills for each game. By dividing expected by the time of the match I can look at the KPM (and DPM) the system is expecting.

And what I see is clearly different KPM for each game mode.

CTF = 1.23, KOTH = 1.55, OB = 1.38, SH =1.35, and SLY = 1.30

What this suggests is we have the same ranking set up as we did in Halo 5. Each different game mode has it’s own KPM, and thus it’s own MMR. And your overall playlist MMR is an average of these (which is what your CSR is set against).

So, if you identify your weakest game mode (mine is Oddball) - this is the best value for money to give your rank a little boost.

Miscellaneous;

Just under 1 in 10 games have been affected by a quit (for or against). Too many.

I made a special note to look at any time I had a winning streak of two games or more - was the first loss unavoidable (ie. rigged). Well, I can honestly say in the 16 such times (yeah, I know - not many), the loss felt avoidable. The best I could come up with of sabotage was a D3 “team mate” who went double figure minus (that hurt).

8 Likes

You playing solo?
Im interested in knowing the percentages of full teams vs partial teams or all randoms.

2 Likes

Came here for the same thing @SafetyM33ting.
@Darwi I’ve kept browsing the forums for a while now and I respect your views on SBMM, but there are definitely some points that I don’t fully agree with you/343 on. Almost ALL of those points center around squads of friends.
SBMM does a fairly decent job IF YOURE SOLO (other than if it cant find a full game of individuals with an acceptable MMR deviation and places you in a game where each team has a deviation of 300 MMR or more between teammates).
When you’re on a squad of friends that MMR deviation is ALWAYS an issue for my friends. Even in non-ranked game modes, the lowest ranked friend gets obliterated every time. I may be the exception, not the rule, but in social, I try to play down to my opponent. I’ll put restrictions on my play (letting my other teammates/opponents grab rockets etc). It still feels like my ‘social’ MMR is too heavily weighted by my ‘ranked’ MMR so that each are affected by each other.
Trying not to turn this into a rant, but I really enjoy the statistics side of it all and would love to know the nitty gritty of how and why 343 uses the SBMM settings they do (past just the basics of how Trueskill2 works).

1 Like

I believe you’re right on this one when it comes to the selection of the map and game mode.

I know that I rarely ever now search ranked solo because I’ve got to have at least one more person communicating. Is this similar for you too or do you search solo?

Great post. Thank you for the informed data on your experience of matchmaking.

3 Likes

Your data is accurate based on how TrueSkill2 and MMR were explained to me.

This 5.2% is too high. It should be 0% since Bazaar is absolute garbage.

2 Likes

Bazaar is actually one of my favourites. :smiley:

Yep. Solo. If I do the same thing again for the Winter Update (I probably will) I’ll look into that.

I’m not a great communicator. Happy to listen and follow. I think I need to learn to be more proactive.

1 Like

This is always tricky.

If the opposition was low enough for your friend to compete - you would be unstoppable.

What we really need is some sort of handicapped mode - where players of all ranks can come together and play on an even playing field. My first script when Forge comes out will be a Slayer game where;

  1. Scores are based on the K/D of who you killed.
  2. If you go on a killing streak of 3 or more you lose your Shields.
  3. On respawn, if your K/D is less than 1.0, you get OS (for up to 90 seconds).

My group ranges from Gold to Onyx. So we’ll see how that goes.

But I don’t think the current system will ever work for providing a balanced experience for everyone on both teams. It’s why they have started putting restrictions in place for ranked.

Keep in mind your MMR can fluctuate from your CSR.

And there is a weighting on the MMR for squads.

You can’t really make a lot of sense of the team balance from your CSR.

Me too. I don’t get how people go into Social playing the same sweaty game style they do in ranked - and wonder why their games end up sweaty.

If your general play is closer to your level - then a few games with your mates isn’t going to do a lot. Your MMR will think you are Smurfing.

You can definitely have a much lower MMR in Social. But you have to be very consistent.

It would be very helpful if we could actually look up our MMRs. Get graphs of MMR over time. Like the ones Josh Menke used to produce for players here on Waypoint.

That, and KPM vs Opponent MMR.

Maybe one day…

This post was flagged by the community and is temporarily hidden.

Josh Menke actually described this in the Halo 5 forums. But at the time it just kind of went through to the keeper.

It makes perfect sense. Especially in terms of weighting personal performance.

And for those who need to rank up a bit - all you need to do is identify what is your worst performed mode - and that’s your best bang for buck on getting your average MMR up.

And when I think of all those people who used brag about how they quit out of certain game modes… :smiley:

Absolutely Agree.

I like the idea of your game mode. My friends too range from Gold-Onyx.

Yep, I’ve actually had the “Average Team MMR” pop up at the end game stats on xbox quite a few times. It’s been a few months since I’ve looked for it though. When I first told my friends about seeing it, they thought I was bullshitting them. Maybe it was a bug? But I’ve definitely seen it quite a few times. But having access to that kind of info would be priceless! (But I completely understand how it could be used to abuse the system)

Unless the MMR calculations for Ranked and Social are COMPLETELY independent, I don’t see how you could get a lower MMR in Social without affecting your CSR gains in Ranked or vice versa. And ever since the infamous Bot Slayer incident, I don’t have much confidence that MMRs from each playlist don’t affect your Ranked MMR.

I’ve been down to playing once a week, and it’s been rough. I can feel my skill slowly fading, but it’s like I’ve played enough games in the beginning of Infinite that my MMR will not fade as fast as it should, so every Tuesday I have a poor experience on Halo with friends and end up hopping off before I even get my challenges done.

This past week was even worse because a random teammate (in a SOCIAL GAME nonetheless) decided to message me and a friend after the game to let us know how horrible we played. He then decided to stalk my HaloTracker stats to let me know that I never won another match that night…and then belittled my next game of choice after having enough with Halo.

I really hope 343 can pull off some magic here to regain some population, because I hate how toxic and bitter the few of us that remain have become…

It’s there at the end of each game.

Under each team on the “Team Line Up” page.

You have a global MMR that contains all your recent games. It has to do this to reduce the data footprint of the system.

Each playlist is then offset from this. But each offset is a MMR curve with mu and sigma - so it behaves like a proper MMR.

So they are not completely independent.

If your MMR for Social is 900 and your MMR for Ranked is 1200 and you raise your global MMR by 25 then your Social moves up to 925 an your Ranked up to 1225.

Think of it as a reflection of “current form”.

Completely different scenario.

This was only brand new accounts who had only played Bots.

When you finally went into Ranked it “borrowed” the only MMR it had to seed your placement. Which was your 2000+ bot MMR.

This behaviour doesn’t happen on normal accounts.

And besides, 343 have since lowered the MMR of the bots significantly. So you can’t do it now even if you tried.

Your MMR is actually set to fade with time away. So it should be mitigated to some extent.

It should fade pretty fast. One or two games is enough to drop your MMR.

:frowning:

But isn’t that flawed logic to assume that everyone will play at the same general skill level regardless of playlist?
For example…If I play a few really good games of Ranked and play above my MMR, then the next night, hop into Fiesta with friends, my ‘global’ MMR will have been elevated from my previous night, leaving me and my friends to play matches of overly-competitive fiesta…let alone the fact that Trueskill assumes that our party is playing at an even higher MMR because we’re partied, but half of us know callouts and the others are trying to figure out which way is up :laughing:

Love to hear Catalyst and CTF are rare for you, looks like I get all the games you don’t, haha :joy: :rofl: :joy: :smiling_face_with_tear: :disappointed_relieved: :sob:

Oof. It HAS been really bad. Not that bad for me but definitely BAD.

Agreed. Ranked quitting needs to be punished by an automatic loss of the maximum amount you would have lost for your bar. That said, before that happens, there needs to be an input tracker every time you open the menu. If the option for leaving the game is selected or ALT+F4 was pressed, it needs to track that as well. Granted, it can’t exactly track ALT+F4 so that’s one that might slip through the cracks… and I vaguely recall legal issues around tracking what inputs are pressed. So who knows.

Regardless, quitters need to be penalized, but we definitely need to protect crashers or folks who lose connection.

AFKers need to be punished too. Mid loss on bar, auto-kick after 1min idle. Because… what are you doing that you can’t set aside a couple of seconds in ranked play? Don’t play ranked if you can’t dedicate the time. You know?

Hate getting AFKers.

2 Likes

I would think it’s pretty safe to assume that good form in one playlist would carry through to a different one.

It’s the logical assumption.

Again. The assumption that you will play a bit better is sound.

And you are only elevating your share of the team average.

If you are dooming the team that’s one hell of an individual form streak. :wink:

And besides. Even if you do get whipped in the first game… All your MMRs will drop quickly. Together.

Word.

It’s pretty much how my parties tend to roll as well. :slight_smile:

But I believe your squad weighting is also team dependant. Too a small degree.

Just joining the boys in social sends them into doom. 2 of us are high diamond/low onyx, and the other 2 are somewhere in Plat. Our MMR is so spread that they struggle to get kills against any of our opponents. They have a better experience playing solo (or as a duo) and it’s pretty much killed our squad’s vibe.

1 Like

Sucks. But I’m not sure what the alternative is?

Apart from sort of handicapping play list as I mentioned earlier.

At least you would work with that. Other Onyx level players I’ve suggested this too were vehemently opposed to having their “fun” (ie. smashing noobs) curtailed.

Some good info here

I wonder if there’s some subjective component to it though. I feel like the majority of my matches are oddball. The map distribution feels right though

1 Like

I’ll try just about any game mode. If the squad’s having fun, I’m having fun.

1 Like