Halo Infinite Ranked.... I'm done with this game

Ken_CTF · February 3, 2022, 3:44pm

I think the main way to identify if there is a large difference in your CSR and MMR is based on how much you gain or lose for a win or a loss.

For example at my current rank, I generally lose 1-2 CSR per game but gain 10 CSR per win. This tells me my MMR is higher than my CSR.

If you are getting similar gains and losses both ways then your CSR and MMR are close together and you are likely at your accurate rank right now.

If you are losing lots of CSR for losses and gaining little CSR for wins then your MMR is likely a bit lower than your current CSR.

SIPIRITIJ · February 3, 2022, 3:45pm

I’m going to copy the whole thing here so we can comment:

I labored through 40ish games on a test account, playing in a very particular way to test the system. For reference, I’m an ok player, nothing special, but good enough to notice what I’m facing. But no need to take my word for it, we have full breakdowns from the Halo Infinite MM API, organized amazingly courtesy of …

Let’s get started.

Hypothesis: there is a hidden MMR rank that affects all of your games, social and ranked, and therefore it should be possible to tank your MMR in social games to give yourself easier ranked games.(Side hypothesis: this effect occurs even after placement matches.)

Method: I started a brand new fresh account, and did the following:

Go into 10 quick play games and lose all of them terribly, going 0-10 or worse.

Play Ranked and actually try, raise get placed.

After placements, play a few more Ranked games to see where the system is putting me.

Now go back to social, and tank another 10 games with abysmal performances.

Finally, go back into Ranked and see what happens.

AS YOU CAN SEE HE’S ACTUALLY TRYING

For reference, here is my test account. Every match you can see the average CSR (the gold/plat/diamond/onyx rank number) and MMR, but these are only team averages, as the API available does not provide individual hidden MMRs.

https://leafapp.co/player/II+Ryot+II/matches

Phase 1: Tank 10 social games Not much to say here. I had to waste about 2 hrs to do this properly, I wanted to really make sure that I ran out and died. My teammates and opponents progressively got worse and worse, but on average my teammates got better than my opponents. In fact, it got lopsided to the point where around game 7 or so it became hard to lose, because even if I went 0-20 in oddball, the system gave me good enough teammates to 3v4.

AS I POINTED OUT, BECAUSE A NON 50-50 W/L RATE IS UNACCEPTABLE

Phase 2: Play my 10 placement matches Oh boy, this was crazy. My very first ranked game, I went 45-1, and that one death I’m pretty sure was from a team nade. I got medals I’ve never seen in my life.https://leafapp.co/game/c3df0b9e-9748-4ce6-b88e-51b320ed6843

But what actually matters over these 10 games is the following: the average team MMRs, both mine and opponents’, kept going up. The games also got harder, and my K/d got worse and worse. Here’s the sequence of 10 games summarized:

Game 1: my team MMR = 293, opponent team MMR = 298, my K/D = 45:1
Game 2: my team MMR = 407, opponent team MMR = 411, my K/D = 8.67:1
Game 3: my team MMR = 566, opponent team MMR = 623, my K/D = 8:1
Game 4: my team MMR = 545, opponent team MMR = 552, my K/D = 11:1
Game 5: my team MMR = 430, opponent team MMR = 430, my K/D = 9:1
Game 6: my team MMR = 565, opponent team MMR = 593, my K/D = 5.5:1
Game 7: my team MMR = 530, opponent team MMR = 522, my K/D = 12:1
Game 8: my team MMR = 784, opponent team MMR = 782, my K/D = 13:1
Game 9: my team MMR = 891, opponent team MMR = 909, my K/D = 16:1
Game 10: my team MMR = 1107, opponent team MMR = 1301, my K/D = 3.57:1

You get the idea, as the MMR goes up it gets marginally harder because opponents get better to the point where they can catch me making mistakes.

MMR RISES AS HE WINS AS IT SHOULD

Phase 3: Post Placement Matches - and I got placed Platinum 3

Alright here’s where it gets interesting. The CSR I got at Plat3 is around 1000-something. But the MMRs I played my next few games in were really tough! In fact, at Plat 3, I lost the next 3 of 4 games. I still slayed really well, but the system gave me teammates that were much worse than the opponents. But guess what? Because my losses were all big slay games, I lost no CSR. Here’s how the next 4 games went:

Game 1: my team MMR = 1304, opponent team MMR = 1245, my K/D = 2.73:1 (loss, -0 CSR.)
Game 2: my team MMR = 1225, opponent team MMR = 1151, my K/D = 3.67:1 (win, +15 CSR)
Game 3: my team MMR = 1320, opponent team MMR = 1368, my K/D = 2:1 (loss, -0 CSR)
Game 4: my team MMR = 1405, opponent team MMR = 1411, my K/D = 2.32:1 (loss, -0 CSR)

THE SYSTEM PLACES HIM IN HIGHER RANKS MATCHES BECAUSE IT THINKS IT BELONGS THERE EVEN THOUGH THIS IS NOT TRUE

See what’s happening here? My individual performance has the system constantly putting me on and against higher MMR teams (meaning my hidden MMR is definitely rising). What we also see: as a Plat 3 I’m playing more Diamonds than Plats, and in every case I’m top frag by quite a bit. My losses lose me nothing, my wins give me huge gains.

Phase 4: Tank more social games. Let’s see if this works after placement. I lost about a dozen or so fiesta games. Or rather, I tried to lose them, eventually the system started giving me really good teammates who could carry me going 0-16. Sometimes it was a fun race trying to lose faster than they could win. We won a 50-49 and I was pissed. Lol, sorry for ruining a bunch of ppl’s MM experiences lately, but it’s for science.

AGAIN AS I POINTED OUT, BECAUSE A NON 50-50 W/L RATE IS UNACCEPTABLE

Phase 5: Let’s go back to ranked and see what happens.

Game 1: my team MMR = 1050, opponent team MMR = 1034, my K/D = 18.5:1 (win, +13 CSR)
Game 2: my team MMR = 1218, opponent team MMR = 1199, my K/D = 9.5:1 (win, +14 CSR)

Wow, and there you have it. I had a ranked and placed account, I then threw a dozen social games, and I’m playing in lobbies 400 MMR lower where I’m fragging like I’m playing bots. Interesting that the next game quickly jumped up the average MMRs involved after I dropped a 33:2 performance lol.

WHY? SEEMS LIKE A GOOD THING THAT IT ADJUSTS QUICKER IF YOU PERFORM BETTER

We can’t figure out the exact numbers involved here though. MMR seems to move down more slowly, and re-adjust to your good performances quickly. For example, throwing those 12 or so fiesta games moved me down from 1400ish lobbies to 1000ish lobbies (400 pts over 12 games). But as we saw in Phase 5, I quickly went up 150 MMR pts on average after just 1 high performance.

12 GAMES IS A LOT BUT NONETHELESS IT’S POSSBILE

Results and Analysis:

Yes, this is broken. You can sandbag in socials to get easier ranked. How much easier? It’s hard to say. If I’m a 1600 Onyx on my main, can I use this method to get to 2000 Onyx (CSR)? Maybe with a lot of grinding, maybe not. I don’t really care personally. But there’s 2 interesting implications here:

From a game theory perspective, if you want the highest CSR possible (e.g. rank, Diamond, Onyx, etc) then you want to use this sandbag method, and you want to avoid slaying hard in bot matches. You want to be better than the average MMR games you’re playing, that way, if you win, you win big (because you’ll probably be top frag), and if you lose, you’ll lose very small as top frag. Also, if you throw in social, because of the team MMR matching, chances are you’ll start getting better teammates (relative to your opponents) because the system wants to give you help.

The other interesting idea is that if you want to play against pros in matchmaking, go slay a bunch of bots on a new account then get into ranked. Your 2000+ MMR might get you matched up against Optic or C9 players Well, the system probably has some breaks in place to not be that extreme, but I’ve definitely seen unranked accounts who have tested this bot method match against the likes of Cratos, Swish, and other pros.

Conclusion:

Matchmaking by 343 is really stupid. Tons of players who over-played casual games will be having a lower CSR than they should in ranked, and a terrible experience getting dunked on. This linkage of social and ranked global MMR means that now we know for a fact that sandbagging works.

If you’re reading this particular reddit, I’d recommend just not giving a s* and playing your games to improve. Maybe try shooting a bunch of bots with 20:1 K/Ds if you want a higher chance of placing into pro or semi-pro matchmaking lobbies lol.

Anyway higher MMR is always better but to lower it means getting easier matches and consequently better K/D. It also means by the tests that if you lose you don’t get punished as the underdog and if you win you win a lot.

Ken_CTF · February 3, 2022, 3:50pm

Not for the first ten games, so his final placement in ranked is inaccurate.

No you will get punished. You are punished more for having a lower MMR. It’s true you are punished less if you lose to a higher rated MMR team, but only if you yourself are not a higher CSR than your MMR.

I cannot stress enough how idiotic the idea is to lower your own MMR in a pursuit for a higher rank. It’s genuinely insane. It’s even worse that someone is spreading this misinformation because it ruins games for other people and players will no doubt try this after reading this widely upvoted post. If it worked I’d at least have to conclude the system needs to be changed, but it literally does the opposite of what people are wanting to do it for.

SIPIRITIJ · February 3, 2022, 3:53pm

Not for the first ten games, so his final placement in ranked is inaccurate.

Why should it be inaccurate? He only played social games. This makes placement matches useless.

No you will get punished. You are punished more for having a lower MMR. It’s true you are punished less if you lose to a higher rated MMR team, but only if you yourself are not a higher CSR than your MMR.

Which is exactly what happens because the system always places him against higher MMR.

I cannot stress enough how idiotic the idea is to lower your own MMR in a pursuit for a higher rank. It’s genuinely insane. It’s even worse that someone is spreading this misinformation because it ruins games for other people and players will no doubt try this after reading this widely upvoted post. If it worked I’d at least have to conclude the system needs to be changed, but it literally does the opposite of what you want it to do.

Stoatbag · February 3, 2022, 3:55pm

To be fair, when I win and play well (usually against Onyx/Diamond players) my rank does tend to jump quite massively, and drops slightly when I lose against them. But it seems inconsistent. For example I had a CTF match against 2 Onyx’s, 15 kills 15 deaths, 1 flag cap and 1 cap assist. My team won 3-0, and my rank didn’t move at all. To me, if a platinum wins with a positive kdr and solid objective performance against a team of Onyx’s and Diamonds the CSR should sky rocket, but it didn’t that time. Other times it does. Maybe over time it will balance out, but the journey is so miserable and makes you feel like an awful player. Don’t understand why they need two different skill measures. Don’t understand why having one doesn’t work

Ken_CTF · February 3, 2022, 3:56pm

Okay here’s the scenario:

Your CSR is 1,500.
Your MMR is 1,550.
You throw buckets of games until system believes you are 1,400.

Your next games are against 1,400 ranked players. Wins are worth next to nothing, losses are huge because the system is keen for your CSR of 1,500 to be closer to your MMR of 1,400.

Had you not thrown games wins would be worth more than losses to help you get to 1,550.

Ken_CTF · February 3, 2022, 4:00pm

Because otherwise there would be no grind I guess. You would just automatically be your true rank. Plus it fluctuates wildly at first so that would be confusing for players. Also, MMR can vary for a number of reasons whereas CSR cannot. Also your MMR goes up or down in games win or lose, whereas at least with CSR you know you will always get some for a win, always lose some for a loss. MMR would take the shine out of it a little, but is more accurate for balancing teams.

I’d say don’t focus on the individual games, it’s the bigger picture, but also the Onyx players may have had worse teammates to balance so even though you did well the teams were still technically very even

SIPIRITIJ · February 3, 2022, 4:06pm

Okay here’s the scenario:

Your CSR is 1,500.
Your MMR is 1,550.
You throw buckets of games until system believes you are 1,400.

Your next games are against 1,400 ranked players. Wins are worth next to nothing, losses are huge because the system is keen for your CSR of 1,500 to be closer to your MMR of 1,400.

Had you not thrown games wins would be worth more than losses to help you get to 1,550.

That doesn’t happen because:
First the difference in MMR before and after throwing matches is wider than 100 (400!!!)
Second: He’s not playing against an equal opponent he’s playing against higher MMRs. Again, why?

It should be stressed that what people perceive is more important than the math behind it and this ranking system makes people feel frustrated and doesn’t give you any sense of progression. If, as I’m pointing out for days the “jumps” between the steps were closer and the system didn’t force you to be destroyed at some point,
scenarios like those in the test where much better players are placed with people who don’t play at all just to make it 50-50 or “harder to lose” wouldn’t happen and we would actually improve our skill instead of bouncing around between too easy and too hard like a ping pong ball without learning anything about the game.

Stoatbag · February 3, 2022, 4:07pm

That’s the bit that I don’t think quite works right now. The team were even in that game. They had two Onyx’s and so did we. But it doesn’t seem to reward me based on my personal chance of doing well in a game. In that game I was the cannon fodder, or at least that is what you would expect. I contributed well but the system said, the match was even and therefore you deserve nothing from it. But from a personal perspective I pulled it out of the bag doing ok against Onyx players. That’s the bit I think needs refined

SIPIRITIJ · February 3, 2022, 4:08pm

Because otherwise there would be no grind I guess. You would just automatically be your true rank. Plus it fluctuates wildly at first so that would be confusing for players. Also, MMR can vary for a number of reasons whereas CSR cannot. Also your MMR goes up or down in games win or lose, whereas at least with CSR you know you will always get some for a win, always lose some for a loss. MMR would take the shine out of it a little, but is more accurate for balancing teams.

I’d say don’t focus on the individual games, it’s the bigger picture, but also the Onyx players may have had worse teammates to balance so even though you did well the teams were still technically very even

“technically” is exactly the problem we need to work with “actually”

SIPIRITIJ · February 3, 2022, 4:09pm

Amen.

Here’s your useless 20 characters.

Ken_CTF · February 3, 2022, 4:17pm

I think that would be more because it’s one individual game and not all games, it’s a progressive thing. Also could be that your team was expected to win perhaps? If you have the game on Halotracker to hand I’ll take a quick peak although usually one game is not enough to say anything about. Typically though if you were predicted to win then that will be the largest indicator of why there was little CSR gain as you are expected to perform better for that reason alone.

Stoatbag · February 3, 2022, 4:22pm

https://halotracker.com/halo-infinite/match/6661a682-aa33-4513-bc60-1321e6fba001

Hopefully that link works

Ken_CTF · February 3, 2022, 4:28pm

In 5 losses, you lost 6 CSR including 50/50 games.

The good news is you are heading straight through Diamond 1 and beyond my friend, as long as you keep playing at your current level.yeah your MMR is definitely higher than your current rank in both playlists. You just need a few wins here and there, keep playing.

Stoatbag · February 3, 2022, 4:45pm

Good to hear, hope it works out that way. I think for me it is a psychological problem that this system causes. It pits you against people at the absolute top end of your skill set, seemingly particularly in platinum and early diamond. That inevitably causes frustration, because you feel like you are rubbish at the game. It hopefully will get me to the right place in the end, but I’d much prefer a system where it felt like progression. You play against similar ranks; you win you move up, you lose you move down. That way you want to keep playing, and then you’ll hit a point where you can’t go any higher and that is your ceiling: That’s the issue to me. It feels like infinite skips past that progression phase where you feel like you are good and improving and just sticks your matchmaking at where it thinks your end goal would be. I.e right at the top end of your skill set, right in day 1. So the feeing of progression is lost, for everyone but the very best players

Ken_CTF · February 3, 2022, 4:51pm

Yeah essentially your opponents never change, just your rank does. If they do a season reset the same will be true. Even though we will all be Diamond 1, Onyx 2,000 players will be playing Onyx 2,000 players, not Diamond 1’s.

It’s definitely a psychological thing though and frustrating if you are losing a lot but due for a climb. The game is never easy, you’re always having to play strong players. But personally I do prefer it this way than stomping people in uneven games.

SoulxFigmentk · February 13, 2022, 8:46am

I hate this! Why are we penalized for having quitters on our teams?