Heart Broken Over Extremely Poor Calibration System for Seasonal Rankings

[brownmagic]

I've been playing DOTA 2 since its Beta.

My current Solo Que MMR: 4,024
My current seasonal MMR (after winning 9 out of 10 calibrating matches): 3,676

I'm not sure what to make of this other than feel extremely heart broken. When I review the win/loss streaks of other players with much higher seasonal MMR, I'm given the impression that I was cheated. With the assumption that they played their 10 calibrating matches back-to-back, I would have had equal if not better stats to all of them. Many of the friends on my friends list can confirm that they didn't win all their calibration matches, but they still received MMR (in some cases) 300 points higher than mine.

I actually don't mind if my MMR is rightfully 3,676 after callibration. I'd just like to understand how or why. I honestly feel cheated here in seasonal especially after playing so well. My games were not easy and I had to do a lot of carrying every so often (as well as be carried). I would think that seasonal would be an opportunity to reset the scale for players who feel lost in their effort at building their solo que MMR.

At least with the current ranked system, the community has set a general bar of 5k+ to be the "top tier" MMR. Can someone please help me understand my seasonal ranking and help set a proper bar for calibration results?

[GermanViet]

Considering it is not possible to calibrate above 5k MMR I think it's normal.

[brownmagic]

Ok, so not being possible to calibrate over 5k, I respect that. But why 3676 MMR? So my 1 loss (-25 MMR traditionally) was equal to 52 victories (1350 MMR)?

Say 5000 was the cap, that means everyone would start at about 2500 MMR while calibrating. That would mean each victory should be worth around 250 MMR. Let's also say that timing matters: first come are served to the higher end of the tier. So the first 250 players to win all 10 calibration matches get ranked between 5000 and 4750 MMR or less, the next 250 with a 9-1 record are ranked between 4750 MMR and 4500 MMR, or just after anyone who wins all his matches.

My calibration ended at 12am PST on May 19th and the compendium came out the 16th at 12am PST I believe. To only receive a score of 3676 MMR, according to the logic above, there would've had to have been 2500 other players that scored the same score as I had before me in that 3-day time span and over 250,000 players who completed their calibrations within that 3 day period (statistically speaking). Not impossible, but how someone else scored 3911 MMR with only 7 out of 10 wins in his calibration match (confirmed record)...that really bothers me...

Instead of helping players measure their skills and set appropriate bars for improvement, an unexplained and seemingly random ranking system sets false bars for everyone in the community and buries talent. It gives a sense of a "no way out/no way to improve" situation that diminishes the value in DOTA 2 as an esport.

Though I've won 9 calibration matches, I wonder if I was given matches competitive enough to measure my skills over those 10 matches? In every match, both teams had an average of 3000 MMR going in. Does this mean near the end of my calibration matches, I was matched with 2k players and expected to carry as a 4k player, or does it mean I only played with 3k players the entire time and that my MMR range was predetermined to begin with...I'd like to believe I was given an opportunity to play and calibrate with the best, but now I'm not sure...

Again, saying this as a devoted fan...this is such a heart breaking development.

[testuser8]

You do not understand, there is barely anyone in unranked with 5k MMR. It's a nonserious gamemode with everyone at 50% winratio.

Most people got to 5k-7k through grinding Ranked queue, which means your unranked MMR is unaffected.
The recalibration works this way:

Your UNRANKED MMR + (-/+500) over 10 games. So you move between 1k mmr beftween your intial MMR with max 500 MMR climb.

You having lower unranked is because you have been losing unranked games or queued with lower MMR players for some time and gotten worse opponents and that affects you when you recalibrate, this is the same system that is used when you calibrate for ranked. You CANNOT ever EVER get a 3k account and calibrate normally to 5k+. Your UNRANKED has to be high, hence why some Tinker players abused the system with unranked and got some extreme MMR after calibration.

Valve did nothing wrong here, you just believed this was a hard reset of your MMR like in CSGO when you first calibrate, but if you ever played dota 2 you shouldve known this was not the case.

[BlinkingRiki182]

You do not understand, there is barely anyone in unranked with 5k MMR. It's a nonserious gamemode with everyone at 50% winratio.

Most people got to 5k-7k through grinding Ranked queue, which means your unranked MMR is unaffected.
The recalibration works this way:

Your UNRANKED MMR + (-/+500) over 10 games. So you move between 1k mmr beftween your intial MMR with max 500 MMR climb.

You having lower unranked is because you have been losing unranked games or queued with lower MMR players for some time and gotten worse opponents and that affects you when you recalibrate, this is the same system that is used when you calibrate for ranked. You CANNOT ever EVER get a 3k account and calibrate normally to 5k+. Your UNRANKED has to be high, hence why some Tinker players abused the system with unranked and got some extreme MMR after calibration.

Valve did nothing wrong here, you just believed this was a hard reset of your MMR like in CSGO when you first calibrate, but if you ever played dota 2 you shouldve known this was not the case.

I call bullshit.

The truth is that you need to win the calibration games with high K/D/A and possibly deal a lot of damage (or another hidden metric).

[Sinix]

I call bullshit.

The truth is that you need to win the calibration games with high K/D/A and possibly deal a lot of damage (or another hidden metric).

You're overrating the importance of the calibration phase for ranked (the 10 games). The ~100 normal games you play before going into ranked have a way higher importance.

[NoneOtherLIKE]

You're overrating the importance of the calibration phase for ranked (the 10 games). The ~100 normal games you play before going into ranked have a way higher importance.

Truth

[innociv]

I won 6-4 and calibrated 3649. And I was 2-2 on games that had an average of over 4,000 MMR.

My friend won 4-6 and calibrated 3699.
I had far higher KDA, GPM, tower damage, etc than him, but calibrated lower with more wins.

The recalibration seems to basically do nothing and be a joke except that it lowers your MMR the more you were over like 4k. It's not a real recalibration as it's mostly based on your previous MMR, or unranked MMR. If it really worked, it probably should have put me over 4k since I went 2-2 on those matches and 4-2 on the ~3.7k ones. And one that I lost was just because someone fed on purpose over 20 times in 20 minutes. Rofl.

There may as well been no calibration matches as it's so heavily based on normal mmr and not real calibration. The mmr is just hidden for Valve to try to trick you into thinking it's not rigged and a joke when it is. It's your normal MMR which mostly affects it even though you never tried in normals.

[GermanViet]

I won 6-4 and calibrated 3649. And I was 2-2 on games that had an average of over 4,000 MMR.

My friend won 4-6 and calibrated 3699.
I had far higher KDA, GPM, tower damage, etc than him, but calibrated lower with more wins.

The recalibration seems to basically do nothing and be a joke except that it lowers your MMR the more you were over like 4k. It's not a real recalibration as it's mostly based on your previous MMR, or unranked MMR. If it really worked, it probably should have put me over 4k since I went 2-2 on those matches and 4-2 on the ~3.7k ones. And one that I lost was just because someone fed on purpose over 20 times in 20 minutes. Rofl.

There may as well been no calibration matches as it's so heavily based on normal mmr and not real calibration. The mmr is just hidden for Valve to try to trick you into thinking it's not rigged and a joke when it is. It's your normal MMR which mostly affects it even though you never tried in normals.

You're overrating the importance of the calibration phase for ranked (the 10 games). The ~100 normal games you play before going into ranked have a way higher importance.

.

[innociv]

Thanks, that's what I said.

[testuser8]

I call bullshit.

The truth is that you need to win the calibration games with high K/D/A and possibly deal a lot of damage (or another hidden metric).

You need to do that in UNRANKED first. Read my post again. You can go 100 kills in your first 10 ranked matches but if your unranked is 1k, you will max get 1.5k MMR after calibration.

[innociv]

Which is pretty stupid, when unranked is "practice" games. Calibration should have more of a spread away from your unranked mmr.

I win over 90% of my unranked games. They're stupidly easy and not representative.

I don't care too much though. If you calibrate at 3.5k and are good you should be able to get to 4.5k or 5.5k or whatever without too much trouble. There's bigger problems than that Valve refuses to address.

[medievil1903]

I win 8 out of 10 but i got 2.840 . It's so strange how do they calibrate who knows?

[testuser8]

I win 8 out of 10 but i got 2.840 . It's so strange how do they calibrate who knows?

Maybe read the thread instead of complaining. This isn't reddit.