Unveiling the Truth: A Critical Look at Dota Plus Lab's Match Evaluation

[DO2N.GAF]

Have you tried out the Dota Plus Lab feature for match evaluation? Well, let me share my experience with it – it's not exactly what you might expect.

First off, the language it uses is overly positive, which can be misleading. Terms like 'great,' 'perfect,' and 'good' are thrown around, but oddly enough, 'fair' is marked in yellow. Now, wait a minute – does that mean fair matches aren't good? It's a bit confusing, isn't it?

Even when it supposedly predicts the quality of a match beforehand, it often falls short. I've been in matches labeled as 'perfect' or 'green' only to encounter griefers, toxic players, or smurfs. So much for the system's accuracy.

Instead of relying on vague words, why not use visual indicators like bars or circles to represent match quality? Frankly, it feels like the matchmaking system is broken if it consistently fails to deliver on its promised evaluations.

Sometimes, it claims the skill range is close, yet there are blatant smurfs wreaking havoc in the game. It makes you wonder what metrics they're using to assess matches.

So, if you're thinking of getting Dota Plus to help you avoid bad matches, think again. It seems like Valve might be out of touch with the reality of matchmaking, especially in lower ranks. I've got a friend in the Legend tier who echoes the same sentiment.

[Absche]

I think it only shows the bracket width for the given matchmaking criteria while not changing them in any aspect. Just to give the players an idea of the balance between expected queue time and expected variations in mm factors.

[﷽﷽﷽﷽﷽﷽﷽﷽﷽﷽]

Ignore the dota 2 defender

Exactly
The dota 2 match evaluation uses subjective language based for technical numerical based term

What is
"great"
"perfect"
"good"
in numerical term

This can be highly subjective and can be manipulated based on various factors that were not disclosed in the announcement but I've found out more after few testing

These factors are based on the your current qued time factor aka based from your matchmaking pool not entirely objective this is dependent for the dota 2 player account what might be good for one set of players will not be good for another set of players which means most players will not get the same equal and even experience an clear flawed logic and method

This is supposed to be highly objective in technical implementation but its not in reality

This is why your own comment of

Originally posted by ÆQUANIMITAS:

"Even when it supposedly predicts the quality of a match beforehand, it often falls short. I've been in matches labeled as 'perfect' or 'green' only to encounter griefers, toxic players, or smurfs. So much for the system's accuracy."

What was supposed to be "perfect" or green good games lead to griefers, toxic players and smurf because the accuracy is based on various factors dependent on
Time of que
Player account type and stats
% variances of the match stats

I can tell you even more the variances differences are legit bad reminds me of the forced 50% matchmaking algorithm its even worse than this
In forced 50% variance difference is 5-10%
For this variance difference are way to big
5-10% is considered "perfect" in dota 2 dev brain
when even in other system of matchmaking that variance has huge problem as this can easily determine the outcome of 90-99% of the games so

10-20% too big difference is labelled as being "good" is bad

over 20% is considered "fair" in dota 2 dev brain

See the problem


They should use actual numbers based on statistics to determine this
CSGO used to do semi do this when you used to que for example current players queing for match, in match and player pool que for your match but they won't show that to players because this would show the true flaw of this game that's why the bad devs from valve won't do it when most multi players games do it

This won't take up server time at all so no excuse for dota 2 devs not to do it

It was just PR announcement as usual nothing actual good for change if good intentions with bad methods lead to bad outcome then its bad in overall sense somehow defenders can't see that

That's not the bad part only bro since they changed how ban is made with prequed bans from players account
The smurfs aka the smartest player population are gaming the system by playing their worst heros with high play rate and winrate then switch to their best hero immediately like I've said previously before its like these dota 2 devs didn't think that this change alone makes their flawed match evaluation completely useless to null nothing even the overwolf dota plus cant help with this

[Jack Frost ❄]

may matchmaking evaluation wont show, why is that.
even if its turn on. it is a bug or what

[DO2N.GAF]

Originally posted by Fox's Sin Of Greed:

may matchmaking evaluation wont show, why is that.
even if its turn on. it is a bug or what

Do you have dota plus ?

[AIR]

Here is the thing, in my perspective:

They could show bars, but that would still be vague. They could show exactly the MMR numbers of each person, but that would be too obvious.

This is pretty much a system to allow selective matchmaking without any strings attached - you can keep on cancelling over and over until you get a team you want. The problem is, the more people get the system - soon it will be available to non-dota plus players aswell, the more you will have a stutter to start the match: you confirm, someone denies... you confirm, someone else denies...

It happened to me already, 4 times in a row, and it was really annoying.

The idea of the system is to know when the matchmaker made a mistake: for instance, put you in a group of people with bad behavior - when you don't want that just because it was fast, so you go ahead and cancel.

Regarding to MMR, I think it should not be more clear than it is: when the system was implemented, people were getting the information of all players and what they were doing? Dodging.

I like the way it is: it shows a keyword and after that keyword, the number of people. It also shows another keyword on top, and another for behavior. So you get to know if the match is, for example, good, and how many people are in that range that is determined good; or how many are in the range of very unballanced, so to speak. Then you get to make a decision.

It isn't meant to be pay to win: each time you queue a game, it is a lottery on the quality of the match. Now, before hand, you get the option to skip if you believe it is too outside of what you determine to be good.

[DO2N.GAF]

Originally posted by Charm:

Here is the thing, in my perspective:

They could show bars, but that would still be vague. They could show exactly the MMR numbers of each person, but that would be too obvious.

This is pretty much a system to allow selective matchmaking without any strings attached - you can keep on cancelling over and over until you get a team you want. The problem is, the more people get the system - soon it will be available to non-dota plus players aswell, the more you will have a stutter to start the match: you confirm, someone denies... you confirm, someone else denies...

It happened to me already, 4 times in a row, and it was really annoying.

The idea of the system is to know when the matchmaker made a mistake: for instance, put you in a group of people with bad behavior - when you don't want that just because it was fast, so you go ahead and cancel.

Regarding to MMR, I think it should not be more clear than it is: when the system was implemented, people were getting the information of all players and what they were doing? Dodging.

I like the way it is: it shows a keyword and after that keyword, the number of people. It also shows another keyword on top, and another for behavior. So you get to know if the match is, for example, good, and how many people are in that range that is determined good; or how many are in the range of very unballanced, so to speak. Then you get to make a decision.

It isn't meant to be pay to win: each time you queue a game, it is a lottery on the quality of the match. Now, before hand, you get the option to skip if you believe it is too outside of what you determine to be good.

Since I made this post, there have been some changes that addressed many of the issues I was having. For example, they added a numerical scale of 1 to 5, which has already made things clearer for me. Additionally, it seems to be more accurate than before.