Liberty75

The Case Against The 3/5/7 Match Maker (READY)

50 posts in this topic

1 hour ago, 8_Hussars said:

 

True, but that is also part of the appeal of the RNG MM system, on small city maps that balance goes in favour of the super heavies, on open maps to the balance of vision and stand off manouverable meds and lights... and its our job as players to figure out a solution.  I don't think approaching NASCAR lineups with equal and opposing tanks is the way either (not suggesting that is your proposal)...

Well you tell me how much you can carry as a team of 5-7 meds and TDs against 5 mice and a type 5 on Karkov, its just a question as to how much damage you can farm and praying they fuck up somehow to win. Nobody wants those matches and they happen all too frequently. 

Share this post


Link to post
Share on other sites

Posted (edited)

Edit: I update the original post with these numbers to bring the sample size up to 400 (from 200). I needed to remove the attachments on this post so I could add others. They weren't necessary anyway. The better charts are the combined 400 game charts.

I took a look at my friend's (Macduff48) replays from before 9.18. I looked at his tier 5-8 battles and left out special tiering tanks and light tanks (he had no light tanks in the periods I picked anyway). The two sets of 100 games were taken consecutively, as long as the battles met the criteria mentioned earlier.

He seems to have had some lucky game-time before 9.18.In the 200 games I sampled, he was top tier 41.5% of the time, middle tier 30.5%, and bottom tier 28%. The frequency of matches with more top tier tanks was higher for him, which would account for him having a personally higher amount of games at top tier than I did. So in a way, the teams that had more high tier tanks, the more likely you are to be the top tier tank. When I looked at his number, when there were 8 to 9 high tier tanks in a match, he was high tier 79%, middle 14%, and bottom 7% in that set. That 7% equates to 2 games and he was on one team with 4 bottom tier tanks and another team with 5 bottom tier tank.

I also looked at his number of horrible games defined as 3 or less bottom tier tanks and 5 or more top tier tanks. That number was the exact same number that I had, 3 games in 200 (1.5%). When I broadened the definition of a challenging match, 5 or more top tier tanks and 5 or less bottom tier tanks, he was bottom tier 17 times in 200 games (8.5%). My own number was 12 (6%).

It seems that the perception of being bottom tier in a horrible match was just a misperception, a myth.

When I looked at this post-9.18 numbers (another 200 games, but with light tanks now), his luck ran out as he had only one (1) top tier battle in 200 games! You read that correctly, that is less than 1%. That is crazy. He had 8 partial top tier battles. So he was only a top tier tank 5% of the time. When I told him what his numbers were, he told me he platoons most of the time (sometimes with me) and that usually screws him with bottom tier matches, and it did. He was bottom tier 50% of the time. That number doesn't include being partial bottom tier. If I include those numbers, he was bottom tier 65% of the time. Players that don't platoon often probably won't see numbers like his, but this is an MMO and a lot of us like to play with our friends. This is unfortunate.

Here are some graphics to illustrate the change he endured.

 

 

 

 

I will combine our numbers in the near future to increase the sample size, even though the trend is still basically the same.

On 7/6/2017 at 6:11 AM, Balthazars said:

... Personally I don't mind being bottom tier so much when there are only 3 top tier tanks. Better than being the 'token bottom tier tank' thrown into a horrific match.

 What breakdown would you consider a horrific match for a bottom tier?

Edited by Liberty75

Share this post


Link to post
Share on other sites
5 hours ago, 1n_Soviet_Russia said:

Well you tell me how much you can carry as a team of 5-7 meds and TDs against 5 mice and a type 5 on Karkov, its just a question as to how much damage you can farm and praying they fuck up somehow to win. Nobody wants those matches and they happen all too frequently. 

Well that's kind of my point you cant talk about tank imbalance without considering the map, put that match on prok, steppes, or sand river its the opposite. 

Share this post


Link to post
Share on other sites
9 hours ago, Liberty75 said:

 What breakdown would you consider a horrific match for a bottom tier?

Personally? Having half the team be top tier - and by top tier, I mean +2mm tanks (so 6-9 tier 10s as a tier 8, for example).

Here's where I'm going to agree with some other posters who said a better distinction here is between +1 and +2 MM spread.

If we're talking about +2MM spread, I'd consider the match-up pretty bad once we're getting to around 50% of the roster being top tier. Given many of the large disparities between +2MM tanks, when 50% of the enemy roster has significant advantages over your tank, the potential role you can play shrinks dramatically. If it's a +1MM spread, then personally the tolerance is higher, as the 'performance gap' isn't as big.

Assuming you're counting 'top tier' as being +2MM tanks, and not just any tank a higher tier than you, your friend's data pre-9.18 actually suggests that situations in which 50% of the roster is +2MM, while not the majority, are still very common, and effectively around half of all games (47% of the matches were between 6 and 9 top tier tanks, 49% with between 1 and 5, with the tiny balance being the rest). Even your own results suggest at least a third of games will have between 6 and 9 top tier tanks. Yes you're probably not going to be bottom tier in all of those games, but someone is, and MM is still spitting out those games with a high degree of regularity. When you consider that if 7 tanks are +2MM, you're likely to then have 1-2 tanks as +1MM, which means you're facing a situation where more than 50% of the roster is a higher tier than you.

The new MM improved things in my opinion because it makes the above scenario impossible. 3-5-7 means you will never be in a situation where 50% of the roster or more is made up of +2MM tanks, and at worst, 50% of the roster will be tanks a higher tier, with the majority of those only +1MM. At worst, you will have 8 tanks higher tier than you in a 3-5-7, or only 5/6 if it's a 5-10 distribution. To me, that's preferable as a bottom tier tank.

Share this post


Link to post
Share on other sites

@Liberty75 would you consider making your raw data available?  I would love to take a look at it with alternate definitions.

Share this post


Link to post
Share on other sites

Posted (edited)

2 hours ago, 8_Hussars said:

@Liberty75 would you consider making your raw data available?  I would love to take a look at it with alternate definitions.

It is spread across a few Excel spreadsheets, but I can post the numbers I have on one spreadsheet with the numbers combined. What is it exactly that you are looking for? I have Pre-9.18 info broken down by Top, Middle, and Bottom tier occurrences in a range broken down by top tier tanks in the match (as my charts indicate). And I have some info for post-9.18 broken down by frequency of being placed in different level Tiers in a match.

I originally started to investigate this because so many people (on the Wot forums) were saying how nice it was to not be a lone bottom tier, or in a small group of bottom tiers. It kept bugging me, because it seemed like that instance would be so rare and I didn't recall horrible matches like that happening often. I figured maybe I was biased since those matches haven't bothered me in years and maybe I was wrong in thinking they were rare. So I stumbled upon WoT Analyzer for other reasons and it occurred to me that I could use that program to look at my old matches to gather a sample. I wasn't sure of what I was going to find. My theory ended up being correct, that those match-ups were rare according to how others defined them to me. Other people may define a challenging match in another way depending on their abilities as a player. We are all at different skill levels and the game should be geared toward accommodating the greatest number of players, even if that leaves me out.

Also, I am not a statistician by trade. The last time I used statistics seriously was in masters class many years ago. So I have been dusting off my brain while doing this. :)

Edited by Liberty75
forgot post 9.18

Share this post


Link to post
Share on other sites

 

I put the 400 together to get a better sample, and this is what I found. I put it into charts so it is easier on the eyes.

04 Frequency Top Tier Tanks and bottom tiers in those matches.jpg

The above chart shows the frequency of top tier tanks in a match broken down into the listed groups. Along with that, I have included the frequency of bottom tier matches for a player in those groups. For example, your concern of matches having 6 or more top tier tanks would occur 44.5% of the time. Which seems horrible. If you look at the likelihood of being on the bottom though, that would only happen 8% of the time. So in 100 matches, you would have about 45 matches with 6 or more top tier tanks in them, but as a player, you would only be bottom tier 8 times in that situation (8 out of 100 games).

The next chart updates the general frequency of matches based on Top, Middle, and Bottom position between the old and new MM and the chart below that one is broken down by the new templates of 3/5/7, 5/10, and Same Tier matches. I think this addresses your +1/-1 comment, but I read it quickly, so I may have misunderstood what you were looking for.

Also, I adjusted some of the Pre-9.18 numbers to reflect 8 matches with zero (0) bottom tier tanks on these charts, so partial top and bottom (ex: 5/10 matches) in the new and old would match up better.

 

01 General Comparison between Old and New MM.jpg

 

02 Player Position Frequency between Old and New MM.jpg

 

I hope this helps to clarify some of your thoughts. If not, and I can help, let me know.

06 Percent Bottom Tier.jpg

05 Number Bottom Tier Matches.jpg

Share this post


Link to post
Share on other sites
1 hour ago, Liberty75 said:

It is spread across a few Excel spreadsheets, but I can post the numbers I have on one spreadsheet with the numbers combined. What is it exactly that you are looking for? I have Pre-9.18 info broken down by Top, Middle, and Bottom tier occurrences in a range broken down by top tier tanks in the match (as my charts indicate). And I have some info for post-9.18 broken down by frequency of being placed in different level Tiers in a match.

Essentially I would hope to look at the data through the 0, -1, -2 MM paradigm and break out Tiers, tank types, battle config (3/5/7, 5/10, and 15) and the pre-9.18 data.  Given you may or may not have that corresponding data in your set.        

Share this post


Link to post
Share on other sites

Posted (edited)

On 7/7/2017 at 1:43 PM, 8_Hussars said:

Essentially I would hope to look at the data through the 0, -1, -2 MM paradigm and break out Tiers, tank types, battle config (3/5/7, 5/10, and 15) and the pre-9.18 data.  Given you may or may not have that corresponding data in your set.        

I essentially used Excel as scrap paper, but I will screenshot that info and edit it into this post.

Here they are. The pre 9.18 numbers first followed by the post 9.18 numbers. The yellow highlighting on the numbers below the data for the pre 9.18 picture was to help me count the battles that had 5 or more high tier tanks and 3 or less bottom tier tanks in them. The red numbers on the same picture indicate the position of the player in that match. I don't think it is relevant to what you asked for, but I figured I would explain it in case you were curious.

Edit. I put the green squiggly line in there to separate the percents from the raw data on the second picture.

Edit 2: Link to google docs info: https://docs.google.com/spreadsheets/d/1l2F_VrSG2tptwDJxNCH9lrJbPAaQi92dyP39q09QAIg/edit?usp=sharing

 

 

(Pictures deleted. No longer needed. See Edit 2 above.)

 

 

Is this what you wanted?

Edited by Liberty75
Added Google Docs link

Share this post


Link to post
Share on other sites

tx, you could also just upload it to a google docs spreadsheet and share it here.

Share this post


Link to post
Share on other sites
12 minutes ago, 8_Hussars said:

tx

It is done. Look in the post above.

Share this post


Link to post
Share on other sites
28 minutes ago, 8_Hussars said:

tx, you could also just upload it to a google docs spreadsheet and share it here.

Crap, I didn't even think of that. Well, there aren't really that many numbers. People that want to play with them can take a few minutes and plug them in. It shouldn't be a hassle.

Share this post


Link to post
Share on other sites
On 06/07/2017 at 3:28 AM, Kolni said:

A huge thing to note as well is that assault games have completely disappeared in tier 10 on EU. 

It used to be the bread and butter of the map rotation before. But that has probably more to do with the new map rotation then the MM.

On 06/07/2017 at 2:19 AM, Dukebarry said:

An interesting issue that I am exploiting is the tier 3 joke. With the new MM tier 3s will never see a tier 5. If you drive tier 10s you have a prety good chance of seeing mostly tier 8s.

Yes. They did fixed low tier now. 

On 06/07/2017 at 6:02 AM, Spartan96 said:

Anyone else noticed a larger number of TDs arty and LTs being placed as the top 3 top tiers of a large number of these 8-10 matches?

Yeah, i played the 132-A and then the 132-1 and you get crazy amount of top tier in 3-5-7.

On 06/07/2017 at 7:56 PM, GehakteMolen said:

3-5-7 means:

  • 3/15 = 20% top tier
  • 5/15 = 33% middle tier
  • 7/15 = 47% bottom tier

This math is only true if 3-5-7 represented 100% of the battles. But it's not. there is 5-10 and 15-0-0.

 

If WG reduce the amount of 3-5-7 and increased the amount of 5-10 and 15-0-0 we could see a greater % of top tier battles. 

 

This is something i posted on reddit on how i see this new formula could be :

The 3-5-7 shouldn't be the most type of game you get. This is how it should look to be enjoyable.

  • 50% One tier game
  • 30% Two tiers game
    • Of which 33% should be top tier and 66% should be bottom (5/10)
  • 20% three tiers game
    • Of which 20% should be top tier 33% should be middle tier, 46% should be bottom tier (3-5-7)

If the MM could keep up with this format this is what we could get (or at least get closer to):

64% top tier, 26.7% mid tier and 9.3% bottom tier. 

hhjpVEK.png

 

Obviously these number could be tweak (let's imagine the following). 

  • 20% One tier game
  • 30% Two tiers game
  • 50% three tiers game

Would equal to this :

40% top tier, 36.7% mid tier and 23.3% bottom tier. 

QwHWRhc.png

 

What i don't understand is WG trying to hold on to the three tier spread so much when 1 and two tier battles could be ever so good in more numbers.

Share this post


Link to post
Share on other sites

I'm probably alone, but I liked the three tier spread when it actually spread you around in a random format. In one game you go out and you're a bully as top tier, the next you play more of a support role as bottom, and as middle you can do a little bit of both depending on the tank and match-up. It was a nice bit of variety most nights. Now it's pretty dull. Restrictive templates in general are dull, to me anyway.

I hate to say it, but I get the emotional feeling that the game is pushing me away, like that girlfriend that is too chicken to break up with you, so instead she treats you coldly. I enjoyed top tier matches. They took it away. I enjoyed the variety of the match maker, and they took that away. I hated the unbalanced vehicle distribution on each team, they left it in. Maybe it is just my personality type and the game is actually better off now than it was before for most players? Sorry guys, just a little gamer-nerd venting here. :)

Share this post


Link to post
Share on other sites
25 minutes ago, Liberty75 said:

I'm probably alone, but I liked the three tier spread when it actually spread you around in a random format. In one game you go out and you're a bully as top tier, the next you play more of a support role as bottom, and as middle you can do a little bit of both depending on the tank and match-up. It was a nice bit of variety most nights. Now it's pretty dull. Restrictive templates in general are dull, to me anyway.

Nah your not alone.  What was lost with 3/5/7 is less emphasis on reading the line up and coming up with a plan to deal with it.  Granted that skill appeals (or is practiced) by a fairly small sub-set of players. The game is balanced around the random pub player (simple game with some skill and luck) where RNG assists them, and not those who prefer the depth of the game mechanics to master and over-perform  in where RNG keeps them in check. 

The Tier 3/5/7 format is better for grinding tanks and the pre-9.18 MM was better for variety.  The problem players continually identified was tank imbalance (and weighting) and WG chose to address it through the 3/5/7 format (which is not necessarily what players wanted) much like all the arty complaints where WG chose to introduce the stun mechanic (which also is not necessarily what players wanted).

   

Share this post


Link to post
Share on other sites
4 minutes ago, 8_Hussars said:

The Tier 3/5/7 format is better for grinding tanks and the pre-9.18 MM was better for variety.  The problem players continually identified was tank imbalance (and weighting) and WG chose to address it through the 3/5/7 format (which is not necessarily what players wanted) much like all the arty complaints where WG chose to introduce the stun mechanic (which also is not necessarily what players wanted).

   

I hear you. Sometimes we should be careful for what we ask for. :)

Share this post


Link to post
Share on other sites

I speak purely from a gamer's point of view without referring to figures and data. For any competitive game there are 2 influential factors, Balance and MM.

Balance so that MM has a solid foundation to work on and individual players has a fighting chance against each other.

MM so that there is a fair match-up to compete on an even ground, and individuals are not subjected to a biased environment unnecessarily.

Low server population should never be an excuse for poor match-up i.e. Asia server, bottom tier frequently, pre or post patch. A good MM algorithm will take into account variation in population size, tank class, tank tiers, so called "adaptive" MM. 

MM continues to be a talking point. Why not +/- 1 tier? The current balancing need not be altered to accommodate this.  Why constant bottom tier? I almost gave up grinding the KV-13 pre-patch because I played at odd hours and was constantly -2 bottom tier, for example being 1 of  the 3 tier7 tanks in a tier9 match, the other 2 tier7 is a LT and an arty. What do you expect a single KV-13 to do in a tier9 match? Go kill their tier7 arty and LT? or try to outpen and out dpm a E-75? I almost went broke from all the apcr spam because of the poor pen gun meeting +2.

The playerbase itself is already a big variable. A wide spectrum encompassing the competitive to the hopelessly apathetic group.   Add that to a less than satisfactory framework of MM and balancing, the end result is a discouraging state of frustration. 

 

 

 

 

 

Share this post


Link to post
Share on other sites

It seems that more people are waking up to the failures of the 3/5/7 template based MM. Content creators are now criticizing it. Maybe WG will finally notice?

Share this post


Link to post
Share on other sites
11 hours ago, Liberty75 said:

It seems that more people are waking up to the failures of the 3/5/7 template based MM. Content creators are now criticizing it. Maybe WG will finally notice?

Citations?

Share this post


Link to post
Share on other sites
56 minutes ago, 8_Hussars said:

Citations?

Multiple content creators questioning it. (ex. AgingJedi, QB, Claus, and others over the past few months)

Recent EU Forum Poll, where most of the complaints are about the post-9.18 MM.

Do you prefer how this game is after 9 18.jpg

 

No one can prove templates are better than a random MM for the general player base. Any supporters of the current template MM base their opinions on perceptions (usually incorrect) and feelings.

Share this post


Link to post
Share on other sites
1 minute ago, Liberty75 said:

Multiple content creators questioning it. (ex. AgingJedi, QB, Claus, and others over the past few months)

Recent EU Forum Poll, where most of the complaints are about the post-9.18 MM.

Do you prefer how this game is after 9 18.jpg

 

No one can prove templates are better than a random MM for the general player base. Any supporters of the current template MM base their opinions on perceptions (usually incorrect) and feelings.

Meh, I can't be bothered to wade though and find applicable citations hence the request.  For now I will leave that as unsubstantiated (hand waving) "multiple content creators questioning..." I hope you would agree that questioning has a different connotation than criticizing...

As for the pole, it does not differentiate between 9.18 MM and the other changes nor the myriad changes since.  (Not to mention its an official forums poll...nuf said)  About all it may highlight is the 9.18 and after changes are not overwhelmingly endorsed.  Nor can we say that those figures are any different than any other poll based on a different patch.     

 

Regardless of anyone's stance on the issue you can't dismiss the discussion with what you posted because conversely to your point, no one can prove templates are worse better than a random MM for the general player base. Any supporters of the past current template MM base their opinions on perceptions (usually incorrect) and feelings.  Both sides base their opinions on perceptions and feels, so until there is a consensus on what exactly are the weighted goals of MM.

As for the proposals to fix (or as an alternate to) 3/5/7 MM here

  1. teams would have the same amount of tanks per tier of top, middle, and bottom. (Implemented in 3/5/7)  
  2. teams will have similar amounts of tank types per each tier with a +1/-1 differential, and a +1/-1 differential per team overall. This means the same amount of TDs, Mediums, etc., with one difference allowed; (Implemented in 3/5/7)
  3. platoons will be placed as close to the same tier level as possible when teams are created or the inferior team will get two (2) lower tier level platoons to counter a top tier platoon on the enemy team; (meh, this was not implemented in either the old MM or new MM.  Is this a really an MM issue?)  
  4. tanks of similar attributes will be identified with each other in the weight/identifier database and placed evenly on both teams at the same tier level (for example, the Bat.-Châtillon 25t and the TVP T 50/51 will both be identified as a medium AND an autoloader so each team has them placed evenly);
  5. the MM will attempt to create all Tier I tank battles first, and only have up to six (6) Tier II tanks per match if it cannot create an all the same tier battle for them;
  6. Tier II and Tier III tanks have +1/-1 matchmaking;
  7. an option in the Settings panel above Grand Battles will allow players to have the MM try to create all same tier matches for them at any tier.
  8. some of the above rules will be relaxed when server populations are low to cut down on wait times.(Implemented in all MMs to date)

Comments 4. and 7. are surprisingly puzzling.  If it's upheld that the 3/5/7 templates are boring (eliminating the rich variety) and that the random Tier amounts of the Old MM were superior; then why is similar tank attribute matching and an increase in reduced Tier spread matches entertained?  Ultimately, they also serve to further "constrain" the matches, reduce the "rich variety", and experience of players.  Regardless of what they are called and even with designed in "looser constraints" they are just templates that meet a different expectation, No?

Whether either/or/neither (old or new) MMs are in the best interest's of "the player base at large" is still open for discussion; however its clear by the forums; (no poll required) is that all most players want is to be "Top Tier" a high proportion of the time and consequently club -2 tanks... and they don't care what form of collateral damage it takes.  Balance, game play, variety, XP, game economics, just don't matter.

I would be interested to see the salt if WG inverted the templates and went to a 7/5/3  and 10/5 template...

Share this post


Link to post
Share on other sites

No hand waving. And I told you, in that poll, most of the complaints in the thread were about the MM. Browse the EU forums. They are mostly against it. I can't read Russian, so I can't speak for them. On the NA forums you have a few players that believe the current MM is good, but when asked why it is better, all you get is some line from WG's press releases about not having bottom tier matches as a single tank or they are happy to have 7 bottom tier tanks in a match. When you show them the numbers that things were never that bad in the previous MM to justify the excessive bottom tier matches now, they just shut down due to some cult-like devotion they have to the new system. The conversation ends up in insults or they just don't want to hear the truth. They think you are lying to them or something. It is rather odd.

Feelings are feelings. Perceptions are not feelings and perceptions can be wrong. It happens all the time and open minded people can see when their perceptions are off when the evidence is there. The stubborn remain ignorant, like the cultist attitudes I mentioned.

Take out the balancing factors that could have been introduced in the previous MM or something like it and It is clear that the 3/5/7 template system is a failure on many levels. We are still waiting for someone to have a good argument as to why a 3/5/7 strict template is better than a team balanced random MM (or even just the previous MM!). So far, nobody has one. It was a bad trade-off compared to what we had before it.

About the proposed MM:
Walls of text about a game mechanic are not fun, so I guess you didn't read the entire text of the proposal. The proposed MM was not presented as a set of changes. It was presented as a model MM that would address the concerns of most players. Your notes of " Implemented in 3/5/7" are correct and many people like those characteristics, so they were kept in the proposed system. Most players acknowledge that the balancing factors are a good thing or at least tolerable. The core of the problems we have with the current MM is the 3/5/7 template and that needs to be removed.

The proposal was put together after months of debate between players. That is why you have more variety AND some balancing factors. The variety crowd gets some of what they want restored to the game and it would also satisfy the balanced teams crowd. It wasn't meant to be an extreme solution one way or the other. It was a compromise between two factions.

WoT is a repetitive game by its very nature. All strict templates are toxic over time because they make the game MORE repetitive and in the end the game will become lackluster for many long term players.

 

Quote

I would be interested to see the salt if WG inverted the templates and went to a 7/5/3  and 10/5 template...

You're funny. Although I bet people would like them more as they would be top tier about 60% of the time and bottom tier in a +2/-2 game only 5% of the time.

Share this post


Link to post
Share on other sites

I updated the first post to reflect the combined sample size. If anyone notices any problems, please let me know.

Share this post


Link to post
Share on other sites

  • Recently Browsing   0 members

    No registered users viewing this page.