Skill based MM in random battles is what some people think the game is missing and there are is a very vocal minority that basically postulates the following:

Random matches should have some sort of skill based MM because there are too many matches that **** due to imbalanced teams!

Well, this reasoning is a perfect example for an Ignoratio elenchi. WoT is very successful, the MM is an integral part of this success, so logic dictates that WoT is successful BECAUSE it has random MM and not despite it, which, in turn, means that everyone arguing for skill based MM, because it would make WoT "better", is missing the point.

And precisely BECAUSE the MM in random battles is one of the major selling points of the game, WG is understandably very reluctant to change anything about it.

That aside, how could skill based MM be introduced and what would likely result from such an introduction?

First of all, WG needs to implement a new method to rate players skill. WNx, Eff, Personal Rating all have one in common: they only work to portray a players skill over a good amount of random battles and only random battles.

Why? Well, answer yourself one question:

What is harder: doing 1k damage against a Unicum player or doing 1k damage against a tomato?

And still, right now, these two things net exactly that same rating but thats ok, because everyone will face the same teams over a large amount of battles because teams are random!

However if skill based MM would be introduced the above would no longer hold true and thus we need a new rating method, something like the one Arpad Elo invented. However note, that this system and all who follow are made for 1vs1.

The advocates for skill based MM then usually go about how this fails to be a problem, as other games have implemented such a rating too.

Well, yeah, not. You hear me, there is not a single game in the world that uses this system with 15 vs 15 and no, it is not even close to 1vs1 because the complexity of finding a battle raises exponentially for every pair of players you add!

So any rating would be complex and therefore expensive to implement. This would be no small feat at all. Yet many people that are all for skill based MM totally disregard this.

But what's with the game itself? Does its structure lend itself to skill based MM well?

Now we need to look at the classical skill based MM games like Starcraft and LoL. Both have two things in common: between equally skilled parties they take much longer than the average WoT battle. Most of this additional time is "build up". You have several minutes in which you basically do "your motions". Most games are very similar during this period and are not extremely confrontative yet still very important and very different for bad and good players.

The equivalent in WoT would be to drive to your first spot...

In other words, random battles in WOT are missing the long buildup phase that allows a good player to get into the game without too much risk and thus to show his skill. Neither SC nor LoL is usually won or lost (for any individual player) in the first 2 minutes of the game (between equally skilled players).

But in WoT that is not true. If you watch vids of Unicums playing, you notice that they usually play very safe during the first 2 minutes of the game, avoiding to be hit. What that means is that they create themselves a buildup phase where they can shape the game without of fear of immediate termination and thus inability to play out their skill fully.

Why does that work? Right, because they have much worse team mates, that don't to that, that spot/take damage/soften up the opponents for them. To quote Garbad: when tracked in the open before any number of enemy guns, the performance of a Unicum and a Tomato is exactly the same!

So the appeal of the game would drop for very good players with skill based MM because, in extremes, games would play out as the WGL Finals - very boring! On the other end the Tomato league would be a hodgepodge of bots, retards and afkers and thus totally bad for the newer player to start their game in. They would never see a good player doing something good and thus would have a hard time learning anything ingame.

Obviously the main advantge would be with the mediocre players. It is no conincidence that most of the advocates for skill based MM are high orange to high yellow players though. They dream of their cuddly orange/yellow world totally missing that there are no more reds to take advantage of and, if they ever get any better, they will be the bottom feeders in the mostly green-blue-purple league where everyone from your own team knows exactly how to take advantage of you.

Sounds good?

OK, as leagues seem somewhat difficult, how about equally skilled teams? Now this approach will likely never work due to the huge computational burdens for the matchmaker. Also Unicums, especially when platooned, would be either put into horrible tomato teams 24/7 or against the one other Unicum platoon all evening long.

The third solution would be a team reshuffle. Basically the MM would do it's work and create two teams in the fast and reliable way that we all love so much (grin). Then the "skilliator" would take over and look wether it could switch team members around to create more evenly matched teams.

Now I actually logged 158 games with screen shots and typed the WN8 together with the tanks and platoons into Calc and tried to work something out. I found that roughly 10% of all games could have been made much "fairer" while still retaining tank/platoon composition. The rest could not. Main problems are the platoons and the good players at low tiers phenomenom.

Hm, while certainly not an exhaustive test, this solution seems to be a lot of work for not that much effect.

TL;DR Skill based MM seems like an expensive, high risk low return venture because a huge part of the great success of WoT is how random works right now.