I am resurrecting old forum topic [Discussion] Balance, Bias, Matchmaking and Battle Ratings - General Discussion - War Thunder - Official Forum so there is proper place to discuss this and the discussion doesn´t flood the Rumor round-up thread.
I will also repost the opening post from the old forum.
War Thunder - is an unique and complex game from different points of view. We are making things that were never ever done before by anyone and our challenge is that we have ambitious goals for our players satisfaction.
Players are fighting using vehicles that reproduce the actual ones with a high degree of reliability. Vehicles in the game are from different dates of release and operations, different countries, types, they were used for different tasks and which cannot be compared directly one to another: bombers, attackers, hydroplanes, fighters, carrier-based fighters, heavy tanks, anti-aircraft artillery, tank destroyers and so on.
In real life most of the vehicles would never be fighting together in one battle (or even at the same theatre of operations).
At the same time the battles themselves are made for random players. A player may choose any aircraft or ground vehicle and press “To Battle”.
Matchmaker (and the battles themselves) should be made so it won’t just create battles, but create them for any vehicles that will be in the queue, for example, there could be only bombers in there, or fighters, or only attackers (in any combination for any country and at any rank), and battle still need to start, since players won’t (and shouldn’t) choose specific vehicles to get balanced teams on both sides.
Matchmaker will try to select the most optimal combinations, but for battles with no ability to re-spawn multiple times, nothing can technically guarantee the ability to select teams based on vehicle combat roles.
Additional nuance is the fact that players progress is personal and not team oriented, a player should have ability to earn, and progress in research and play in any vehicle he chooses.
No real balance based on “technical characteristics”, or “historically fought against each other”, or “production date” is even theoretically possible.
For example, B-17 was produced in same years as I-15, it even has similar speed characteristics, but it is obviously not an equal foe for I-15. In an imaginary duel of B-17 versus I-15 (even against four I-15) the B-17 will most likely will be the winner.
At the same time in hypothetical battles (its impossible right now by ranks of the planes and their BR), where in one team there were one or two B-17’s and in another team - one or two I-15’s, victory can be achievable by any side given decent team play - more so, the personal effectiveness for the I-15 may be higher than for the B-17 (because there are obviously more aircraft than just B-17s on the other team).
The matchmaker system (and the battles themselves) are calculated this way:
- it will gather battles for any vehicles in the queue
- it will provide the maximum possible balance in sessions (in average)
- it will provide maximum equal personal efficiency (based on research points and silver lions amounts) in battles (in average)
Matching occurs only based on the BR and the vehicle rank: vehicles of 3 rank differences cannot be in the same battle, only one rank or two, the BR difference will not be more than 1 for aviation and 0.7-1 for ground forces (we are not counting that there is a possibility to get weaker vehicles into a battle while in squad or in games with multiple re-spawns).
We are calculating BRs based on statistical data. It is gathered from the amounts of fly outs, average life time, shot down enemies, lost vehicles, critical hits, destroyed AI vehicles, destroyed bases and generally by everything what affects victory and personal progress, but without accounting if the team won or lost (we collect that information as well and analyse it to evaluate matchmaker job and also balance of the mission, but since teams are not necessarily created balanced by the type of vehicles, the fact of the victory and loss is not considered for BR calculation).
This happens approximately once per month and also usually after 2 weeks after changes in FM/DM/Missions etc. The BR update is not always happening right after it.
Incoming changes to BR are aimed to complete two goals.
For aviation - it is just a regular re-calculation based on the statistics. We have taken into account players opinions and, possibly, will change only the most changed planes BR (by statistic) from last patches. It will allow us to better analyse those changes and make them smoother.
For ground vehicles we also plan to increase possible brackets, while saving the possibility of any tank is able to destroy any tank it meets in battles. The old brackets (for 0.7 BR that is more than 20 possible brackets) was leading to the possibility of very long queues for tank battles, especially at 3-5 ranks, and at the same time variety of the vehicles in combat was very limited.
Discussion and further changes for matchmaking
Players always discuss matchmaking saying that its not ideal. Partly it happens because people instinctively think that vehicles with equal BR should be “equal” and “fair” (based on technical characteristics, date of construction or participation in real battles), while forgetting that comparison of the incomparable - different types of vehicles made for different tasks and often never encountering each other in real life battles, like carrier-based fighters or long-distance fighters and frontline attackers and fighters.
Often players forget that specific aircraft can meet other planes more often from the list of possible opponents, than those which they are comparing one to each other, vica-versa, its rare that there will be teams made from one type of vehicle only. At the same time balance should be based on teams and within the limits of our in game missions.
But we are listening to every constructive suggestion and are ready to participate in any dialogue.
Of course we cannot physically participate in dialogue with all the hundreds of thousands of our players at the same time.
So if you will collect suggestion for matchmaking (with taking into account specific points, like, for example, not excluding from the system any type of vehicle and not creating the situation that some player may never get into the battle if the rest of the players select same type of the vehicle as he did) - then we will look into it and will answer to them. At least to those that take into account all the specifics tasks required of the matchmaker.
We are also ready for online discussions and answers to questions in real time with “emissaries” of the community - of course you will need to select those who will be presenting your case and asking your questions.
I hope you understand where I am coming from with this and would ask, rather than assume we are not listening, try and appreciate we are really only doing our best for the entire community, not just one group or another.
Todace (Kirill Y)
This topic was allowed to be left open as a discussion about BR Rating and matchmaking - I would request you stay on topic and take any other discussions elsewhere, this is an emotive topic, as many of our merged and herding topics are, and we don’t need you mixing them up. It doesn’t help anyone.
This is the Match making and BR! Look at the pinned post as to what we are supposed to be discussing in here, not connectivity, not balance. Please look at the other “major” topics we have pinned in the general area, as you can see, there are clearly other areas where you can discuss them. It isn’t here. If you prefer, we can merge them all into one, but as im sure most of you would agree that isn’t practical.
Stay on topic.
As designers the devs have to choose 1 of 6 options
Historical matchup, obviously this will be good for some of us (Historically accurate), situations like reserve pz II meeting the KV1 would obviously narrow that group even more as although it is a historically correct matchup, it would be a very emotive encounter generally. Of course there are a group that would not want to even entertain this.
Performance matchup - vehicle stats alone - some might say this is satisfactory, but in opposition to number 1 above many match-ups would be way of the mark in terms of reasonableness, this would tend to upset the fair balance of some players in game as a good performing vehicle isn’t always easy to fly for example.
Player performance - if we based match-ups on this, although some would welcome it, probably many thinking about, people like me or sim guys would really get annoyed with it. I consider myself a team player rather than a high scorer. So I would in effect be put into games where the teams are less goal orientated (random flying or driving about shooting at seagulls).
Absolute balance (where vehicles are exactly the same as you in battle) red vs blue - no interesting encounters - purely competitive matches (good for some tourney work i guess).
No mm at all - well i guess we can imagine the games that would be played using this.
A mix of 1 to 3 above (mixing 4 or 5 into the formulae negates their effect)
So 6. is where we went with this, it may not suit all, but it suits more than the numbers not happy with any other choice.
The answer was always going to be complex but we strive to reach the ultimate balance.
edit - i firmly believe the key to this is spread as many have observed, but whilst we can edge aircraft to a narrower band due to sheer numbers of vehicles available, ground vehicles are not quite there yet. We could narrow it right now sure, but this would create more chance of FOM standard matches (a la wot) which would tend to be more individually competitive. I’m sure there are some that this would suit, but I do believe the majority would find it boring.
So as we add more vehicles to the lineups (and this includes aircraft too as that could still be narrowed) the situation will improve. Hence why we have to add vehicles with every major.
Relating to BR - what you see in game is what you get in terms of rating (failing errors) but that a players performance is not checked on the mm sort (BR isn’t adjusted on the fly), it is done on a broader scale or indirectly if you prefer, before the BR is set in game (and periodically updated), we don’t weight players knowingly on entry into a match.
For BR changes to occur periodically -
Player a) performs well in non premium specific vehicle - that is treated as a " + " (fractionally) in terms of BR rating so the aircraft will tend to be uprated (higher) obviously this isn’t obvious when only one or two perform well with it but rather a large chunk of players performing well in it will generally cause it to go up.
Player b) performs badly in exactly the same vehicle (all counts) but in the premium line. this will be treated as a " - "(fractionally) and so the BR rating will tend to go down (in reality again it takes a fair chunk of players to cause this.
After BR changes have been applied by the formula (we list changes when they happen) we observe the behaviour in game and follow WTPC and player feedback on the changes and may manually adjust them soon after the change.
At this time, we do not apply player performance in terms of MM directly instead relying on the use of the vehicles before to determine what is “balanced” for the grouping within the matchmaker.
Many MM problems(not all) seen by players are a result of poor choice of lineup or indeed a complete ignoring of BR tending rather to look at tier.
As seen by recent chronicles - many in the community cannot work with historical matchups which is why we tend to favour aircraft/player performance to provide the balance. The theory is that ultimately the mix will balance itself out and provide no OP or UP vehicles in a battle. Every time we add a vehicle via updates, we actually help the situation and ultimately this will allow us to reduce the spread and keep “Flavour of the month” vehicles to a minimum. Obviously it isnt the only answer and it is quite a complex system, especially when you also consider that QM is taken into account which adds another filter into the mix to make sure that team power is balanced or even pre chosen when used exclusively. Obviously it isn’t fool proof yet, (as we need to manually adjust from time to time) but it is getting there.
Within the MM, the BR will only be applied after the Tier (or year grouping if you prefer) has been applied thus keeping the greater number of year groups separate.
BR changes (when they occur) are patched in and the changes/patch notes are made avail at same time (sometimes pre-planned alterations are announced in develop or news ahead of time).
Again changes to BR come with patches (major or minor) which are never specific about their release. As far as technological advances go. This doesn’t mean what you think it does. It is not this tank is more advanced-therefore it shouldn’t fight this other one or vice versa.Simply put in war especially WWII. Each side enters into a race to make all aspects of their prospective arsenals better in one way or another until they win (or that is the goal). So for instance the German’s used a 88mm cannon variations (a lot). There were tanks specifically designed with defeating THAT tank family that they were obsolete after the war and new stuff was needed by Korea, and the coming Cold war (interested parties designed things for possible war with the other). That is natural occurrence and cannot be compared to MM putting better or worse vehicles into a game nor influence it’s BR. The tanks are listed in 5 development eras their BR’s adjust to how well they do based upon player usage) Its complicated but it’s simple,
Russian Bias! A statement that is very often discussed on forums and communities…
"My favourite one! There is well-known Russian Bias in War Thunder for the International community, but what is much less known for these communities is the U.S. Bias and German Bias, which are very lively in the Russian community, where people hadn’t heard about Russian Bias. So, depending on the country you are coming from, Gaijin is biased against it. We are Russian biased because we are Russians, we love Russian vehicles and thus make them overpowered and everyone else underpowered. We are also biased towards the U.S., because players from the USA pay more. So, we make their vehicles overpowered as a reward for the money they are investing, and the Russian players have to suffer because of that. We are also Germany biased, for two combined reasons. They also pay more than Russians, and in addition to that - it’s untranslatable, but in Russian it just means we “simply love Germans” (“Немцефилы”).
So, I think if all communities would speak one language and we could combine them, the whole Bias thing would just “annihilate”. But before that, we have to live with mainly three biases. Well, there is also British Bias because Spitfires are cool. The only bias we really don’t have now is Japanese Bias (or we simply haven’t get it translated from Chinese\Korean community)."