Jump to content
  • Sign Up

Recommended Posts

Hello Friends !


Recently i was thinking about making an excel sheet to help determine the best match up combination based on your team composition and the enemy team composition.

The idea would be to fill 4 cells with your team composition and 5 cells with enemy team composition and then press a button to generate what class would best fit the current composition in order to win the current match.

Of course this is not easy to overcome and i'm looking for experienced players feedback as to what criteria i should put into this excel sheet.

Link to post
Share on other sites

Well, matchups is just a pseudo belief about what the other classes are playing, because it assumes all classes in a match are playing a meta build. Matchups are based on builds, not on classes...so the spreadsheet will always be inaccurate to some degree.


But the best place to start is to look out how a neural networks would do this kind of calculation. A neural network would look at this problem, by calculating a kind of value for properties on the level of individual skill matchups and then going up from there. an example of a skill matchup would be if Skill A versed Skill B, if Skill A is supposed to win, then the system gives it a value of either 1,0 or -1, where 1 is a win, 0 is a tie, and -1 is a loss.


Instead of parsing the system by skills, which there are thousands of them, you would have a layer in your network that parses skills on a behavioral or mechanical level instead. So as an example you define W X Y Z as properties of skills (W being condition based, X being power based, Y being interruption based, Z being defensive based) then If Skill A has property X Y and Z, while Skill B has property W and X, then whatever way you have your system defining the matchup, the system would define each skill as a win, loss or tie, by looking at the sum of it's properties rather then looking at the matchup of skills...again since there are thousands of skills and skill combinations.


So just an example :

W -> X  = 1

W -> Y  = 1

W -> Z  = -1

W-> W = 0


For all skills with a property of W, this value is imposed in a calculation, and you would do one for each iteration.


In the next layer, if a meta build (1) has Skill A B C and D in the build, each skill is parsed into those properties, and then each property determines a value for each skill, and each skill when compared to each skill get's it's own value...not a summation of the values, but it's own -1, 0 or 1 against other skills...final layer is again the same process until you end up on the level of classes.


In this way, all you have to do is a design a calculation that looks at a small fundamental number of properties, to determine some calculation for a complex object like a class using 20-50 skills.


The above examples are ugly approximations of the actual games mechanics, so you'd have to do the diligence of looking into how to define these game mechanics and parse them into ways you can actually use as a meaningful metric as for what mechanics win over others. Some of these are obvious like Boons vs Boon strips...but others are less obvious like boons vs Condition based CC.


Look at this video, it will help you make your spreadsheet using a neural network.


Link to post
Share on other sites

This idea would be rendered completely inaccurate when taking into consideration that during 2s, people swap utilities, traits rather often, which can completely change the dynamic of what class is good vs what. IE: Reaper vs. Reaper with no Corrosive Poison Cloud in play, it is a balanced 1v1. But when one Reaper brings Corrosive Poison Cloud he is able to completely negate the use of the other Reaper's Lich and now he is countering that Reaper.


So if you were serious about this idea, take this into consideration. The meta for 2v2s or even 3v3s is not really a meta designed around "one build does it all" kind of specs for a given class. The meta for 2s & 3s is more about "What can a Necro do for swap play when he is able to swap between Core Necro, Reaper, Scourge, and various utilities/traits".


In other words, don't base your spreadsheet on meta builds. Base it on Core Class potentials.


lol but if you want to cut to the chase this season, here is my opinion of what I've seen so far:


S Tier duos:


  1. Scourge + Prot Holo
  2. Scourge + DH
  3. Scourge + Spellbreaker
  4. Double Necro in general who know when to toggle between Core, Reaper, Scourge

Anything else falls under these.

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Create New...