Matchmaking Algorithm for Normal Warzones

Meh. In sub-50 I noticed with my sage that I am closer to breaking 200k or more when I am the sole healer as opposed to when there is another healer in the group, but DPS 100k even there is not hard to hit unless someone else is doing the killing and I am relegated to just the objectives. That is sub-50 though. There is a MUCH larger health pool in the 50 bracket to go around.

EDIT: Oh and yes please on the matchmaking. I generally PUG, but don't mind facing premades if others on my team all know whats up. If you think a like, which with the cookie cutter warzones, there isn't all that much to think about; you don't need voice chat. I just hate it when I know we are against a premade and the people I am PUGed with have "weird" ideas about how to approach the situation. or if they stay and whine the entire time. Or if they keep quitting.

I would like some clarification on "premades kill effectively". I mean I have put up less than 20k and won games of huttball but you need to put out damage to win that match but you are not going to win on any other warzone where you need to clear nodes/doors to cap objectives.

Awesome work. When you said in the other thread that you'd think about an algorithm i absolutely didn't expect that.

Did you consider including gear in the equations? As in, 300k damage in full recruit is better than 300k dmg in minmaxed EWH. Once people have WH gear i think it's fine to not consider gear, and since the grind to WH is super quick it's probably not necessary, i'm just curious on your thoughts on this.


also

It would also help premades, because if there are decent players in the queue you get a much more fun game, which is imo the whole point of this. That said it would probably be smart if Bioware gave an exact maximum time after which the matchmaker just tries to get any match going, ie it increases the range until after 3 minutes, it just ignores rating alltogether. If they don't do that people will blame there long queue times on the matchmaker wether it's true or not...

Pretty much in all of them if you take nodes quickly and crush the opponent, they usually just end up sitting on one node or they trickle in one at a time. I don't know about the new one since I've barely played it, but this was true in Huttball, NC and ACW. They've changed the timing on VS but sometimes if you started on offense and got ahead of the opposing team quickly you could just keep opening doors fast and that game would end quickly too. At most you'd get some damage on defense, and in fact VS used to be the only one where I could get any decent type of big stat score while doing effective damage.

All the top premades were getting really low damage scores back when I was actively playing this game around April-June or so. This included teams like STS and T.W.A.T. who came a little later but were good teams. I think they all quit. My own premades and all the PVP guilds I was in also disappeared, probably everyone got bored. It's awfully hard to get good premades against each other in this game because nobody sticks around too long and the gear grind makes too much of a difference at the highest end. So it's generally just PUG rolling, or possibly weaker premade rolling, and rarely competitive back and forths.

it would definitely be something interesting to look at, but like you said since WH is very easy to obtain there would probably not be a lot of practicality to doing it.

my thoughts exactly, didn't read the last 2 pages I'm sure people *****ed about your thought and said you are stupid, it would be the puggers that get their *** handed to them.

unless cross server ques were added then I would be down for something like the OP posted.

Can you please state the goal of the algorithm more precisely (in mathmatical terms)?

If I am setting the goal it would be: more close matches is better, more one sided matches is worse.

I would use a histogram of match scores to evaluate different match making algorithms.

So an algorithm that makes most everyone's win rate close to 50% but 95% of the matches were one sided results would be bad according to the above criteria.

is focusing only on close / lopsided matches enough or should the criteria include a histogram of win rates? I.e. do we want to minimize the percentage of players who's win rate is 0-10%? Might this naturally fall out of the "close matches" criteria?

Only after we settle on a stated goal for the match making can we evaluate various algorithms.

Also, for regular WZ, we must include an algorithm for back filling. This is probably trivial after solving the main problem but it is still worth discussing.

It doesn't really matter if they can handle it or not, what matters is if BW ever decides to stop adding yet another WZ for a release cycle and instead focus on a few quality of WZ issues like this one.

OP is definitely trying hard to offer a constructive suggestion and I applaud him for that.

As others have noted, the algorithm could include time limits that relax matching criteria to avoid super long waits.

i think what you are saying is that you want it to follow a bell-curve? where most people are in the middle w/ ~50% W/L ratio, 10% have high W/L ratio and 10% have very low W/L ratio?

i disagree with including wins or losses in this algorithm. the point is that it is a measure or individual performance and skill. why should you benefit/suffer based on the performance of people you did not choose to play with/did not choose to play with you?

if you want pug matchmaking, that is how it has to be.

i suggested this exact thing in like the 3rd paragraph of the OP; before all the math even. start the search out for players with +/- 100 rating. every 60s, expand the search by another +/- 100 rating. at some designated time period, say 5 minutes, the criteria is removed and will match the next available players for a match.

How do I generate so much internet nerd rage?

This is like my 4th insult stalker this week...