Forged Alliance Forever

Re: Scientific Balancing?

2018-10-13T10:36:24+02:00

You can, technically, consider the entire FAF unit database as a "gene" for some sort of an optimization algorithm with a %ratio of wins by faction to be used as a goal function. After a few (thousand) rounds of this the game should be pretty balanced.

Statistics: Posted by uzurpator — 13 Oct 2018, 10:36

Re: Scientific Balancing?

2018-10-13T05:05:49+02:00

(A) and (C) are already collected and have been shown in Softly’s FAF data thread: /viewtopic.php?f=2&t=13449

I wouldn’t be surprises if he could also get (B) if he really wanted to.

You don’t need to collect data to find what the most popular and effective units are at tech levels. You just naturally learn that stuff by getting good at the game and the balance team is quite full of dudes that can play at a competent level. I mean this might look like objective data but you’re still going to be subjectively determining what the “true” death rate of a unit “should” be.

moses_the_red wrote:
I'd like to see an expansion personally, I haven't explored either modding myself or what mods exist out there but my understanding is that it IS possible to create new units and features.

Using something like kickstarter to fund further development might make such ideas viable.

If you mean trying to kickstart some mod that creates units, no one would stop you. If you mean kickstarting to get new units into FAF; I doubt that will ever happen. T3 MAA is up there in top 5 most controversial FAF changes and that unit is probably as uncontroversial as it comes in implementing a new unit to fufill a purpose.

Statistics: Posted by FtXCommando — 13 Oct 2018, 05:05

Re: Scientific Balancing?

2018-10-13T04:11:48+02:00

This sounds like a good idea, but it also requires a very expensive skillset, people that are good at datamining make the big bucks doing it professionally.

That said, perhaps a kickstarter for something like this would be worthwhile, if the powers that be would allow such a thing. I'd like to see an expansion personally, I haven't explored either modding myself or what mods exist out there but my understanding is that it IS possible to create new units and features.

Using something like kickstarter to fund further development might make such ideas viable.

Statistics: Posted by moses_the_red — 13 Oct 2018, 04:11

Scientific Balancing?

2018-10-12T04:40:10+02:00

Is it possible to balance the game (FAF) scientifically, that is to say empirically, from real game outcomes?

Firstly, a game has a meta-balance; the way it is balanced tactically and strategically by primary concepts and design decisions. I am not talking about that level of balance. Rather, I am talking about the balance of the game (FAF) as it exists now. Please bear with me if these are not all new ideas. It might also help to methodically list existing ideas.

The first thing to do is collect and analyze real game data. I assume this is being done now. The place to start (but not finish) would be to analyze ladder 1v1 games.

(a) What are the faction win rates?
(b) What are the faction win rates broken down by map type (size by terrain Type)?
(c) What are the faction win rates broken down by rating categories?

Then;

(d) What are the build stats and death rate stats for each combat unit type (including fixed weapons) analysed overall and by tier?

From (d) we can deduce some things like (i) the most popular units at each Tech tier by faction, (ii) the most popular units for high rated players at each Tech tier by faction (on the assumption that high rated players know and use the good units well), (iii) The most effective units, their survival numbers and kill numbers compared to their production numbers and so on.

Starting to look at these stats might help us to home in on OP and UP factions, OP and UP units and other aspects of balance. Of course, it's likely none of these ideas are new but are they being looked at? It just takes a bit a data mining and spread-sheeting I guess.

Statistics: Posted by Crazy Cossack — 12 Oct 2018, 04:40