2023-09-17
Preamble
Bringing attention to issues is useful for game development, balance, and most importantly, fun. I do not think it is productive to just complain without offering solutions, so here is my input. My goal is to present a fair system for evaluating the effectiveness of squads, so we can have better matchmaking (MM). As game systems become more complex, solving them typically becomes intractable. This is nerdspeak for “the best build is impossible to determine”. If it’s not a difficult problem, it’s usually indicative of a very simple, or unbalanced system. The current meta (6k+ hangars) has a short list of 8 mechs(1). There is a wide variety of weapons being used, which I feel is a good sign for this game since the meta isn’t “solved”(2). Because both sides in a match don’t have identical pieces, a system like the ELO ranking in chess is problematic for matching people to have a “fair”(3) match. If the matching system is fair, then the outcome of each match is determined by the combination of map, mech abilities, weapons, and player skill. When I parsed matches, it was clear that the MM was probably not based on squad power (SP), rather it appeared to be based on the highest rank and rarity of equipment in the squad. I believe that we need a more sophisticated calculation than SP as the sum of all the parts of squads, which I call Match Squad Power (MSP).
Elephants in the room
MM is clearly broken, and also being abused. SP dropping should be trivial for the programmers to fix – just re-calculate MSP after players are queued for battle, and if no similar-powered players are available, fill the match with bots. If there’s an issue with token passing or queue priority, someone should hire me to fix it.
The problem with current MM
Matching by SP is not ideal because it’s easily abusable: A squad of 1-2 strong mechs will typically destroy a balanced squad of 5 with the same total SP.
A more reasonable assessment of the power of the squad would factor the power of the top mechs, and project that power across all 5 slots, even if empty. This would also encourage people to build more balanced squads instead of stacking everything into 1-2 mechs. My understanding of the MM from last season doesn’t change what I’m suggesting.
My suggestion, in loose terms, and using current power assignments:
Build bots as follows:
- Sort hangar of Player A in descending order of power
- Assign Weight = 9
- Player B1 = A1
- Player B2 = (Weight x A1 + A2) / (Weight + 1)
- Player B3 = (Weight x max(A2,B2)+A3)/(Weight + 1)
- Player B4 = (Weight x max(A3,B3)+A4)/(Weight + 1)
- Player B5 = (Weight x max(A4,B4)+A5)/(Weight + 1)
And then apply some minor variance of up to +/- 5%
Use anything similar to the current bot matching configuration, and adjust pilot, implants, and weapons to fit these numbers.
Matching players: Calculate MSP, and pair within 5%
- Sort hangar of Player A in descending order of power
- MSP = sum (A1x5 + max(A2,A1x.9)x4 + max(A3,A2x.9)x3+ max(A4,A3x.9)x42+ max(A5,A4x.9)x1) / 3
Rationale: assign weighted power, resulting in at least 90% of the power of the previous mech. The final minimum power, based on the highest power mech is 90%, 81%, 73%, and 66%
This intentionally punishes players with relatively weaker mechs on their squad, or empty mech slots. This should only be applied when players are expected to have 5 mech slots. The final division by 3 is simplified division by 15 (the sum of the multipliers) then multiplied by 5 to show
These systems rely on the current assignments of power, which I feel are flawed for the following reasons:
- Weapon power is based on rarity and energy, not actual damage, rate of damage, or effective range.
- This is further compounded by implants and pilots. If additional damage is sufficient to kill the opposing mech without a reload, this allows the winning mech to move to a better position instead of risking taking damage during a firefight.
- Mech power is based on the stats, and some of the power assignments are questionable, such as Panther.
I haven’t parsed matches vs bots at different loadouts to extrapolate what appears to be the current matching system this season (it seems very different than Season 22), and a new matching system has been mentioned, so I hope there is time for some of these points to be considered and possibly implemented.
(1) current meta, in alphabetical order:
-Aegis
+Bastion
+Eclipse
-Gatecrasher
+Orion
+Panther
-Redeemer
+Surge
The +/- indicates popularity
(2) Solved normally means a winning strategy that guarantees (has a high probability of) a win, loss, or draw for one of the players. In abstract terms, every strong mech has a viable counter in the meta, Since we see regularly new weapons and mechs in the game, the meta is expected to change over time. The fact there are active players with 9k+ meta builds from last year and still winning 80% of their matches is a strong indication the game is healthy and not solved.
(3) in Probability, “fair” means an equal chance of success. For example, a coin is considered fair if there is an equally likely chance of heads and tails.
edit 2024-05-19 I’m wondering if instead of sorting mechs in a hangar in descending order, it might be better to calculate the max SP from the current hangar; biggest mech with biggest pilot modifier, biggest 2 weapons (regardless of energy) including biggest pilot modifiers for that weapon, and then assign 100%, 90%, 81%, 73%, 66% with some variance as originally posted.