64 replies to this topic

[Wingbreaker]

Hoax415, on 30 August 2014 - 06:01 AM, said:

No you don't understand a system where a Kit Fox is the same for matchmaking as a Locust is clearly superior to anything else. Wingbreaker says so with zero evidence or argument or data to back him up so it must be true.

The 14 SM-SL vs 1 AS7-D exercise is nothing new, it has been used to show how bad BV is for years.

The honest truth is that the Atlas probably wouldn't even last 5 rounds. He will effectively have the opportunity to kill 1 every round. If he's lucky. These are moving at 216, after all. Even without calculating the pilots, and assuming a maximal atlas pilot, the simple numerical chances of engine criticals from that many small lasers being fired means that the Atlas is dead.

Again, BattleValue suggests that larger weapons are worth more. In reality, large amounts of smaller weapons are remarkably more useful simply because of the percentages possible.

Let's imagine this in relation to MWO. A Gauss Rifle is worth, by TT values, 320 +40 per ton of ammo. This means a standard gauss loadout equals 440 points. At this rate, a single gauss rifle is worth ~9.5 medium lasers. So for the price of 'that guy' that brought a GR, you'll have outfitted an entire Jenner and a half.

Is a single Gauss Rifle worth 9 medium lasers? Hell no.

Have fun trying to implement this in any sort of logical or sane manner. Are you going to up the costs of smaller weapons? Are you going to put a much heavier weight on the piloting factors (elo)? Either way, it means you're balancing one good player against many mediocre with weapons loadouts that make them far more reliable.

Edited by Wingbreaker, 30 August 2014 - 06:14 AM.

[Jonny Taco]

I've got to agree with you about the elo system in this game Flash Frame... It honestly feels like doing well in the game results in a punishment via being matched with paint chip eaters to come to a "balanced" team elo...

A mm should NEVER EVER EVER build games around the concept of particular players completely carrying the team... It should be based around grouping people of similar skill levels with and against each other...

[kailii]

lartfor, on 30 August 2014 - 06:03 AM, said:

kaili, i just wanted to highlight your point about dropping the game back to 8v8. 8v8 is easier for the match maker, more in line with the vast majority of maps, and imo provided a much better gameplay experience allowing exceptional (good or bad) pilots actions to have a much heavier weight on the outcome of the game...

Better yet! Some maps should be 4v4, some should be 8v8, some should be 12v12.

The idea was for the Matchmaker to start with 8 Mechs each, but to fill out the teams to 12 vs. 12 (or say 9 vs. 12) before the game starts.
But i fully agree, varied team sizes based on map design (or even randomly!) would be a great feature.

[Jonny Taco]

Guess i miss read then, my bad.

After looking through your post again (sorry it's early) you've really put together something reasonable.

[J0anna]

I agree that adding BV will help balance matches, it should be combined with elo (which should not be based solely on wins and losses, but rather should consider w/l, k/d, assists per match, and damage per match). Otherwise a person with 50 drops under his belt would look the same as someone with 10,000 drops.

I like the 10 v 12 idea, especially since IS will be 3/3/3/3, you could expect clans will be 3/3/2/2 or even 3/3/3/1. Adding in BV would help, but the BV must be based on MWO weapon values, not TT values. The matchmaker would then pick a clan team first, say setting up 3/3/2/2 and mix IS teams to counteract that. It would help games where one team seems to get a mismatch in missiles (they would have higher BV) or ecm (which would have a very high BV). Would it be perfect? No, nothing is perfect, but it should certainly be better.

[van Uber]

Wingbreaker, on 30 August 2014 - 06:06 AM, said:

Have fun trying to implement this in any sort of logical or sane manner. Are you going to up the costs of smaller weapons? Are you going to put a much heavier weight on the piloting factors (elo)? Either way, it means you're balancing one good player against many mediocre with weapons loadouts that make them far more reliable.

This.

When the Locust is considered the "worst Mech in the game", the solution is not to introduce a convoluted, arbitrary BV-system on top of what we have. The solution is to buff the Locust, either by the current values or by introducing chassis specific quirks to make it more competitive.

[kapusta11]

van Uber, on 30 August 2014 - 07:14 AM, said:

This.

When the Locust is considered the "worst Mech in the game", the solution is not to introduce a convoluted, arbitrary BV-system on top of what we have. The solution is to buff the Locust, either by the current values or by introducing chassis specific quirks to make it more competitive.

Do you honestly believe PGI is capable of/willing to fix things like obsolete or overpowered mechs/weapons/modules/game mechanics?

[InspectorG]

If we needed a BV i would urge simplicity.

A basic tier ladder for chassis and builds and weight them accordingly.

Something like a Mad Cat may be worth 2 Thunderbolts? Then calculate for player skill which i would break down into 3 groups: green, vet, pro.

May how often a player drops with others and their rank. Most will level out to vet level. Pro will likely group drop with pro for training and comp purposes. Constant green is either a very casual player or a youngster.

Mech tier weight/ player rank in relation to tonnage for the match, something like that.

I think it needs to be simple to not bog the MM or have too many variables for PGI to drown in.

Caveat:: no system will be perfect, why not go for the simplest that brings the relatively happiest results?

[STEF_]

Wingbreaker, on 30 August 2014 - 06:06 AM, said:

The 14 SM-SL vs 1 AS7-D exercise is nothing new, it has been used to show how bad BV is for years.

The honest truth is that the Atlas probably wouldn't even last 5 rounds. He will effectively have the opportunity to kill 1 every round. If he's lucky. These are moving at 216, after all. Even without calculating the pilots, and assuming a maximal atlas pilot, the simple numerical chances of engine criticals from that many small lasers being fired means that the Atlas is dead.

Again, BattleValue suggests that larger weapons are worth more. In reality, large amounts of smaller weapons are remarkably more useful simply because of the percentages possible.

Let's imagine this in relation to MWO. A Gauss Rifle is worth, by TT values, 320 +40 per ton of ammo. This means a standard gauss loadout equals 440 points. At this rate, a single gauss rifle is worth ~9.5 medium lasers. So for the price of 'that guy' that brought a GR, you'll have outfitted an entire Jenner and a half.

Is a single Gauss Rifle worth 9 medium lasers? Hell no.

Have fun trying to implement this in any sort of logical or sane manner. Are you going to up the costs of smaller weapons? Are you going to put a much heavier weight on the piloting factors (elo)? Either way, it means you're balancing one good player against many mediocre with weapons loadouts that make them far more reliable.

Not agree.
The fact is that in Mwo a jenner is chased by others lights or medium. For this reason in BT exist BV AND class weight, such as IS lances, and clanner star system.
You cannot compare mechs with different classes, .... and, of course that jenner must to go close, if it manages (chased by others).

The core of the problem is that the actual elo system works ESACTLY contrariwise to common sense.
The solution of course is simple: match-maker should associate players with similar elo.
So we would have fair team.
Good players would enjoy a competitive match.
Noobs enjoy learning without being stomped, or irritating good players.
What do we have now? Good players with headhache. Noobs stomped. Too few people enjoying the game.

[FDJustin]

Wingbreaker, on 30 August 2014 - 06:06 AM, said:

The 14 SM-SL vs 1 AS7-D exercise is nothing new, it has been used to show how bad BV is for years.

The honest truth is that the Atlas probably wouldn't even last 5 rounds. He will effectively have the opportunity to kill 1 every round. If he's lucky. These are moving at 216, after all. Even without calculating the pilots, and assuming a maximal atlas pilot, the simple numerical chances of engine criticals from that many small lasers being fired means that the Atlas is dead.

Again, BattleValue suggests that larger weapons are worth more. In reality, large amounts of smaller weapons are remarkably more useful simply because of the percentages possible.

Let's imagine this in relation to MWO. A Gauss Rifle is worth, by TT values, 320 +40 per ton of ammo. This means a standard gauss loadout equals 440 points. At this rate, a single gauss rifle is worth ~9.5 medium lasers. So for the price of 'that guy' that brought a GR, you'll have outfitted an entire Jenner and a half.

Is a single Gauss Rifle worth 9 medium lasers? Hell no.

Have fun trying to implement this in any sort of logical or sane manner. Are you going to up the costs of smaller weapons? Are you going to put a much heavier weight on the piloting factors (elo)? Either way, it means you're balancing one good player against many mediocre with weapons loadouts that make them far more reliable.

It's definitely never an easy thing to come up with. Best you can do is hire a doctor of maths, give them the data and hope they can come up with a reasonable value.
So does the system just give the gauss to much value, or the SL's too little? Or is the problem boating?
It it's boating, I suppose the system could add a cumulative penalty for each additional weapon. But what about when you reach the ghost heat point? Reduce the cost again?

It would definitely, always be gameable, and a sytem that takes into account the power of boating leaves itself open for builds that are just slightly less efficient for dramatic point savings.

Let's say there is an accumulative 10% boating penalty up to the ghost heat, then it's reduced to 5% per each extra.
MG is worth 10 points
MG Ammo is worth a flat 1 point / ton.
SL is worth 10 points.
ML is worth 50 points.
LL is worth 120 points.
Heatsink is worth 10 points.

A 6 ML Jenner would cost 50, + 55, + 60.5, + 66.55, + 73.21, + 80.53, or 385.79 points.
An 8 ML Firestarter would cost 50, + 55, + 60.5, + 66.55, + 73.21, + 80.53, + 84.56, + 88.78, or 559.13 points.
Even with the reduced inflation after ghost heat, the Firestarter doesn't being that much more 'valuable' than the Jenner, so changes it just a bit and probably winds up with a better mech that's much closer in value.
A 6ML, 2 SL Firestarter with an extra heatsink would cost 50, + 55, + 60.5, + 66.55, + 73.21, + 80.53, + 10, + 11, + 10 or 416.79 points.

A spider with 1 LL, 4 MG, 2.5 tons of ammo would be worth 150, + 10, + 11, + 12.1, + 13.31, + 2.5, or 168.91 points.
A 4 ML, 4 MG Ember with 2.5 tons of ammo would be worth 50, + 55, + 60.5, + 66.55, + 10, + 11, + 12.1, + 13.31, + 2.5, or 280.96 points.
Once again, the Ember pilot decides to shave off some points, so drops one ML, replaces it with a SL, and upgrades to 3 tons of ammo. Now it's worth 224.91 for being just a little less efficient, and a lot closer to being dropped with that spider.

I expect the value of the weapons to be much closer actually, but I chose those values to show how a few little tweaks here and there can, and will be used to shave off large percentages of BV without shaving off large percentages of actual combat worthiness.

Ideally, you would have mechs with the same weight having similar BV's, with more emphasis put on weapons than other equipment.. Or we could just have the match maker try and equalize the tonnage on each side, because that's probably going to be just as close to balanced at the end of the day.

[kapusta11]

MM and ELO works mostly fine, there are however some issues with ELO not knowing when exactly to stop rising difficulty resulting in lose and then, as the result of ELO dropping to low, win streaks. What it does not account for is what mech/loadout/modules you use so if you play in meta Highlander and switch to non elited Awesome MM demands you to play as good. For this very reason, to balance things MM cannot, BV is good.

Edited by kapusta11, 30 August 2014 - 08:15 AM.

[BLOOD WOLF]

Stefka Kerensky, on 30 August 2014 - 07:37 AM, said:

The core of the problem is that the actual elo system works ESACTLY contrariwise to common sense.
The solution of course is simple: match-maker should associate players with similar elo.
So we would have fair team.
Good players would enjoy a competitive match.
Noobs enjoy learning without being stomped, or irritating good players.
What do we have now? Good players with headhache. Noobs stomped. Too few people enjoying the game.

[Tsig]

Stefka Kerensky, on 30 August 2014 - 07:37 AM, said:

Not agree.
The fact is that in Mwo a jenner is chased by others lights or medium. For this reason in BT exist BV AND class weight, such as IS lances, and clanner star system.
You cannot compare mechs with different classes, .... and, of course that jenner must to go close, if it manages (chased by others).

The core of the problem is that the actual elo system works ESACTLY contrariwise to common sense.
The solution of course is simple: match-maker should associate players with similar elo.
So we would have fair team.
Good players would enjoy a competitive match.
Noobs enjoy learning without being stomped, or irritating good players.
What do we have now? Good players with headhache. Noobs stomped. Too few people enjoying the game.

I see the words, but I can't seem to put them into anything resembling complete sentences.

[FDJustin]

Tsig, on 30 August 2014 - 08:08 AM, said:

I see the words, but I can't seem to put them into anything resembling complete sentences.

Translation: I disagree. You can't compare a Jenner to an Atlas in a vacuum, since there are 12 v 12 teams, and that Jenner shouldn't have an opportunity to duel an Atlas unabated. SL's are short ranged garbage which force high risk engagements.

The matchmaker doesn't pit every player with others of reasonably similar skill levels, so you find new players overwhelmed by experienced players who have fully kitted out mechs, while more experienced players either find it less challenging, or q.q over having inexperienced players thrown on their team to 'balance' it out.


I'm not sure what he means by the weight class used as a balancing factor though.

[Xarian]

Tonnage matching works much better. For mechs that are the best or worst of their tonnage, they can get added quirks to offset the balance.

For example, Awesomes can get heat bonuses (i.e. less heat generation) for PPCs, with one variant instead getting that bonus for LRMs. Timberwolves stay as they are, but Summoners get a big bonus to jump jet thrust to make up for their other weaknesses. Warhawk gets a quirk that makes targeting computers more effective. Victor gets its torso speeds un-nerfed, Griffon gets a jump jet bonus, Locust gets a huge damage reduction to falling damage, etc.

[STEF_]

Tsig, on 30 August 2014 - 08:08 AM, said:

I see the words, but I can't seem to put them into anything resembling complete sentences.

Come on, use intuition!

Edited by Stefka Kerensky, 30 August 2014 - 08:48 AM.

[Xarian]

FDJustin, on 30 August 2014 - 07:44 AM, said:

It's definitely never an easy thing to come up with. Best you can do is hire a doctor of maths, give them the data and hope they can come up with a reasonable value.

There's not enough data for a "doctor of maths" to work with unless you want to answer very simple questions. For example: is a medium mech pilot more likely to win or lose using 1 ERPPC?

You can't use a simple one-number system to determine value. Why? because BV is multidimensional - in other words, your mech's value depends on your opponents' mechs' values. A LRM boat, for example, may have a high BV against normal opponents but low BV against teams with ECM.

If you want, you could probably break it down into a few parameters... role value (RV)? Brawl Strength vs Support Strength vs Sniper Strength vs Artillery Strength

[STEF_]

FDJustin, on 30 August 2014 - 08:24 AM, said:

Translation: I disagree. You can't compare a Jenner to an Atlas in a vacuum, since there are 12 v 12 teams, and that Jenner shouldn't have an opportunity to duel an Atlas unabated. SL's are short ranged garbage which force high risk engagements.

The matchmaker doesn't pit every player with others of reasonably similar skill levels, so you find new players overwhelmed by experienced players who have fully kitted out mechs, while more experienced players either find it less challenging, or q.q over having inexperienced players thrown on their team to 'balance' it out.


I'm not sure what he means by the weight class used as a balancing factor though.

Thank you Justin, you're very kind.

Btw, I wanted to say that it is possible to compare BV between the same weight class, but not, for example, between an Atlas with a jenner. For this reason BT uses BV AND a lance built system.

[Quick n Fast]

why not base it on, hr's played + agv% of shots landed + W/L... the first 2 are exp/skill based n w/L as luck factor.

Any good pilot knows its not the mech thats dangerous its the pilot.

[FDJustin]

Xarian, on 30 August 2014 - 08:45 AM, said:

There's not enough data for a "doctor of maths" to work with unless you want to answer very simple questions. For example: is a medium mech pilot more likely to win or lose using 1 ERPPC?

You can't use a simple one-number system to determine value. Why? because BV is multidimensional - in other words, your mech's value depends on your opponents' mechs' values. A LRM boat, for example, may have a high BV against normal opponents but low BV against teams with ECM.

If you want, you could probably break it down into a few parameters... role value (RV)? Brawl Strength vs Support Strength vs Sniper Strength vs Artillery Strength

Exactly. Best you can do is aggregate all that data sitting on our profiles, assign a general value to each metric (10 damage is worth this much. Burn time modifies it by this much. Weight and crit space removes that much. Bonus critical chance or damage is worth blah.) and take an educated guess.

There are already far too many variables to take into account player skill, or roles. There's no need for role value because the missiles, ecm, tag, ams, etc. Would have it's own BV.

I think it should just try to equalize tonnage on each side, that'll probably bring you closest without too much gaming. All the equipment should have a very close value just by how it's made to balance within an individual mech already.