24 replies to this topic

[Nesutizale]

TL:DR

- People complain about T1 beeing to Meta.

- Solution: Drop Tiers and PSR and give each mech an indiviual BV (based in the indiviual build + Pilots W/L with that chassis) and let the MM sort teams by that.

- Advantage: More interesting matches and faster Q times.

- Possible option: One could make asymetrical matches and give Clantech back its "OMPF" because Clams would have a higher BV and so a team with a Clam would most likely have less members but even that could balance itself out by both sides haveing a similar number of Clams because the teams need similar BV to start a match.

--- The long version ---

Reading through here there where resently some people saying stuff like:
"T1 is basicly one Meta tactic and only meta builds" and that is terrible.

Thinking about it I noticed that in other games too. Like WoWs where the higher you go the more you have to stick to a certain meta or go back to harbour early.

Point is that is to be aspected. The higher the skill is in any given game the more optimised the player has play. That narrows down options to choose from.
Most appearant is this in stuff like fighting games... Street Fighter for example. Still I think MWO won't be an exception to this in the long run.
When you don't follow the comming tactics you will see the mechbay earlier then later.

Because of that I had just a thought. WoWs allready dropped the try to do a matchmakeing by skill and replaced it with a system that sorts people by ship used.
Since WoWs ships have mostly fixed loadouts, armor and stuff its pretty easy to balance those to each other...also they sometimes make questionable decissions...but that aside matches still work and come together quickly.

It also seams that people sort themself into their favorite Tier so the skill level sorting is done that way....mostly.

Comming back to MWO.
The idea is to drop Tier and PSR completly. Instead Mechs will have a Battlevalue applied.

This means that the mech will have a base value from its tonnage. This is then further modified by things like number and kind of hardpoints as well as their placement.

Quirks would also be taken into account.

Next you would get points for each piece of equipment, every weapon, heatsink etc.
There can also be some modifiers for maybe heat efficiency, mechspeed and so on. This will need some detailed work.
In the end each mech would have an individual BattleValue (BV).

For the purpose of matchmaking the MM would then go and try to sort players by BV and try to create teams following a 3/3/3/3 setup if possible. Primary focus will be BV.

Some might notice that this sound like the BV system from Battletechs Tabletop and yes its basicly that but we need to modifiy that a bit for MWO I think...but can be a starting point.

[Edit]
Added "(based in the indiviual build + Pilots W/L with that chassis)" to the TL:DR list

Edited by Nesutizale, 20 July 2020 - 10:01 AM.

[Scout Derek]

No.

[Nesutizale]

Why?

[Scout Derek]

Because that's completely unnecessary. any of it.

The main outlier will be and always will be the lack of skilled players in the queue constantly. You're taking into account that people are equally skilled in all chassis, regardless of their stats or weight with this new value system for mechs.

And by doing so you can basically turn matches into something Akin to the tonnage disparity we've been seeing lately.

[Vyx]

PGI has stats for every player specifically in every mech chassis they have ever played.

A BV could be derived for the individual player using their specific mech choice for every match.

This would reflect the value that the player - in that mech chassis - brings to the match.

Alternatively, you could generalize it a bit and work a player's BV from all their past performances in Light mechs, Medium mechs, Heavies, and Assaults.

Tiers could be categorized this way as well. For example, a specific player may be Tier1 in Heavy mechs, but only Tier2 in Assaults, and Tier3 in Lights and Mediums, based on their past performance and stats in those mech classes.

Obviously it would not take into consideration loadouts, but it would be better. Thoughts?

[OneTeamPlayer]

I think it's an interesting idea and i like the hustle.

I also think the last bit of major change added to MWO was originally internal, then crowd sourced at the very last minute, then implemented in a way that the creator themselves didn't ask for- which indicates a project that would require this level of coding to the game is unlikely to be implemented.

[Nesutizale]

@S. Derek
Yes I leave out the player and only take into account the chassis. The reason for this are:
- We have to few in numbers
- People notice now that those high skill based plays are not fun but more like work
- High BV mechs are most likely to be used by high skilled players because they know how to build a mech. In that regard knowledge/skill of the player is in some way taken into account.

Tonnage disparity should be less of a problem since the tonnage is taken into account and the MM will try to even out the BV of both teams.

Vyx, on 20 July 2020 - 07:44 AM, said:

PGI has stats for every player specifically in every mech chassis they have ever played.

A BV could be derived for the individual player using their specific mech choice for every match.

This would reflect the value that the player - in that mech chassis - brings to the match.

Alternatively, you could generalize it a bit and work a player's BV from all their past performances in Light mechs, Medium mechs, Heavies, and Assaults.

Tiers could be categorized this way as well. For example, a specific player may be Tier1 in Heavy mechs, but only Tier2 in Assaults, and Tier3 in Lights and Mediums, based on their past performance and stats in those mech classes.

Obviously it would not take into consideration loadouts, but it would be better. Thoughts?

Forgot about these stats....yes I can definitly see that beeing added to the calculation. Maybe as a modifier to the BV of the given mech. Good idea. Should even out the differance in skill even more.

[Brauer]

Scout Derek, on 20 July 2020 - 07:29 AM, said:

Because that's completely unnecessary. any of it.

The main outlier will be and always will be the lack of skilled players in the queue constantly. You're taking into account that people are equally skilled in all chassis, regardless of their stats or weight with this new value system for mechs.

And by doing so you can basically turn matches into something Akin to the tonnage disparity we've been seeing lately.

This idea also assumes that a mech is running a decent loadout. I've seen absolutely trash loadouts on some of the best chassis in the game. PSR already considers whether or not a player mostly run trash mechs ineffectively or meta mechs and drives wins.

[Nesutizale]

PSR takeing build into account? Don't make me laugh. Its only counting if you can do damage with it, nothing else. That dosn't mean that you bring a good build to the match. With BV you can at least to a certain degree quantify if the build is worth something.

The pilot is a force multiplier to that. Thats why I like the idea mentioned above to have a modifier by skill per chassis.

[Brauer]

Nesutizale, on 20 July 2020 - 08:04 AM, said:

PSR takeing build into account? Don't make me laugh. Its only counting if you can do damage with it, nothing else. That dosn't mean that you bring a good build to the match. With BV you can at least to a certain degree quantify if the build is worth something.

The pilot is a force multiplier to that. Thats why I like the idea mentioned above to have a modifier by skill per chassis.

Yeah, PSR takes into account the build because if the build is bad enough that you can't do damage or get wins you're less likely to go up in PSR. There are certainly ways that PSR or the match score system can be improved so that the ability to drive wins is measured much better, but it does account for it to a degree. A BV that accurately measured the ability of a mech to drive wins is not going to happen. You'd need to account for the strength of the chassis, the strength of the specific build including armor distribution and skills, and the player's playstyle.

Just deciding "Oh this player is in a Sleipnir, that's a top 10 assault so it gets a really high BV" doesn't account for if the player is running a high tier Sleipnir build (2xUAC10+2xUAC5, or something), if they're running a bad but not totally useless build (say 4xLBX5+ERMLs), or a complete dumpster fire of a build (4xHMG, 4xSLs, LAMS, giant standard engine).

Charting each player's ability per chassis seems likely to fail simply because people don't play all their mechs all the time, people run multiple builds and playstyles on specific chassis and individual mechs, and you're suggesting that they take a single variable (PSR) for each player and multiply it potentially by at least the number of chassis in the game (846 iirc). Seems like that's an awful lot of extra work and data behind the scenes. It's better to recommend improvements to make PSR better reflect a player's ability to drive wins.

[Nightbird]

PSR is terrible but Battle Value is even worse.

[C E Dwyer]

You going to put the money up to fund this thing that will never work.


Using wargamming as an example is also not a very wise choice, as BvP in WoW fails

[Nesutizale]

Brauer, on 20 July 2020 - 08:16 AM, said:

Yeah, PSR takes into account the build because if the build is bad enough that you can't do damage or get wins you're less likely to go up in PSR. There are certainly ways that PSR or the match score system can be improved so that the ability to drive wins is measured much better, but it does account for it to a degree. A BV that accurately measured the ability of a mech to drive wins is not going to happen. You'd need to account for the strength of the chassis, the strength of the specific build including armor distribution and skills, and the player's playstyle.

Just deciding "Oh this player is in a Sleipnir, that's a top 10 assault so it gets a really high BV" doesn't account for if the player is running a high tier Sleipnir build (2xUAC10+2xUAC5, or something), if they're running a bad but not totally useless build (say 4xLBX5+ERMLs), or a complete dumpster fire of a build (4xHMG, 4xSLs, LAMS, giant standard engine).

Charting each player's ability per chassis seems likely to fail simply because people don't play all their mechs all the time, people run multiple builds and playstyles on specific chassis and individual mechs, and you're suggesting that they take a single variable (PSR) for each player and multiply it potentially by at least the number of chassis in the game (846 iirc). Seems like that's an awful lot of extra work and data behind the scenes. It's better to recommend improvements to make PSR better reflect a player's ability to drive wins.

Don't we just see the PSR system totaly fail with LRM boat builds? When PSR is an indicator of people using good builds then everyone should be playing LRM boats and nothing else.

As for the BV composition, yes its exactly as you said, individual BV for each build. As I stated in the long version. Each mech would have its own indiviual BV based on the chassis, the hardpoints, placement of hardpoints, reactor, weapons, quirks...everything that makes this mech even down to the heat efficiency to create its own individual BV.
That way you can sort builds very well by its to aspected performance.

As for the Pilot. We can take his skill with the mech very easly into account. When you look at your stats page you will notice that PGI allready tracks your stats for each indiviual mechchassis. No extra work needed here. When you use the W/L ratio as a modifier in addition to the BV you get a pretty good picture of the mechs potential + how good the pilot is to win games in that specific chassie.

What I agree on it can't track if you use, for example, 3 different Warhamm -R configurations. On the other hand its very likely that people performe quite similar with each configuration.

[Brauer]

Nesutizale, on 20 July 2020 - 09:57 AM, said:

Don't we just see the PSR system totaly fail with LRM boat builds? When PSR is an indicator of people using good builds then everyone should be playing LRM boats and nothing else.

As for the BV composition, yes its exactly as you said, individual BV for each build. As I stated in the long version. Each mech would have its own indiviual BV based on the chassis, the hardpoints, placement of hardpoints, reactor, weapons, quirks...everything that makes this mech even down to the heat efficiency to create its own individual BV.
That way you can sort builds very well by its to aspected performance.

As for the Pilot. We can take his skill with the mech very easly into account. When you look at your stats page you will notice that PGI allready tracks your stats for each indiviual mechchassis. No extra work needed here. When you use the W/L ratio as a modifier in addition to the BV you get a pretty good picture of the mechs potential + how good the pilot is to win games in that specific chassie.

What I agree on it can't track if you use, for example, 3 different Warhamm -R configurations. On the other hand its very likely that people performe quite similar with each configuration.

1) LRM boats are not as reliable as other meta chassis. Plenty of people who have spammed LRM boats don't have terribly impressive stats.

2) There's basically no way that you can set a single battle value for every single build out there. The number of possible builds is gigantic so even creating a BV for every one is a ridiculous task, not to mention that the BV will simply not be 100% accurate. There will be builds that are vastly under and overestimated and it'll be a complete mess. It's much better to just look at how often people win, or how high of a match score they accumulate (despite all the flaws in match score).

3) Seems there would be extra work because the mech stats would have to be used somehow in matchmaking rather than just displayed occasionally when someone looks them up. The matchmaker has already proven to be overburdened when balancing the variables it has at hand, it doesn't need to have another variable mashed into it.

[Willard Phule]

In theory, the idea is interesting. Imagine what the BV of an LRM Atlas is. Here, the loadout doesn't matter. An Atlas is an Atlas, no matter how you build it. BV was designed to include the loadout on TT.

Problem is, taking theory to actual application. I doubt it's within PGI's limitations to do something like this. And even if they tried, I think it's pretty safe to assume that it would be just as cockeyed as anything else they've done.

[Nesutizale]

Brauer, on 20 July 2020 - 10:39 AM, said:

1) LRM boats are not as reliable as other meta chassis. Plenty of people who have spammed LRM boats don't have terribly impressive stats.

2) There's basically no way that you can set a single battle value for every single build out there. The number of possible builds is gigantic so even creating a BV for every one is a ridiculous task, not to mention that the BV will simply not be 100% accurate. There will be builds that are vastly under and overestimated and it'll be a complete mess. It's much better to just look at how often people win, or how high of a match score they accumulate (despite all the flaws in match score).

3) Seems there would be extra work because the mech stats would have to be used somehow in matchmaking rather than just displayed occasionally when someone looks them up. The matchmaker has already proven to be overburdened when balancing the variables it has at hand, it doesn't need to have another variable mashed into it.

1) LRM boats may not be reliable but they are also not off the table. In that regard PSR dosn't care if you reach T1 in a LRM boat or in a Sleipnir build with 2xUAC10+2xUAC5. The only thing PSR cares about is your damage output. Not how you archive it or if that contributed to the team or if its efficient.

In a BV system you could assign LRMs an BV value that reflects their ineffectiveness. For example by reduceing their BV so the team has room for a much better meta build with a high BV or by makeing LRMs so BV inefficent that takeing them ruins the build. Either way you can balance for the overall team BV or make people take LRMs less.

2) I have a guess what your line of thought is. You think of haveing a list of all the thousands (?) of different combinations in a list that someone has to type. Should that be it...yes that would be terrible.

There is an easier solution to it, also its still on the work intensive side, no illusions there.
- Create a list of the base value of each chassie (this value is based around tonnage, # of hardpoints and quirks)
- Give every equipmentpart its own BV

When you now build your mech the BV of your mech is calculated like your tonnage. You drag something into your mech and the computer adds the BV to the basis. Like the tonnage of a ML is added to the current tonnage of your mech.
Your W/L modifier is at this step also taken into account.
When you hit save the BV of that indiviual configuration is save clientside.

3) The data is there. What you can do to save time is to have the BV, that is calculated in the mechbay while you build your mech as discribed in point 2, add the chassis W/L modifier to the mechs BV right at that step and just save the value clientsided.

When you start to look for a match the presaved BV from you PC is then send to the MM. He just gets a number that he need to compare, no calculations needed.

Willard Phule, on 20 July 2020 - 10:57 AM, said:

In theory, the idea is interesting. Imagine what the BV of an LRM Atlas is. Here, the loadout doesn't matter. An Atlas is an Atlas, no matter how you build it. BV was designed to include the loadout on TT.

Yes BV wasn't the perfect system last time I used it in TT but matches back then where pretty balanced, even though the teams where hardly matched by skill.
So the indiviual players skill was less important then the BV to my experiance.

Quote

Problem is, taking theory to actual application. I doubt it's within PGI's limitations to do something like this. And even if they tried, I think it's pretty safe to assume that it would be just as cockeyed as anything else they've done.

Agreed. This is more on the line of theorie crafting then PGI doing it. If anything at all I say the best bet is to push PGI for an MWO-Open approach and let the community handle the figuring out of the system and PGI only implements it when its all done.

[Yiryi-Sa]

Brauer, on 20 July 2020 - 08:16 AM, said:

...or a complete dumpster fire of a build (...LAMS...)

Hey now! Leave the LAMS out of it.

[Brauer]

Nesutizale, on 20 July 2020 - 11:52 AM, said:

1) LRM boats may not be reliable but they are also not off the table. In that regard PSR dosn't care if you reach T1 in a LRM boat or in a Sleipnir build with 2xUAC10+2xUAC5. The only thing PSR cares about is your damage output. Not how you archive it or if that contributed to the team or if its efficient.

In a BV system you could assign LRMs an BV value that reflects their ineffectiveness. For example by reduceing their BV so the team has room for a much better meta build with a high BV or by makeing LRMs so BV inefficent that takeing them ruins the build. Either way you can balance for the overall team BV or make people take LRMs less.

I said lrms were less reliable. That means I think you can drive wins better, get a higher KDR, and get a higher average match score with direct fire builds than with lrms. I didn't simply mean they kill inefficiently, I was also referring to how they perform poorly on some maps and have much higher match to match variance with a much lower floor than other builds.

Quote

2) I have a guess what your line of thought is. You think of haveing a list of all the thousands (?) of different combinations in a list that someone has to type. Should that be it...yes that would be terrible.

There is an easier solution to it, also its still on the work intensive side, no illusions there.
- Create a list of the base value of each chassie (this value is based around tonnage, # of hardpoints and quirks)
- Give every equipmentpart its own BV

When you now build your mech the BV of your mech is calculated like your tonnage. You drag something into your mech and the computer adds the BV to the basis. Like the tonnage of a ML is added to the current tonnage of your mech.
Your W/L modifier is at this step also taken into account.
When you hit save the BV of that indiviual configuration is save clientside.

That does not work because a piece of equipment does not have a static value. The value of a particular piece of equipment depends on your overall build. This is true both explicitly as in the case of a targeting computer (which behaves differently on different builds) or an active probe (which is wasted tonnage on many builds, but essential on others), and because the overall effectiveness of the equipment is more than the sum of its parts. None of what you described deals with the issue that BV would be massively inaccurate, implementation issues aside, this inaccuracy is enough reason to shelve the idea. It'd be far better to move away from arbitrary values like PSR and BV and toward things like WLR which are direct measures of effectiveness.

Quote

3) The data is there. What you can do to save time is to have the BV, that is calculated in the mechbay while you build your mech as discribed in point 2, add the chassis W/L modifier to the mechs BV right at that step and just save the value clientsided.

When you start to look for a match the presaved BV from you PC is then send to the MM. He just gets a number that he need to compare, no calculations needed.

The matchmaker is struggling enough as is and does not need another variable to muddy things up and either increase wait times, reduce how well it can pursue other priorities, or make worse matches.

Personally, I think the best approach would be to split groups into their own queue again, add an option for solos to opt into group queue, and then see if it would be possible to implement a more WLR focused approach. Using the specific mech WLR would be best most likely, but I don't think that'd be practical.

[Nesutizale]

All equipment has lots of static values like tonnage...or is your laser lighter because you equip an TC?
Adding a TC will be taken into account, also it has a multiplayer instead of a simple addition. (if this continues I have to post the entire rules and calculations and values *sigh* )

W/L is a system that was made for 1:1 and fixed teams that allways play together in their specific league. It works like a charm for Solaris and FW. Takeing this to play with randoms...not quite to my experiance.

I have recorded all my matches since the reset, around 50.
When we go by W/L only I should went up in Tier but I devinitly shouldn't. My average matchscore and damage output are clearly saying T3 at best.

The most likely explenation is that I was lucky to be carrierd. At this point I get the usual "Yes but in a hundret games it will level out". Well if it levels out it dosn't prevent beeing carried or that one can so easly manipulate W/L.

The nice point about the BV system is that it covers several points at once.
- It takes the tonnage you bring to the field into account. So we can reduce the amount of tonnage unbalance we see now.

- It takes into account the build you bring. Is it a meta build or scrap? That way both sides would field a similar number of meta and scrap builds.

- With the W/L per chassis modificator you can will also even out skill of pilots and it gives lights a bit more value as a good light pilot will be higher placed in value. So the composition of teams is more similar.

- Smurfs and wanting to gameing of the system will be hard. You can loose matches on purpose or create a new account. The meta build you play will still put you in a spot that is pretty close to where you should be. You would have to ruin the experiance for yourself realy hard to brake it like going without armor or weapons. PS: BV also accounts for that and reduces your BV, so it frees up BV for others to take. You would just go around and die a lot. No seal clubbing because either there is another Meta player you have to look out for on the other side or you restricted yourself so much that you are not a problem for even a new player.

- The MM has a much bigger pool of players as there is only one pool where it matches the BV on both sides to be similar.

- As people noticed gameplay at high tiers becomes stale at its a few tactics and builds that are valid. Since teams are placed to have similar BV you can put people of much more places together. It won't prevent stomps, nothing can. Still games should be more varied then just T1 matches.
Personaly I think QP should be more focused on variaty and fun games while people that realy want a more competetive experiance should have their comp Q.

Edited by Nesutizale, 20 July 2020 - 02:31 PM.

[Brauer]

Equipment values are not static. Stick a TC1 on an mg piranha, it's a net loss to the build, not a benefit. Stick the same TC on a grasshopper and it offers some utility.

BV is simply an overcomplicated system that won't even work on its own terms. No to BV.