14 replies to this topic

[Torage]

Tournament recommendations

PROBLEM

The scoring system

Assists giving 15 points make it extremely important that you damage every 'Mech on the other team, failing to do so will ruin your score for the match. This means that if someone on your team pits a light as it runs past with a lucky/skilled x2 AC/20 shot before anyone could poke them, that match is ruined for everyone ells.

Here is an example of the problem:

2 'Mechs on the same team, the team wins and both survive.

'Mech A "pokes" each 'Mech on the other team and does 150 damage in the process.
(15*8)+(150/30)+20+20 = 165 points

'Mech B battles it out and scores 3 kills and 2 assists dealing 600 damage.
(20*3)+(15*2)+(600/30)+20+20 = 150 points

See what I mean?

Now I understand they can't lower the points gained from damage or assists because it leads to "damage hogging" and "kill stealing" and I DO NOT RECOMMEND BOOSTING POINTS FROM DAMAGE AND KILL SHOTS but something that will fix both problems in one move.

SOULITION
Make each part of a 'Mech worth 5 points (totaling 40 pts)
assists will still be 15 points
Examples:
'Mech A blows the an enemy 'Mechs arm off and Mech B cores them.
'Mech A gets 15 from the assist and 5 for the arm = 20
'Mech B gets 35 points for the kill (5 pts for each remaining part)
'Mech C finds a Jenner with one leg and no arms/side torsos and puts the poor guy out of his misery for 15 points (Leg+CT+head)

The numbers can be changed but hopefully this presents a good idea of the theory. Hopefully the scoring system is set up that this kind of information can be used and not just what shows up on the scoreboard (although the page where it shows how many C bills you earned shows this information).

Come to think of it the C-Bill earning system is a much more accurate gauge of match performance than the current tournament point system (obviously minus cadet/hero/premium bonuses).

Other than that I hope you offer some real prizes in the future. Pocket change in MC is kind of meh for your top ranking players but I understand it's still in the testing phases (not saying to give more MC necessarily but like a badge or something that shows you're a boss).

[Soy]

Ok basically right now if 2 lights go gank a light to start out a round right now,

might as well quit match and drop new mech.

[Odnir]

That's kind of an interesting idea, but I don't think it will change things too much. For example if a mech is fully intact except for the CT, you can get in the last few taps and get the full 40 points. 40 points is a lot for a kill, and a lot for a lucky shot. Also hello. We were 1 and 2 on the Dragon board yesterday but have since been bumped down a couple spots.

If there's anything I would want to see different in a future tournament it's the intangibles that contribute to a win. I noticed yesterday I was playing differently in order to try to score well on the leaderboard. I was a more aggressive attacker than I usually am. I'd stay on a target till it fell, even when a team mate could have done the job as well as I, and when I was needed elsewhere. But it's hard to score for that. It's not easy to measure how many times you successfully played squirrel. Or how much damage you successfully soaked away from the team via avoidance / great twisting. And all those other little bits that measure your true "team value". (edit: which I think made the game less fun - and it's a weird thing to me to play a game for a reason other than fun. I've stopped playing this account for the weekend and just hope when the dust settles I am still in the top 5)

But the title of the competition is "Heavy vs. The World" and not "Great Team Players" so I'm not really complaining. I think it's actually a pretty decent system.

I too would like to see a different reward structure though. Descending prize values would be more fun, i.e First place 2000, second place 1000, 3 place 500, 300, 200, etc. maybe even down to a little 50MC prize just for being on that top 15.

Edited by Odnir, 23 June 2013 - 10:00 AM.

[Torage]

Odnir, on 23 June 2013 - 09:57 AM, said:

That's kind of an interesting idea, but I don't think it will change things too much. For example if a mech is fully intact except for the CT, you can get in the last few taps and get the full 40 points. 40 points is a lot for a kill, and a lot for a lucky shot. Also hello. We were 1 and 2 on the Dragon board yesterday but have since been bumped down a couple spots.

If there's anything I would want to see different in a future tournament it's the intangibles that contribute to a win. I noticed yesterday I was playing differently in order to try to score well on the leaderboard. I was a more aggressive attacker than I usually am. I'd stay on a target till it fell, even when a team mate could have done the job as well as I, and when I was needed elsewhere. But it's hard to score for that. It's not easy to measure how many times you successfully played squirrel. Or how much damage you successfully soaked away from the team via avoidance / great twisting. And all those other little bits that measure your true "team value". (edit: which I think made the game less fun - and it's a weird thing to me to play a game for a reason other than fun. I've stopped playing this account for the weekend and just hope when the dust settles I am still in the top 5)

But the title of the competition is "Heavy vs. The World" and not "Great Team Players" so I'm not really complaining. I think it's actually a pretty decent system.


I too would like to see a different reward structure though. Descending prize values would be more fun, i.e First place 2000, second place 1000, 3 place 500, 300, 200, etc. maybe even down to a little 50MC prize just for being on that top 15.

The idea would be that with each component on a 'Mech being worth points that there would be more points to go around so if you miss as assist your not SOL, you can make up the lost points elsewhere. But I can see the flaw you point out here. I think that the earning of C-bills might be best, you get C-bills for kills and wining and assists but also for component destruction and spotting assists.that greatly help your team. The only adjustment it might need I think is that I'm not sure you are penalized for dying (you should be anyway if you aren't already that I don't know of imo).

The more i think about it the more sense it makes to use the scoring system they have already fine tuned with the C-bills.

EDIT: They should also award points when your 'Mech does a flip after you die.

Edited by Torage, 23 June 2013 - 02:11 PM.

[Ozasuke]

While I agree with the problem, I disagree with the original post's solution. Scoring should be normalized, but this can't be entirely possible without more variables to extrapolate data effectively.

However, with the data that is available, it's still possible to strike a balance between kills & assists, damage
dealt, and win rate & survivability. I've restructured the Heavy tournament formula using the three aforementioned categories:

(Kills x 3.5 + Assists x 1.75) x (Win x 0.5 + Survived x 0.5 + 1)
+ (Damage Dealt - Team Damage) x (Kills x 2 + 1)/(Assists / 2 + 1) x (Assists + 1)/(Kills + 1) / 37.5

The key is the relationship between kills & assists and the damage dealt across those kills and assists. It's designed to mitigate kill stealing and assist-whoring while normalizing points awarded by damage dealt.

Let's look at some scenarios (assuming max wins and survivals and Medium-class 'mechs):

• A game with 8 kills dealing 800 damage will yield 96 (56+40) points.
  
• A game with 8 kills dealing 100 damage (kill stealing) will yield 61 (56+5) points.
  
• A game with 8 assists dealing 100 damage (assist-whoring) will yield 33 (28+5) points.
  
• A game with 8 assists dealing 480 damage will yield 51 (28+23) points.
  
• A game with 4 kills and 4 assists dealing 525 damage will yield 84 (42+42) points.
  
• A game with 2 kills and 6 assists dealing 480 damage will yield 72 (35+37) points.
  
• A game with 2 kills and 4 assists dealing 480 damage will yield 64 (28+36) points.
  
• A game with 1 kill and 3 assists dealing 540 damage will yield 53 (18+35) points.
  
• A game with 1 kill and 0 assists dealing 614 damage (an Atlas) will yield 32 (7+25) points.

Yes, a player who manages to steal all 8 kills will get a good amount of points, but stealing 8 kills is an incredible feat that will require running around the battlefield without yielding much damage or simply having rare luck (considering that each target typically has more than one person firing at it). Yet a player focusing on each target, laying in damage, and scoring a mix of kills and assists will hit higher average points.

I'd like to improve on this formula further by knowing what PGI has access to for match data. I can assume everything is available, but I don't want to waste time on assumptions.

Cheers,
Oz

Edited by Ozamis, 23 June 2013 - 09:15 PM.

[Torage]

@Ozamis

I looked over your formula and ran a few games through it and couldn't find a way to break it. Not sure what ells to say on the topic other than I hope someone at PGI reads this.

Edited by Torage, 24 June 2013 - 08:36 AM.

[hammerreborn]

Torage, on 24 June 2013 - 08:35 AM, said:

@Ozamis

I looked over your formula and ran a few games through it and couldn't find a way to break it. Not sure what ells to say on the topic other than I hope someone at PGI reads this.

I hope not, mainly for one thing...the only thing once again that's rewarded is damage. It's so ******* stupid. You want to have a weekend with not everyone playing 30 hours with 4 ppc cataphracts and saying that they are skilled like they found the hidden best way to win? Reward something other than how much damage you do. Give points for capping, give points for TAG/NARC, give points for spotting bonuses, savior kills, defender kills.

When TAG was a bonus, guess what happened? Lots of catapults running TAG + LRMs. And know what the best part was? THEY WEREN'T PPCS/GUASS. There were actual varied builds running about as people were not just content with going for max damage knowing they could make the points up elsewhere. Lights with TAG, Atlases with LRMS. Brawls and tactics actually occured rather than snipehumping hills or poptarting.

Get rid of the ******* exclusivity to damage bonuses and we can have a real tournament and a real game. It's bad enough damage is the only thing rewarded in the c-bills/xp game, it could at least be different in the tournies.

[Deathlike]

hammerreborn, on 24 June 2013 - 09:02 AM, said:

When TAG was a bonus, guess what happened?  Lots of catapults running TAG + LRMs.  And know what the best part was?  THEY WEREN'T PPCS/GUASS.  There were actual varied builds running about as people were not just content with going for max damage knowing they could make the points up elsewhere.  Lights with TAG, Atlases with LRMS.  Brawls and tactics actually occured rather than snipehumping hills or poptarting.

The problem was that TAG wasn't really a great platform for heavies. IIRC, Dragons won over other heavies in the tourney because of it. If it were part of a lights vs the world challenge... that would be OK...

Quote

Get rid of the ******* exclusivity to damage bonuses and we can have a real tournament and a real game.  It's bad enough damage is the only thing rewarded in the c-bills/xp game, it could at least be different in the tournies.

Well, this tourney didn't reward damage that much, and it changed the tourney meta in a not-so-optimal fashion...

Edited by Deathlike, 24 June 2013 - 12:33 PM.

[hammerreborn]

Deathlike, on 24 June 2013 - 12:31 PM, said:

The problem was that TAG wasn't really a great platform for heavies. IIRC, Dragons won over other heavies in the tourney because of it. If it were part of a lights vs the world challenge... that would be OK...



Well, this tourney didn't reward damage that much, and it changed the tourney meta in a not-so-optimal fashion...

It rewards damage even more than usual because Kill shots become the end all be all due to the new weighting, and a 40-60 point pinpoint alpha is far better at getting kill shots than an LRM salvo or lasers.


And who cares if Dragon's won because they were better at TAGing? GREAT. A MECH OTHER THAN A ******* STALKER OR CATAPHRACT BECOMES VIABLE AND IS REWARDED FOR PERFORMING ITS ROLE? HOLY ****! No wonder they stopped making that an incentive. Can't let those PPC boats not get free MC for hopping up and down behind a ******** hill all game. Now those are the skilled players we should be rewarding.

[Deathlike]

hammerreborn, on 24 June 2013 - 12:38 PM, said:

It rewards damage even more than usual because Kill shots become the end all be all due to the new weighting, and a 40-60 point pinpoint alpha is far better at getting kill shots than an LRM salvo or lasers.

I'm trying to recall if the meta at the time was pre-LRM nerfage (back when LRMs were at 1.8 or so).

I think it was a really nerfed LRM meta at the time...

Quote

And who cares if Dragon's won because they were better at TAGing? GREAT. A MECH OTHER THAN A ******* STALKER OR CATAPHRACT BECOMES VIABLE AND IS REWARDED FOR PERFORMING ITS ROLE? HOLY ****! No wonder they stopped making that an incentive. Can't let those PPC boats not get free MC for hopping up and down behind a ******** hill all game. Now those are the skilled players we should be rewarding.

Actually... changing the tourney meta because you don't like the current meta is not a good solution.

Besides, Catapults had an inside track to winning until Dragons abused it...

IOW: It was a mediocre tourney anyways. You didn't really get enough people to use LRMs to make that meta workable, thus making the tourney a lot more painful than it should. It would probably work better in the current meta where LRMs are passable as weapons.

Besides, NARCs completely got the shaft (it was part of the TAG/NARC bonus deal), but the thing is that they are yet to be truly viable in the current meta...

Edited by Deathlike, 24 June 2013 - 01:03 PM.

[Enigmos]

I thought the formula used in the June 22 weekend tourney was fairly right on the money. The problem with placing too much emphasis on damage is that you end up with too much focus on scattering damage. Load-outs like dual gauss or AC40 kill quickly and don't usually spread the damage around so if damage is too big a factor you end with everyone taking scatter weapons like LBX and SRM6. Balance is the key where the desire is to reduce the alpha strike builds and encourage load-out and therefore mech-build diversity. Factoring only 30% of the damage was pretty close to right: I could run dual gauss or a gauss/er ppc build and still get a decent match score. Bottom line I think their formula was at worst on the right track, considering what went before. I think the over-emphasis on damage done is a mistake on the part of those attempting to evaluate pilot skill. I know some excellent light pilots who do very little damage but play a crucial role in drawing off elements of a well prepared defensive line that would have been impenetrable had those brave pilots not drawn off the ready defense. I know some pilots who lured the OpFor into the best terrain for their team. I've seen pilots using their jumpjets to make a crucial kill versus a poptart and turn a match around. Skill like that isn't measured in damage done. Don't make a complex game into a simplistic arm wrestling match.

[Deathlike]

It's not 30%... it's /30. Even on very good day (say 900 damage), you'd only overcome say 2 assists... but that doesn't factor in how this tourney is VERY kill focused and kill-stealing tends to be a preferable solution. That's not how it should happen either. Under the current system... you don't really get that opportunity to produce 900 damage because of the kill stealing...

Edited by Deathlike, 24 June 2013 - 05:18 PM.

[Ozasuke]

hammerreborn, on 24 June 2013 - 09:02 AM, said:

I hope not, mainly for one thing...the only thing once again that's rewarded is damage. It's so ******* stupid. You want to have a weekend with not everyone playing 30 hours with 4 ppc cataphracts and saying that they are skilled like they found the hidden best way to win? Reward something other than how much damage you do. Give points for capping, give points for TAG/NARC, give points for spotting bonuses, savior kills, defender kills.

I totally agree. It's why I mentioned in my response to this thread that I'd like PGI to share what data is retained after each match is closed out so that I or any clever statistician can create a weighed and normalized formula that can dynamically cater to most--if not all--play styles in a stat-tracking setting.

The key terms are 'weighed' and 'normalized'. Simply adding multipliers to things like assists, damage, and spottings only adds weight, but without normalizing, the scoring becomes unbalanced and easily gained like we've observed in this tourney and previous events (kill stealing, assist-whoring, etc.).

The first step is to identify the weights. I can immediately think of three categories: kills/assists, damage, and support & recon. There are components that can make up each category, but those are the three I can think of. Heck, it could be just two categories: combat and support/recon.

Next is to identify the scale. How do we go about balancing these categories? We set a baseline for points in each and aim to even them out. If you look at my refactored formula for this past Heavy tourney, I struck a balance between the ideal point potential for kills/assists and damage by creating a relationship between the targets you hit vs. the damage you deal. That solved the major problems with that tournament but left out other key elements that make great MechWarriors.

Now let's add a new weight: scouting and recon. Dropping this on the scale now unbalances the rest of the equation, so we'll need to evaluate how spotting targets, using tag, etc. augments the scoring for kills/assists and damage weights.

Finally, it's a matter of tweaking based on conjecture and community feedback as to possible ways to gain the system.

I want to continue to speak in theories and hypotheticals, but I don't want to spend the cycles without hearing from PGI first in terms of what historical data is stored in order for us to theorycraft. I'd love to start an Ozamis Weighed and Normalized (OWN) campaign and see them in future tournaments or at least support someone else who's more knowledged in equalized statistics.

Cheers,
Oz

Edited by Ozamis, 24 June 2013 - 06:10 PM.

[hammerreborn]

Deathlike, on 24 June 2013 - 01:01 PM, said:

I'm trying to recall if the meta at the time was pre-LRM nerfage (back when LRMs were at 1.8 or so).

I think it was a really nerfed LRM meta at the time...



Actually... changing the tourney meta because you don't like the current meta is not a good solution.

Besides, Catapults had an inside track to winning until Dragons abused it...

IOW: It was a mediocre tourney anyways. You didn't really get enough people to use LRMs to make that meta workable, thus making the tourney a lot more painful than it should. It would probably work better in the current meta where LRMs are passable as weapons.

Besides, NARCs completely got the shaft (it was part of the TAG/NARC bonus deal), but the thing is that they are yet to be truly viable in the current meta...

But you can do both. Why the constant exclusive focus on damage? If you want scouting to exist, fire support to exist, and move away from pinpoint alphas, then you need to start rewarding it. Right now you get rewarded diddly squat for any of those things, while damage is a payday from heaven, and now people are getting MC because they can 6 ppc stalker killsteal rather than play in an intelligent fashion that supports their team and results in wins rather than giant e-peen stroke contests on who has the biggest gun.

Know what my favorite mech is? My Jenner K with 3 meds, TAG, SRM4, UAV, sensor range, target decay, and seismic. I'm the king of scouting. I can light up entire enemy teams for LRM fire and it has my highest win rate of all my jenners, but also the lowest K/D ratio as well.

Guess which mech is completely unfeasible in tournament play?

[Deathlike]

hammerreborn, on 28 June 2013 - 08:26 AM, said:

But you can do both. Why the constant exclusive focus on damage? If you want scouting to exist, fire support to exist, and move away from pinpoint alphas, then you need to start rewarding it. Right now you get rewarded diddly squat for any of those things, while damage is a payday from heaven, and now people are getting MC because they can 6 ppc stalker killsteal rather than play in an intelligent fashion that supports their team and results in wins rather than giant e-peen stroke contests on who has the biggest gun.

Know what my favorite mech is? My Jenner K with 3 meds, TAG, SRM4, UAV, sensor range, target decay, and seismic. I'm the king of scouting. I can light up entire enemy teams for LRM fire and it has my highest win rate of all my jenners, but also the lowest K/D ratio as well.

Guess which mech is completely unfeasible in tournament play?

Spider-5K. Then again, the mech as currently constituted is primarily garbage.

TAG/NARC is not the solution.. but extracting other info should be.

For instance, the XP formula for an assist is variable, depending on the damage you do (yes, it's another damage example). I do recall that getting a direct heatshot kill gave me INCREDIBLE XP though (although, this was on a DC mech, but XP counts all the same). So, I believe that existing formulas related to the "match score" can be used in some form to determine scoring. This includes capping on Conquest (although, some special modifiers should be applied with Conquest matches vs Assault).

The thing is... ANY tourney for the most part will alter your way of play SIMPLY due to scoring. It doesn't mean it's good or bad, but understand that promoting a behavior that isn't generally common to the game is NOT A SOLUTION. (Sadly, kill stealing is common, but it's mitigated by XP distribution of the kill.) TAG/NARC has never met that standard (outside of the missile boat carrying its own TAG with the occasional light/fast mech spotter that has TAG). Even the lame "assist tourney" is not common to the game (although, people do try to get cheap assists for a very small XP bonus because they are in dire need of it).

Now, only if PGI could figure out how to reward scouting.. then we'd have something here. However, the game is dedicated to combat, so DAMAGE unfortunately should be the #1 reason for success (assuming, the damage actually destroys a mech efficiently, instead of taking one's sweet time, and also is not just an outright kill steal).

Edited by Deathlike, 01 July 2013 - 11:21 AM.