54 replies to this topic

[PhoenixFire55]

justcallme A S H, on 07 April 2019 - 05:38 PM, said:

It is all about understanding the game. Lucky that I do and I can at least, attempt, to stop the constant misinformation several users post around here.

For once I disagree with you about Khobai's post here ... Damage still is a primary contributor towards match score. However, what important isn't who gets more matchscore in a scenario "Piranha with 500 dmg and 4-5 kills vs a Supernova with 500 dmg and 0 kills", but rather "a Piranha with 500 dmg and 4-5 kills vs a Supernova with 1200 dmg and 4-5 kills", because we compare the "effectiveness" of damage both mechs required to kill the same 4-5 enemies. Needless to say 1200 dmg beats 500 dmg with other factors being equal, while in fact it shouldn't, because obviously most of it was quite irrelevant, and most likely achieved by far less effort.

Edit: Grammer.

Edited by PhoenixFire55, 08 April 2019 - 02:46 AM.

[PhoenixFire55]

Tiewolf, on 07 April 2019 - 10:49 PM, said:

My Question is if something did change (MM, playerpopulation, anything in the envoironment) that could explain the results or did I win the lottery of bad luck?

Well the only significant recent change in game balance was lock-on mechanics change and corresponding nerf/buff (depnding on who you ask) of lock-on weapons.

[justcallme A S H]

Khobai, on 08 April 2019 - 12:39 AM, said:

wrong. damage is the same

um were talking specifically about the matchscore derived from damage itself.

Damage is not the same.

I pointed out there are different types of damage quite clearly and it's impact to match score.

Edited by justcallme A S H, 08 April 2019 - 07:29 AM.

[Verilligo]

I would argue that solo kills actually ARE worth a ton of match score as opposed to pure damage being the dominant factor, but it occurs to me that I have no idea the exact score value given to anything you do in a match. Does ANYONE know how match score is derived? I get the feeling PGI has never released any of that data. I can't exactly rely on my own anecdotal numbers, either, because generally when I get a high number of solo kills I also have a high damage score. It would be interesting to see, though, at what point securing (solo) kills gets beaten out by just dealing all the spread damage ever.

[O L L O]

High KD but low WL suggests that you do too little early damge or dont share enough armor or are a skilled coward. No offence, just me reading stats, I may be completely wrong.

Edited by Ollollo, 09 April 2019 - 03:16 AM.

[Shock21]

Ollollo, on 08 April 2019 - 03:50 PM, said:

High KD but low WL suggests that you do too little early damge or dont share enoough armor or are a skilled coward. No offens, just me reading stats, I may be completely wrong.

This is a possibility. Do you find yourself surviving a lot of matches your team loses? Are you focusing too much on keeping yourself alive instead of doing what it takes for the team to win? I have a poor K/D ratio in my Atlas but a better than average Win/loss because I soak up a lot of damage while the rest of the team does the damage/gets the kills.

[MischiefSC]

Tiewolf, on 07 April 2019 - 10:49 PM, said:

Thx for the differentiated reply. Even though matchscore is debatable but for sure dmg or K/D is not a good measurement for performance or your overall contribution to the match. I never meant that because I share armor and play to win not to grind useless stats by kill stealing, seal clubbing or using the team as meatshields.
My problem is that i get constantly over significant matched to bad teams. The ones where you know no player name, the team splits up from the beginning, at least 2 yolo players die in the first 2 min of the game, no matter what you do the teams persistently camping even when get flanked from all sides, players try to be the lone rambo and all that kind of stuff. Yes from time to time there are the matches that I mess up myself by bad positioning and games that I win but most of the time no matter how hard I try to carry my team dies like flies.
First I thought it would be just a silly gut feeling but my W/L ratio is telling me, that my guts do not lie. And over 400+ games in one season the probability that it is just bad luck with the MM is unlikely. Normaly when i drop below 1 i am testing something or try to skill up the most crappy mechs there are in MWO. But thats not the case atm.
My Question is if something did change (MM, playerpopulation, anything in the envoironment) that could explain the results or did I win the lottery of bad luck?

I get what you're trying to say but mathematically it's not possible. You're swimming in the exact same pool as everyone else. Your teammates are the exact same caliber as the ones everyone else gets. The matchmaker doesn't look at Jarls List or anything else - just tiers. Once you're Tier 1 you're matched and valued exactly like every other Tier 1.

I don't know you or how you play and I'm not trying to make any sort of judgement or moralization. What I can say for absolutely certain is that if what you're doing is netting you damage and kills at a reasonable amount but not winning matches, then how you're getting that damage and those kills isn't driving wins.

The matchmaker is little more than a theory at this point. It can only build matches out of the people available. Ironically this makes your W/L even more indicative of how your behavior drives wins because it means there's less total variation in teams between players.

Win/loss is zero sum. As in for every 12 people who win, 12 people lose and there's only X number of games that have ever been played and are being played. There's not infinite variables and the outcome of every match is always 1/2 win, 1/2 lose. The other 23 players are variable and change as they do for everyone else but they change out the same people as you get. You, however, are always the same ~8% value of your team. Hence on the average after a sufficient sample size (about 80 matches) your w/l is a reasonably accurate reflection of just how good your particular 8% is at winning matches.

If W/L was actually random than every single player would have every single month shift randomly between 0.00 and 100.00 w/l. That's not what happens, not even close. Instead you'll see players stick within a pretty narrow band with a few variations and, generally, showing some improvement over time.

The worst possible tool for identifying trends is human memory. It's only slightly better than worthless and totally unreliable for statistics. Confirmation bias, the very nature of memory (you don't remember an event - you remember the last time you remembered an event. You absolutely do change your own memories over time, especially if they don't jive with what you wanted to happen or feared would happen). Just look at the numbers.

By the numbers how you're getting damage and kills isn't on par with driving wins relative to other players doing comparable damage and getting comparable kills to you. If you want to win more you should evaluate how you're doing damage and getting kills and what you can do to drive more wins.

[MischiefSC]

PhoenixFire55, on 08 April 2019 - 02:37 AM, said:

For once I disagree with you about Khobai's post here ... Damage still is a primary contributor towards match score. However, what important isn't who gets more matchscore in a scenario "Piranha with 500 dmg and 4-5 kills vs a Supernova with 500 dmg and 0 kills", but rather "a Piranha with 500 dmg and 4-5 kills vs a Supernova with 1200 dmg and 4-5 kills", because we compare the "effectiveness" of damage both mechs required to kill the same 4-5 enemies. Needless to say 1200 dmg beats 500 dmg with other factors being equal, while in fact it shouldn't, because obviously most of it was quite irrelevant, and most likely achieved by far less effort.

Edit: Grammer.

Even more to the point - a Supernova with 1200 damage may have had a minimal impact on driving the win while a pro grade squirrel in a Commando with 300 damage and 0 kills may have tied up 4+ very deadly mechs long enough for his teammates to turn that 11 v 8 situation into a roll. The Mando may have absolutely driven the win and, conversely, may have a great W/L and a mediocre match score as an average.

End of the day the guy with match score average of 350 but a W/L of 1.2 is not as valuable a teammate at driving wins as the guy with match score average of 250 but a W/L of 1.8. That's not random, that's not luck (if they're both in the hundreds to thousands of matches to shake out variables) that's a direct reflection of how those players and how they play drives wins.

[Jman5]

Without watching some of your games to see what is going on it's hard to say precisely what you need to improve. I honestly have no idea what the typical WLR is for someone with your average matchscore so I don't know if it's unusual or not. Though a 1.0 WLR would indicate that you're actually in the perfect spot to provide you with a great challenge.

Here are some general thoughts if you want to win more:

1. Do you drop call or provide any instructions in chat frequently? If you do, it's possible that some of your calls on some maps may need to be reconsidered. (Nothing wrong with that. I rethink my map tactics all the time) If you don't, then providing solid calls might up your winrate. Also, how you communicate can be just as important.

2. Keep working on your fundamentals. Be hungry to do more damage. Practice torso twisting your mech better and more frequently. Get better at poke-trading from cover. Work on your situational awareness by occasionally stopping and reading your minimap. Don't tell me you do all these things fine already. Everyone can do it better.

3. Don't blame your team when you lose. Yes a bunch of them played badly, but there is nothing you can do to make them play better. You can't force matchmaker to give you a better team either. All you can do is play your mech better. After the loss ask yourself what you could have done better. What mistakes did you make, could you have positioned better? Were you moving and shooting well? etc... Don't fixate on your dopey teammates who dropped the ball. IMO, you have plateaued and the only way you're going to break it is by looking hard at your own play and incrementally improving it.

Edited by Jman5, 08 April 2019 - 05:18 PM.

[PhoenixFire55]

MischiefSC, on 08 April 2019 - 04:35 PM, said:

Even more to the point - a Supernova with 1200 damage may have had a minimal impact on driving the win while a pro grade squirrel in a Commando with 300 damage and 0 kills may have tied up 4+ very deadly mechs long enough for his teammates to turn that 11 v 8 situation into a roll. The Mando may have absolutely driven the win and, conversely, may have a great W/L and a mediocre match score as an average.

End of the day the guy with match score average of 350 but a W/L of 1.2 is not as valuable a teammate at driving wins as the guy with match score average of 250 but a W/L of 1.8. That's not random, that's not luck (if they're both in the hundreds to thousands of matches to shake out variables) that's a direct reflection of how those players and how they play drives wins.

Oh I agree, but its actually hard to track movements like these and also their contribution since it requires team to capitalize on the created opening, which isn't always the case. The effectiveness of damage on the other had is very easy to track. Since the game knows which component destruction led to each mech death (i.e. head, CT, both STs, both legs etc.), then it can also track who did damage to that particular component and how much they did. Hence it can give KMDDs and solo kills not based on total damage to a mech, but rather only to the parts that were absolutely necessery to kill said mech. I.e., if a mech died because it lost its legs, then only people who did damage to legs get rewards according to who did what % and all damage dealt to arms, head and torsoes isn't credited nearly as much or at all.

The other factor in winning matches for example is taking away as much firepower as quickly as possible. So if you take for example a Dragon with 3xAC2s, then destroying that ballistic arm should probably be worth more than destroying the entire rest of the mech combined. Its not hard to implement either, since obviously game knows exactly what % of firepower is located in what component on each mech. That will also IMO make people more willing to understand and learn the game, since in order to get good match reward they'll need to know mechs, their loadouts and deal with them effectively.

But of course thats never happening. Not in this game and not while PGI is at the helm anyway.

Edited by PhoenixFire55, 09 April 2019 - 12:18 AM.

[FRAGTAST1C]

The game can track how many components are destroyed. Perhaps it can be made to track the number of weapon systems destroyed as well?

[MischiefSC]

PhoenixFire55, on 09 April 2019 - 12:00 AM, said:

Oh I agree, but its actually hard to track movements like these and also their contribution since it requires team to capitalize on the created opening, which isn't always the case. The effectiveness of damage on the other had is very easy to track. Since the game knows which component destruction led to each mech death (i.e. head, CT, both STs, both legs etc.), then it can also track who did damage to that particular component and how much they did. Hence it can give KMDDs and solo kills not based on total damage to a mech, but rather only to the parts that were absolutely necessery to kill said mech. I.e., if a mech died because it lost its legs, then only people who did damage to legs get rewards according to who did what % and all damage dealt to arms, head and torsoes isn't credited nearly as much or at all.

The other factor in winning matches for example is taking away as much firepower as quickly as possible. So if you take for example a Dragon with 3xAC2s, then destroying that ballistic arm should probably be worth more than destroying the entire rest of the mech combined. Its not hard to implement either, since obviously game knows exactly what % of firepower is located in what component on each mech. That will also IMO make people more willing to understand and learn the game, since in order to get good match reward they'll need to know mechs, their loadouts and deal with them effectively.

But of course thats never happening. Not in this game and not while PGI is at the helm anyway.

Easier is just have an Elo for each mech and a modifier for different weapon systems. Still never going to be precise but if, say, X variant of Y mech tends to get a lot of wins, especially with Z weapons than the players Elo gets a bump for using them. If you're in a lore build/trash mech your value in the Elo calculation is lower.

It's doable, absolutely doable. However given the current population level I'm not sure how much difference it would make in the actual matches created.

Far more valuable to the Matchmaker would be a 'Filler' queue.

You have to create a dropdeck for IS and Clan, you need patron and mech selected for each Solaris queue and a mech of each weight class selected for QP. You have all these options pre-selected and you get to choose 'I prefer FW first, then Solaris, then QP' or whatever order you want and it fits you in a match with whatever of those is going to fill by your presence. Super fast matches! It also gives a very valuable boost to the Matchmaker.

So the Matchmaker has, say, 30 people for FW, 40 for QP and 7 for various Solaris divisions. The 30 in FW may or may NOT make a match because it could be a 12man for Clans and for IS 2 pugs, a 7 man and a 9 man. This means those 30 people are queued but no matches created so those 30 people are available but tied up with 0 matches created. The 40 for QP could be piss-poor matches for tonnage and PSR plus it's got 16 left over after it fills a match and it spends 60 seconds queued for the match because it's waiting for the release valves that let it say '**** it, PSR and Tonnage matching sucks but drop the chumps in and turn the blender on'. The 7 in Solaris divisions probably have 3 people waiting for their division at any time.

What's all that mean? It means that of the 77 players in that 60 second window of time in queue (not even in matches, just in queue) only 28 are actually getting matched. So 49 players at stuck queued.

If, however, you've got even just 10 'Fillers' queued out of that 49 you get 1 more pug to get the 9man + 3 pugs and BOOM. They and the 12man in a FW drop. Drop 3 into Solaris and they're all filled, that leaves 6 who've got 1 mech in every weight class for filling that QP drop with way better tonnage (and possibly PSR) matching without the 60 second wait for release valves.

Result? faster, better match for QP and 24 (1/2 of) the people stuck in queue are now in matches - which means they finish games faster and are released back to the queue to fill more matches.

You get a powerful cumulative value from this. Faster matchmaking = more people available each 60 second window to fill other matches. This would let you get better quality match making and less wait time with your existing resources.

I think it would be a worthwhile expenditure of resources for PGI.

[PhoenixFire55]

MischiefSC, on 09 April 2019 - 08:28 PM, said:

Easier is just have an Elo for each mech and a modifier for different weapon systems. Still never going to be precise but if, say, X variant of Y mech tends to get a lot of wins, especially with Z weapons than the players Elo gets a bump for using them. If you're in a lore build/trash mech your value in the Elo calculation is lower.

It's doable, absolutely doable.

Oh absolutely, I've talked about this a lot back in the days on the forum too. Need a slight adjustment value for the amount of skill point in a mech as well. The amount of matches needed to "normalize" the adjustment for each mech of a certain player would be quite a lot, however, each mech can have an additional adjustment value based on the average performance of said mech across all players. Same with loadouts. Again, it can all be done, sure, just takes effort, and it means it'll never happen.

MischiefSC, on 09 April 2019 - 08:28 PM, said:

However given the current population level I'm not sure how much difference it would make in the actual matches created.

I will repeat myself once again. The size of the population you need for a balanced solo play 12v12 match is exactly 24 people. Population level is nothing but a lame excuse.

MischiefSC, on 09 April 2019 - 08:28 PM, said:

Far more valuable to the Matchmaker would be a 'Filler' queue.

You have to create a dropdeck for IS and Clan, you need patron and mech selected for each Solaris queue and a mech of each weight class selected for QP. You have all these options pre-selected and you get to choose 'I prefer FW first, then Solaris, then QP' or whatever order you want and it fits you in a match with whatever of those is going to fill by your presence. Super fast matches! It also gives a very valuable boost to the Matchmaker.

No. Because people don't like being forced into things. If someone doesn't want to play 1v1 Solaris matches he will rather quit the game then endure them happening when he wants to drop 12v12.

[Xiphias]

Dragonporn, on 07 April 2019 - 03:20 PM, said:

A while back folks were convincing me, that if your personal performance is good, your average match score and W/L would be higher. Considering how it works in reality, as you can easily hit like 10+ match streak where you barely do anything and just get wins after wins, while you can also hit over 10+ losestreaks where you consecutively score over 500 each match and 3-5 kills. I argued that your personal performance isn't as important as sheer luck factor of hundreds, if not thousands, different factors you can not have any control over in QP. Prolly I was wrong after all, eh.

It is true. I pug a lot (when I'm actively playing) and I've been able consistently maintain both a high match score and a decent WLR ( > 1.5) playing pure solo. While you might have a long losing streak, on average you absolutely can do better. However, to have a significantly higher average you have to play significantly better than the average. If you aren't seeing it it means you aren't contributing enough to make a big difference. You're argument is directly opposed to statistics. Just because you can't see it doesn't mean it isn't happening and the data bears that out.

Arnold The Governator, on 07 April 2019 - 04:07 PM, said:

My friend and I are having the same problem lately. No matter if we solo que or group up, 90% of our games end up in a loss. The main problem is no one is willing to work as a team and communicate properly. I can't really blame the mechs that we are in (although not all of our mechs are maxed out in the skill tree but I highly doubt that is the reason) since him and I know our defined roles. He wants to uninstall the game and I really can't blame him for him.

Not having max skilled mechs can have a decent impact, but the bigger issue isn't so much about communication. Skill is more important. A skilled team with no communication can easily stomp a weaker team that is communicating. Communication is good, but it only really matters if the skill level is close. If neither team is communicating it's not the communication (or lack thereof) that is causing the losses.

justcallme A S H, on 07 April 2019 - 05:38 PM, said:

Now I know you're gonna find a screenshot with a PIR thats had say 4 kills and 450dmg and be lower match score than a SNV-A with 550dmg... But that's the thing - if the match score IS lower then the PIR was not getting the KMDD/Solo/Blows, it was simply jumping in for killshots at the last second without doing a hell of a lot.

While I agree with most of what you said (damage =/= matchscore) KMDD and solo kills don't always reward as they should. Take the case of an assault who gets sandblased on the front armor by LRMs and the later gets backstabbed by a light who kills it from behind. The LRMs contributed nothing to the kill, but will still get the KMDD because they did a lot of damage where the light did efficient damage. I run into this a fair amount when I am picking off crit components and targets, but not getting KMDD because someone else sprayed it with damage (e.g. 1 streak salvo = 2 pinpoint alphas). Sure, you'll get some of that back with component destruction, but spray damage often farms the KMDD.

InspectorG, on 07 April 2019 - 07:19 PM, said:

That or get really good at vulturing with a light mech at end of match.

I do pretty well in solo running lights. I attribute it to a couple of things. 1) Controlling the battlefield, 2) Killing crit targets quickly 3) being able to burn down damaged mechs at the end of a match (AKA, vulturing). Vulturing is an important skill to swing those close matches, but the first two are important to getting to the last step more frequently.

Tiewolf, on 07 April 2019 - 10:49 PM, said:

First I thought it would be just a silly gut feeling but my W/L ratio is telling me, that my guts do not lie. And over 400+ games in one season the probability that it is just bad luck with the MM is unlikely. Normaly when i drop below 1 i am testing something or try to skill up the most crappy mechs there are in MWO. But thats not the case atm.

My Question is if something did change (MM, playerpopulation, anything in the envoironment) that could explain the results or did I win the lottery of bad luck?

At 400 matches the odds of it being bad luck are extremely small, it definitely sounds like a gameplay issue. More information would help though, what kinds of builds/weapons are you mainly running?

Something like a streak boat could get a decent KDR by killing light mechs and get good damage numbers while being pretty ineffective at doing significant effective damage to heavier mechs. Without seeing you play it's hard to say exactly what going wrong. I lot of people will say communication, but you can effective even without using a mic. Here are a few tips based on my experience pulling a high WLR in solo pugs.

1) Use your positioning and firepower to control the flow of battle.

2) Prioritize targets and kill efficiently

3) Get good at mopping up the end of a match

1)
Understand how your team and the enemy team react to actions you take. For example, as a light I will often see the team start to follow me at the start of a match. This can be good or bad. I can tell the team not to follow me (communication) or just try to lead them to a good location.

Shooting at enemies also controls their movement. For example, if you are watching a corner, a mech pokes, and you shoot it then it will usually go back behind cover. If you do this a couple of time that will will either die (good) or will rotate to a different position to try and get better angles, you've forced them to move in a direction that you want. Using your firepower carefully can corral enemies into moving to a location that is better for your team. Being a force multiplier is also important. Being outnumbered is bad, engaging in even numbers can lead to slugfests, but having multiple mechs shoot the same target will force them to either retreat or die, increasing your chances of numerical superiority.

Assaults that look like they are alone will tend to draw light mechs. Light mechs will draw streak mechs and turn assaults. Shooting mechs in the back will cause them to turn. Hitting mechs from the side will distract them and split their attention. Your team is more likely to shoot a target that is locked. This type of knowledge can help you manipulate the battle without having to communicate at all.

2)
Target priority is incredibly important. Killing mechs quickly is what wins games, not picking up kills at the end once a match is decided. Obviously, being good at spotting and picking off crit targets is important, if you can keep a mental damage checklist of the enemy team it helps a lot (cycle targets frequently, look for smoke). Focusing down dangerous targets is important because they can quickly turn a match. Distracting a target can be almost as good as killing it if you can keep it from being effective in the fight for long enough.

3)
Plan for the end game all match. Know what is dangerous to you and work from the start to kill it. As a light I prioritize getting damage in on streak boats early and try to bait them into my team. Once they are dead I have a lot more freedom and I don't have to worry about them if the match gets close. Lights I tend to mostly ignore early on (unless I have easy shots), once the game starts getting closer to the end (e.g. both teams have only a few mechs left) I will often prioritize light mechs above heavier targets because they will be the ones that are able to chase me down. If I can leg/kill them I can usually wear down their heavier teammates.

As an assault I would probably also try to kill lights in that situation because they would have an advantage in a 1v1. As a medium, heavies/assaults would probably be a bigger threat. Kill the biggest threats while you still have teammates to help.

Being able to quickly switch targets for a shot or two is also important. Keep targeting the assault, but if a light runs into your fire line and you can get a free shot take it. It will hurt the light a lot more than it will hurt the assault. If you can't get a solid shot on the light you're better off shooting the assault. By willing to jump targets in an instant when the opportunity presents itself.

These are the things that I find to be most helpful at edging my team towards victory in solo. It won't be able to win every match, but you'd be surprised at how big a difference it can have in the long run.

[death390]

honestly i have noticed a lot more stomp matches recently. i just went up to T4 and am a recent join to the community (~500 games). before the LRM nerf (yes i consider it a nerf) i enjoyed 3 types of playstyle: LRM fire support (around 500-600m and armor sharing/ scouting in MDD-H), Sniper (shadow-cat ERPPC+ERLL+3xLMG), and flanking (with same shadowcat). since the changes to LRMs i have been playing less and less since it is not as fun for me anymore due to the horrendous IDF lock timers making it nearly impossible to be fire support (MDD was my first buy, and i enjoyed it a lot).

when it comes to the games i play i honestly don't care about Last hitting (who gets the kill) i prefer to open someone up, call it out and GTFO since i am often squishier than my target, or running LRMs and ripping apart their front torso. i used them A LOT when i first started and was learning the maps so i fired quite often into walls/ ceilings/ roofs and my accuracy is still around 40% with a LRM20.

with the LRM changes i might not stick around much longer because the ONLY mech i have found to still be fun is my shadow cat since the Mad dog is useless at this point (i know it can work but have to use ATMs and brawl which i hate). i mean i took my MDD out for about 5 matches using LRMs after AMS died down to a resonable level and if i did 200 damage over 1600 missiles it was a "good" match. i was used to seeing 500-700 damage as fire support before the nerf. and i think that is part of the problem, since there is basically no fire support anymore the game rewards those who can pinpoint the most damage. double gauss, backstabbers, laser vom, ect. the pinpoint damage game is 10x higher since people are untouched until everyone is in each others effective ranges (the classic frozen city i might as well power down since i am in effective range and die at a glance).

MDD build was LRM20x3, HMLx3, Tag.