18 replies to this topic

[Duke Nedo]

I have complained a lot about unnatural streaks in MWO, especially after 3-3-3-3 was introduced (I think) in solo drops. I took the time to collect the statistics to see if my "hunch" that you are more likely to loose if you select one of the over-populated classes (i.e. Heavy or Assault). This is what I found from my personal dataset:


Red bars is the "baseline" so to say for matches played before clan mechs were launched, and blue are my clan chassi games.

For my stats after clans were launched, I have about the same kill/death ratio and comparable damage/game in all classes, but my win/loss ratio is much better for lights and mediums. Thus, it is true that I have been significantly more likely to loose a game if I launch in one of the overpopulated classes, i.e. Heavy or Assault. In contrast, the historical data pre-clans shows a weaker and reverse trend, there I was more likely to win in an assault, which somehow makes sense since there was no 3-3-3-3 so the more you brought the more your team ended up with.

I have all my loss streaks when trying to skill heavies and assualts. Does that make any sense at all?

For PGI: would it be possible for us to reveal a bit about how the MM goes about it when it releases the 3-3-3-3 constraints? Is it possible that when you are a "filler" that you somehow are more likely to be placed on the team most likely to loose?

For fellow mechwarriors: How does your stats look? Is this only me or is this a pattern? NB: only solo drops!

Edited by Duke Nedo, 29 September 2014 - 01:08 AM.

[MrMadguy]

Every mech class has separate ELO, so may be you just have lower rank for those classes? For examle, you have abused overpowered timbers and dires too much, so your heavy/assault rank is too high. But at the same time I totally agree about lose streaks, while leveling new mechs.

Edited by MrMadguy, 29 September 2014 - 01:20 AM.

[Duke Nedo]

According to the historical stats I may have a bit higher elo for the assault class, maybe, since my w/l ratio is a bit higher there but I doubt there is any big difference in elo. My Timberwolves have a w/l ratio below 1.0 so I am pretty sure they did not inflate me...

And, even if there was a big difference in elo between my classes, that should give me good opponents AND good teammates so it really should not affect the w/l ratio dramatically. Surely not enough to result in any win- or loss streaks.

Also: If we ignore lights because of jesus-boxes and locusts, before the introduction of 3-3-3-3 I performed worse in mediums than in heavies/assaults, but after 3-3-3-3 I suddenly get much more wins in mediums. That goes for both Novas and Stormcrows, no difference between the two (though sample is very very small). Also the fact that the Timber and Whale are both relatively much more overpowered than the chassi I had better ratios in the archived data, like Dragons, Awesomes, Catapults, etc... the exception being the DDC, that one counts.

Perhaps its just a feeling, but it certainly feels like something changed dramatically!

Edited by Duke Nedo, 29 September 2014 - 01:38 AM.

[Chrithu]

You win some, you lose some, most of the time I have a lot of fun. Only stat that matters is if what I did helped the team no matter if we lost or won.

[Duke Nedo]

Jason Parker, on 29 September 2014 - 01:41 AM, said:

You win some, you lose some, most of the time I have a lot of fun. Only stat that matters is if what I did helped the team no matter if we lost or won.

Sure, I agree that fun is what matters, and for me close games are fun no matter if you win or loose.

[Dracol]

Duke Nedo, on 29 September 2014 - 01:05 AM, said:

For my stats after clans were launched, I have about the same kill/death ratio and comparable damage/game in all classes, but my win/loss ratio is much better for lights and mediums. Thus, it is true that I have been significantly more likely to loose a game if I launch in one of the overpopulated classes, i.e. Heavy or Assault. In contrast, the historical data pre-clans shows a weaker and reverse trend, there I was more likely to win in an assault, which somehow makes sense since there was no 3-3-3-3 so the more you brought the more your team ended up with.

Going by damage alone, if you are getting the same damage from Clan mechs as you did IS mechs, then you my be under performing in your clan mechs. Clans can deal more damage over time than an IS mech because they don't have the pin point accuracy to take out mechs quickly.

[Griggio]

You would need to run some numbers for overall average damage to get a better picture of whether under performance is the issue. Last time I pulled my numbers the difference between my clan mechs and what would be considered the best IS mechs were much closer than one would think minus the Dire Whale which isn't surprising.

Edited by Griggio, 29 September 2014 - 03:01 AM.

[Charronn]

I don't know about being deliberately put in bad teams but what I did notice is when i'm grinding a different weight class from my usual or if i'm grinding a new mech my winrate usually takes a nosedive or if grinding a mech I don't like.
I tend to stick to heavies and assaults so when I jump down to mediums or lights it takes a fair bit of getting used to.In my case anyhoo I think it's down to the guy behind the keyboard.

[Vezm]

Am I the only one for whom the picture is tiny?

[Duke Nedo]

Griggio, on 29 September 2014 - 03:01 AM, said:

You would need to run some numbers for overall average damage to get a better picture of whether under performance is the issue. Last time I pulled my numbers the difference between my clan mechs and what would be considered the best IS mechs were much closer than one would think minus the Dire Whale which isn't surprising.

Sure, damage per match is 30-50% higher post-clans and k/d ratio also higher for all classes so no problem there...

Charronn, on 29 September 2014 - 03:18 AM, said:

I don't know about being deliberately put in bad teams but what I did notice is when i'm grinding a different weight class from my usual or if i'm grinding a new mech my winrate usually takes a nosedive or if grinding a mech I don't like.
I tend to stick to heavies and assaults so when I jump down to mediums or lights it takes a fair bit of getting used to.In my case anyhoo I think it's down to the guy behind the keyboard.

I am pretty sure its not about me at all, I don't want to overestimate my impact on the match result. I perform about the same in all classes I tend to believe. It's just a weird thing that whenever I jump into a heavy or an assault mech I stand a bigger chance of ending up with a bad team. In a heavy I still have a k/d ratio of 1.75 and avg damage per match of 430, where both are higher than pre-clans even though I was on the winning side much more before clans... so I do better as far as you can judge from k/d and dmg, but I still loose more.

I don't know, nothing of this makes sense, so that's why I am asking if others see the same trend? Perhaps I am just unlucky, I mean the sample is small... but its ~200 drops per class now post clans so its starting to become quite significant. But again, doesn't make any sense at all to me.

So, the key question for me is exactly what does the MM do when it gives up on 3-3-3-3 and starts filling up the match. When dropping in lights/mediums I should be placed in a game before the constraints are loosened up, but when dropping in heavy/assault I should quite often be placed in a game after the constraints are gone. Will for example ELO still be in play at this stage, or is has the ELO restraints also been released? etc etc...

Edited by Duke Nedo, 29 September 2014 - 05:26 AM.

[Hotice]

There are nights that I just couldn't get a win. I have gotten 5 kills and over 800 damage done but still loose 6 to 12.... There is nothing anybody can do. I have seen my team couldn't kill one mech with multiple people "focused" on the enemy mech. It is so sad and funny at same time. My Atlas walked 1800m to get to the said target and poped it with two alpha hits. Not sure why MM pitch the group together like that. I'm not sure if this is pure RNG luck with MM system though.

[Mawai]

Most data is needed

Do you drop in clan or IS mechs?

If you drop in IS mechs ... part of the numbers could be due to the effectiveness of clan heavies and assaults ... particularly the timberwolf. If you drop IS, you may slightly weaken your team compared to dropping in the top clan mechs (depending on your mech and build) ... statistically, this could result in a reduction of your win/loss after the introduction of clans.

On the other hand, if you are dropping in clan mechs in the assault or heavy classes ... since they are all new ... you may not have unlocked any of the skills for these mechs yet. This reduces their effectiveness. If you have been working on leveling clan mechs then your win/loss will be lower since you may not have been as effective as possible ... though even an unleveled timberwolf is pretty effective

I also noted that you die surprisingly infrequently.

I tend to die in about 25% to 50% of my wins (assuming I die every loss since I usually play skirmish/assault). In lights it is more like 50% while in heavies it is a bit less. On the other hand, your numbers seem to indicate that you die in about 10 to 15% of your wins. This significantly affects KDR.

So ,,, it is possible that your numbers are affected by the game mode you play.

If you play conquest ... clan mechs have more speed for a given firepower making them much more effective in conquest than comparable IS mechs, If you are playing IS mechs in conquest then I could see the mobility difference also playing into the difference in W/L.

Without knowing your playing habits/mechs/game modes ... there is no way to reach any conclusions regarding the numbers you posted for pre and post clans.

(As someone else mentioned .. you have a separate Elo for each weight class so you could have different competition in each class ... also your skill level migth be changing relative to the overall population which would also show numbers like these ... i.e. more folks dropping in clan mechs ... they skill them up and get better ... your W/L in that class is affected since the folks you are matched against based on Elo are playing more effectively in the new mechs).

[Mawai]

Hotice, on 29 September 2014 - 05:52 AM, said:

There are nights that I just couldn't get a win. I have gotten 5 kills and over 800 damage done but still loose 6 to 12.... There is nothing anybody can do. I have seen my team couldn't kill one mech with multiple people "focused" on the enemy mech. It is so sad and funny at same time. My Atlas walked 1800m to get to the said target and poped it with two alpha hits. Not sure why MM pitch the group together like that. I'm not sure if this is pure RNG luck with MM system though.

Do you say anything in team chat at the beginning or during the match to try to organize your team?

I was in a match yesterday (solo PUG only) ... I usually start matches by saying "regroup, stick together, focus fire " ... if they listen we usually do pretty well. I accidentally sent that to "all" in one match ... so decided I had better do more organizing ...

I got everyone together ... we waited for the assaults to catch up ... (this was on Tourmaline) ... and then we pushed forward all together more or less ... the other team was doing the usual peeking stuff ... we had 3 or 4 mechs firing every time they peeked ... we kept moving ... they split to come at us from different directions but didn't coordinate it so we took out 3 or 4 in front then turned and took out 3 or 4 behind ... we won the match 12:3 or 12:4. Someone on the other team was complaining in all chat about having a useless team. All I could do was LOL ... I told them what to do at the beginning by accident ... then I followed up more that I usually do with the following 3 sentences ...

"Let's meet up and push together ok?"
"Wait for those slow assaults"
"lets go" - and I started walking forward and everyone else did too

About 14 words ... took all of 10 seconds to type. Made the difference in this match.

And you know what ... in many matches when I don't say anything ... NOBODY does. So if you find yourself in matches where it feels like your team failed ... if you didn't say anything ... then the fail is not with the team exclusively

Note: Trying to organize doesn't always work ... folks still decide to do their own thing (which is fine) ... you don't always win either .. especially if the other team is effectively more organized (I was in another solo PUG match on the weekend where the other team felt as coordinated as a large group ... they were all moving together ... spread across a firing line (not single file) ... pushing when needed ... they wiped us 12:2 but they gave the appearance of being far more organized than my team).

P.S. focus fire is one of the harder elements to coordinate but among the most effective. Use R to target ... focus on weakness and KILL opponents if at all possible (though chasing a rabbit light mech is usually also a good way to lose a match ).

[Duke Nedo]

Mawai, on 29 September 2014 - 06:11 AM, said:

Most data is needed

Do you drop in clan or IS mechs?

If you drop in IS mechs ... part of the numbers could be due to the effectiveness of clan heavies and assaults ... particularly the timberwolf. If you drop IS, you may slightly weaken your team compared to dropping in the top clan mechs (depending on your mech and build) ... statistically, this could result in a reduction of your win/loss after the introduction of clans.

On the other hand, if you are dropping in clan mechs in the assault or heavy classes ... since they are all new ... you may not have unlocked any of the skills for these mechs yet. This reduces their effectiveness. If you have been working on leveling clan mechs then your win/loss will be lower since you may not have been as effective as possible ... though even an unleveled timberwolf is pretty effective

I also noted that you die surprisingly infrequently.

I tend to die in about 25% to 50% of my wins (assuming I die every loss since I usually play skirmish/assault). In lights it is more like 50% while in heavies it is a bit less. On the other hand, your numbers seem to indicate that you die in about 10 to 15% of your wins. This significantly affects KDR.

So ,,, it is possible that your numbers are affected by the game mode you play.

If you play conquest ... clan mechs have more speed for a given firepower making them much more effective in conquest than comparable IS mechs, If you are playing IS mechs in conquest then I could see the mobility difference also playing into the difference in W/L.

Without knowing your playing habits/mechs/game modes ... there is no way to reach any conclusions regarding the numbers you posted for pre and post clans.

(As someone else mentioned .. you have a separate Elo for each weight class so you could have different competition in each class ... also your skill level migth be changing relative to the overall population which would also show numbers like these ... i.e. more folks dropping in clan mechs ... they skill them up and get better ... your W/L in that class is affected since the folks you are matched against based on Elo are playing more effectively in the new mechs).

All post clan drops are in clan mechs, I have just been skilling them. But, that's basically true for the IS mechs before clans as well, I mainly skilled mechs and bought new ones and skilled them... Game mode has constantly been random. The major factors that make the pre-clans and post-clans dataset different works in the opposite direction as far as I can see. After clans I have dropped in better mechs, I have scored much more damage and kills, I am a better player now than in the early beta and I "carry harder" now. All these things should make me win more now, and I do for lights and mediums... but for heavies and assaults it's the opposite. I still score better and all, but win significantly less.

Anyways, I agree I have much too little data here, that's why I am so curious if there are other 100% solo players out there that would have a look at their stats and see if it fits my data or if it's completely different....

Edited by Duke Nedo, 29 September 2014 - 08:32 AM.

[FupDup]

One explanation might be that the better players in this game are less likely to pilot sub-par mechs than "meta" mechs. They play to win, so they prefer to drive the robots that are better at helping them win. And in this case, that usually means heavies and assaults.

[Duke Nedo]

What does that have to do with anything?

[FupDup]

Duke Nedo, on 29 September 2014 - 11:59 AM, said:

What does that have to do with anything?

Your original post raised the question of if someone is more likely to get a bad team when dropping in overpopulated classes. My post was a guess at why that might be. Most commonly (but not always) this is heavy and assault mechs. Reiterating my previous post, my guess was that the "good" players often use mechs that are better at winning, which just so happens to line up with those overpopulated heavy and assault classes. This means that on average, there are probably fewer "meta" players in the lower classes, hence their lower performance.

[Duke Nedo]

I interpret what you are writing as "players playing the underpopulated classes are worse players", which is something else. What I was observing in my stats was that if I drop in a popular class, I stand a higher chance of losing the match after 3-3-3-3 was introduced than before.

Anyhow, the whole thing doesn't make any sense so probably just coincidence. Would be fun to see more data though.

Edited by Duke Nedo, 29 September 2014 - 02:10 PM.

[Deathlike]

FupDup, on 29 September 2014 - 12:47 PM, said:

Your original post raised the question of if someone is more likely to get a bad team when dropping in overpopulated classes. My post was a guess at why that might be. Most commonly (but not always) this is heavy and assault mechs. Reiterating my previous post, my guess was that the "good" players often use mechs that are better at winning, which just so happens to line up with those overpopulated heavy and assault classes. This means that on average, there are probably fewer "meta" players in the lower classes, hence their lower performance.

So... that means more shaking fists @ my "meta" Cute Fox?

I named mine Cute Poopy Fox for a reason.

Edited by Deathlike, 29 September 2014 - 02:25 PM.