40 replies to this topic

[Yellonet]

A while back I had the idea that I should collect some statistics on Clan vs IS mechs in the matches I played, and I started to collect screen shots of the match result.

What I've done is that I've only looked on matches where the losing team was completely wiped out and where no one had disconnected so that the mech name or status was hidden. Other than removing a few such matches these are 100 straight matches, I can provde the screenshots if anyone would like them for "proof" that I haven't just made up the numbers.

Now what I've collected is the number of Clan vs IS mechs in each team and also how many Clan mechs and IS mechs that survived the match (obviously only winning team).

I know it's just 100 matches, but that's all I had the energy to go through and collect numbers from...

Anyway, this is the result:





But what we can see here is that the winning team have a higher average count of Clan mechs than the losing team and in 55% of the matches, the team with most Clan mechs win.

I'm not really a statistics guy so I'm hoping that someone that is can take the numbers and get some more interesting info out of it.

Here is the spreadsheet if you would like the copy the numbers.




Oh yeah, almost forgot, Clans now proven OP easy mode!

[FupDup]

Surviving the match is a strange statistic to focus on, given that pretty much any mech in the game can be the last man standing if it uses its teammates as meatshields.

How about things like kills, damage, damage per kill, etc.?

[Yellonet]

FupDup, on 08 August 2016 - 11:58 AM, said:

Surviving the match is a strange statistic to focus on, given that pretty much any mech in the game can be the last man standing if it uses its teammates as meatshields.

How about things like kills, damage, damage per kill, etc.?

I would have liked to collect everything, including mech model, but then I would have had to have a good OCR program, writing down this by hand and then punching it into a spreadsheet was painful enough

[FupDup]

Yellonet, on 08 August 2016 - 12:04 PM, said:

I would have liked to collect everything, including mech model, but then I would have had to have a good OCR program, writing down this by hand and then punching it into a spreadsheet was painful enough

For mech models, you could probably get away with lumping all variants into one. It's not perfect because some variants are the best (e.g. compare Spider 5V to 5D for maximum hilarity) but it makes it so your life is easier.

Another way to record this might be to take a screenshot at the end of each match? Makes writing it down easier.

Edited by FupDup, 08 August 2016 - 12:10 PM.

[Pilotasso]

Clans are indeed stronger, but this game has balanced it out by giving clans a longer cooldown as well as a longer shot duration specially for larger weapons such as large lasers.

I found out I could make more fun mechs with IS models because I can use 3, 4, or 6 LL boats that just keep fire continuously while on clan units doing the same is just asking to get shot in the face while the equivalent weapons cool down. So they always require smaller weapons to be mounted and is a way to waste tonnage and alpha (not always true when you have lots of hardpoints to spam fill with small lasers, but still)

Moreover I think the IS mechs have generally more options for interesting weapon placements (lasers on arms and cannons on top of torsos are my favs for example) as opposed to clan generally weirder and asymmetrical placements.

So yeah, the clans give you slightly better chance of winning, but it is the IS that gives me more fun to drive.

Edited by Pilotasso, 08 August 2016 - 12:37 PM.

[Yellonet]

FupDup, on 08 August 2016 - 12:10 PM, said:

Another way to record this might be to take a screenshot at the end of each match? Makes writing it down easier.

That's what I did, didn't you read the OP?
But, it took long enough to just write down these things, I really don't want to do it again and with even more things. As I said though, if I had a good OCR software perhaps with batch functions it should be easy...

[SuomiWarder]

With armor being roughly equal, I would expect the side that can blast out more damage will win the majority of matches. So it is not unexpected to see that having more clan mechs on your side increase the odds of winning slightly.

[Nauht]

All this just shows is that clan mechs may be more popular. You don't have the data to show that they're more combat effective.

I'm a grognard and of all the old IS mechs I only like the WHM and MAD. Even with them though I have a mental picture of the Unseen in my head rather than PGI's variants. The BM was on my list but I don't like PGI's version of it. If they ripped the Griffin cockpit and put it on the BM I'd be all over the BM like Trump on Clinton.

Whereas I love the look of nearly all my clan mechs - they more closely resemble the look of the old art. I love the look of my Vulture, Timberwolf, Hellbringer, Summoner, Nova.

[Wolf Ender]

I would say at first glance this seems statistically insignificant.

[Alistair Winter]

Wolf Ender, on 08 August 2016 - 03:34 PM, said:

I would say at first glance this seems statistically insignificant.

Basically this. The smaller your sample is, the bigger difference you need for the numbers to mean anything. A 55% chance of Clan-dominant teams to win across 1 million matches is meaningful. But in only 100 matches, it doesn't mean much.

It's like flipping a coin. Do it 100 times, there's a fair chance it will be tails 55% of the time. But flip a coin 1 million times, it's extremely unlikely to be tails 55% of the time.

Besides... Clans vs IS isn't that interesting unless people are playing the best mechs from each faction. Kit Foxes and Cheetahs are apples and oranges.

[Pariah Devalis]

Alistair Winter, on 08 August 2016 - 03:38 PM, said:

Basically this. The smaller your sample is, the bigger difference you need for the numbers to mean anything. A 55% chance of Clan-dominant teams to win across 1 million matches is meaningful. But in only 100 matches, it doesn't mean much.

It's like flipping a coin. Do it 100 times, there's a fair chance it will be tails 55% of the time. But flip a coin 1 million times, it's extremely unlikely to be tails 55% of the time.

Besides... Clans vs IS isn't that interesting unless people are playing the best mechs from each faction. Kit Foxes and Cheetahs are apples and oranges.

Hell, I'd argue that even if it was a 5% advantage out of 1,000 games, being within 5% of equal, given the attempt at making both techs different but equal, is impressive. There is no such thing as perfect balance, and 5% rarely makes or breaks something.

[Wolf Ender]

commendable for putting in the work and taking the time to record all these games though.

[Yellonet]

OK guys, everyone understands that only 100 matches is a very low count to base anything on, we should have about 10,000 to be able to show anything really. But I sure as hell can't do that myself. Also, the data I collected is just a part of what should be collected.
If 100 people do this and we aggregate all of it we get 10,000 matches (minus a few that are the same matches).... but yeah, that's not going to happen.

I guess I just wish that PGI would show us things like this because they can of course create statistics like this if they want.

Which is the best mech for instance (highest average killrate, score or whatever) and so on... but I guess we'll never see things like that because PGI probably think stuff like that will hurt them.

[zagibu]

Yellonet, on 08 August 2016 - 12:04 PM, said:

I would have liked to collect everything, including mech model, but then I would have had to have a good OCR program, writing down this by hand and then punching it into a spreadsheet was painful enough

You don't need true OCR for this. I've written an AutoIt script in the past that let you control a Dwarf Fortress arena automatically to conduct different weapon material against armor material experiments. All you need is image fragments of all the letters and numbers and then "find" them in the screenshot. Granted, the transparency of the results screen makes it a bit harder than it was in the DF case, but it should still work with sum of average distance. If you are interested, I could try to find the code of the script.

[CK16]

Yellow should anyone take your IS v Clan balance debate seriously with you signature?

With out even reading most it just seems like "Bad guys are OP, us Good guys need to win and thus must have better stuffs) I get the same crap on WoW or Horde are the evil bad guys they must lose and die it's only right...see where this is going....

[Ex Atlas Overlord]

Yellonet, on 08 August 2016 - 11:50 AM, said:

But what we can see here is that the winning team have a higher average count of Clan mechs than the losing team and in 55% of the matches, the team with most Clan mechs win.

Clans have been better than IS since day one. This is nothing new. PGI doesn't care. The people that want lopsided "balance" certainly don't care.

[PhoenixFire55]

Yellonet, on 08 August 2016 - 11:50 AM, said:

Other than removing a few such matches these are 100 straight matches, I can provde the screenshots if anyone would like them for "proof" that I haven't just made up the numbers.

I take it its 100 matches from solo QP ?

[Yellonet]

zagibu, on 08 August 2016 - 10:30 PM, said:

You don't need true OCR for this. I've written an AutoIt script in the past that let you control a Dwarf Fortress arena automatically to conduct different weapon material against armor material experiments. All you need is image fragments of all the letters and numbers and then "find" them in the screenshot. Granted, the transparency of the results screen makes it a bit harder than it was in the DF case, but it should still work with sum of average distance. If you are interested, I could try to find the code of the script.

Yeah, that might work, the background for the characters are always quite dark if that helps. I guess one could always use some levels settings to make it completely black and white too.
Is it possible to batch your script so that it automatically looks through each file and then sends the text to a txt or how is it used?

CK16, on 08 August 2016 - 10:33 PM, said:

Yellow should anyone take your IS v Clan balance debate seriously with you signature?

With out even reading most it just seems like "Bad guys are OP, us Good guys need to win and thus must have better stuffs) I get the same crap on WoW or Horde are the evil bad guys they must lose and die it's only right...see where this is going....

The sig is somewhat tongue in cheek, but as I said I can supply the screenshots from where I've taken the data.

PhoenixFire55, on 09 August 2016 - 02:00 AM, said:

I take it its 100 matches from solo QP ?

Yes.

[PhoenixFire55]

Yellonet, on 09 August 2016 - 07:51 AM, said:

Yes.

Well Clan mechs with their superior weapon ranges are typically much better/easier to farm PUGs with. Add it to the fact that Clan mechs are more expensive and thus better players / more experienced players (i.e. those who earn more c-bills) tend to have them more often and you get the idea why your data shows what it shows.

Otherwise yeah, Clans OP, need moar nerf.

[Yellonet]

PhoenixFire55, on 09 August 2016 - 08:01 AM, said:

Well Clan mechs with their superior weapon ranges are typically much better/easier to farm PUGs with. Add it to the fact that Clan mechs are more expensive and thus better players / more experienced players (i.e. those who earn more c-bills) tend to have them more often and you get the idea why your data shows what it shows.

Otherwise yeah, Clans OP, need moar nerf.

Do you have any basis for saying that clan players are better in general? Couldn't it just as well be that Clan mechs allow these players to be a bit better?