496 replies to this topic

[Commodore_Sailracer]

I wasted a lot of time this morning combing through all the players in two large clans gathering some statistics.  I wanted to see what the actual data shows.

 

I took the player's average win rate and their win rate in the goose (as well as their win rate in the Me-410 if they also had it).  I excluded planes with 5 or less battles.  I then made a scatterplot and added a trendline on the data.  If the Goose were neither overpowered or underpowered, the data should end up with a trendline running from 0:0 to 100:100- meaning if you are a 50% player, you would win 50% of the time in a perfectly balanced plane 80% would win 80% of the time, etc. 

 

If a plane is more powerful, the line would shift up from the perfect average.  It means  you would expect to win, say 55% of the time if you are a 50% player flying that particular plane or 85% if you are an 80% win rate player.  If it is less powerful you would expect to win 45% of the time in that plane if you are a 50% player and 75% of the time if you are an 80% player.  This obviously doesn't work on an individual basis, but with a large sampling it should provide and idea of how the plane performs.

The data shows that the Goose on average has about a 5% better winrate for a given pilot than their average.  It honestly isn't as high as I was expecting, but I don't know how that compares to other planes (except the Me-410 which the chart shows).  There were a number of outliers- very low scores on the goose by both good and bad pilots that drag the line down.  I wonder if I had more data if that trend would hold.

 

Additionally, it can provide some interesting data on the skill level required to fly the plane.  A high skill / high reward plane I would expect to have a steeper line- that is a good player will do better than their average in it by a larger amount than a bad player will do better than their average in it.  You can actually see the opposite with the goose- a 90% win rate player could expect a 92% win rate whereas a 50% win rate player could expect a 55% win rate.  This would suggest there is no great skill level required to fly it.

 

The Me-410 data surprised me a lot- it actually has an even shallower slope implying good pilots win less in it than their average by a pretty serious amount and bad players to slightly better than their average.  I was expecting it to have more reward for good players based on the thrashing I have received from a good player or two in it, but the data doesn't show that.

 

(The forum won't take an image so here is a link to the charts: https://drive.google.com/file/d/1luiL_ZnYX4kEhk1CfEbyQqdFlUNl4vOT/view?usp=sharing )

 

I also started tracking my battles with XP-54s in them (including battles I fly in it).  So far the Goose side is 4 for 4 with a Goose taking 1st place.  (Though there was also a goose on the other side in one of those, so that Goose took a loss despite a 21K point showing).

[Captain_Underpants53]

[_Bronze_]

Commodore_Sailracer, on 31 December 2021 - 08:02 PM, said:

(The forum won't take an image so here is a link to the charts: https://drive.google.com/file/d/1luiL_ZnYX4kEhk1CfEbyQqdFlUNl4vOT/view?usp=sharing )

 

[Trauglodyte]

Commodore_Sailracer, on 01 January 2022 - 01:02 AM, said:

<snip>

 

A statistical man, after my own heart.

 

I'm curious, when you factored the win rates, did you do so by overall (aka. "with") versus "without"?  You wrote "I took the player's average win rate and their win rate in the goose (as well as their win rate in the Me-410 if they also had it)." but I can't tell if that is to mean that you took their overall average and compared it to that of the Goose or you took their overall average and backed the Goose impact out of it.  Essentially what I'm asking is if you calculated lift.  Because ideally what you want to look at is a player's overall play with, their play without, and then compare Goose win rate versus the win rate, of other tier 6 HFers of those same players.

Unfortunately, what you cannot do is factor out the statistical noise, of flights.  Clannies will flight up and there is nothing wrong with that.  But, win rate is greatly impacted, in a positive manner, because of it.  So, if you compare two players, both with 5k games played, and player A flights 70% of the time and player B flights <1% of the time, the noise is going to create a statistically unsound picture.  That isn't your fault but rather a limitation of the data available.

 

As to your musings of the Me 410, I'm honestly not surprised.  For all of its power, the Me 410 comes with a lot of drawbacks.  It is the most rigid, of the bunch.  Turning, like a dump truck, means that mistakes can be easily punished.  It is also the slowest, in terms of acceleration.  So again, a loss of energy is especially punishing.  On top of that, it has to manage 3x different ranges and velocities, on its weapon suite.  The MK-108s are notoriously derpy.  Less so, than the 57 mm on the Ki-102.  But, you also get less literal "bang for your buck" - the Me 410 has the 2nd lowest damager per trigger pull, at 118.7 damage.  The worst, is the Mosquito, but that is to be expected.  The largest is the Ki-102, at 221 damage per pull and the Goose is 2nd at 207.2 damage per pull.  Essentially, this means that the Me 410 requires more shots on target, to put out its overall damage (which is compounded by the different ranges, velocities, and derpiness).  The other planes, removing the Mosquito, are able to take greater advantage, of landing shots, than the 410.  And as one popular streamer has noted, "The German 30 mm isn't an anti-light fighter weapon." so its prey menu shrinks.

[Dennez]

so if i condense all this down,i think you agree with me-its the player not the plane.

in just about any plane a pilots win rate in it will pretty much mirror their overall win rate give or take a little either way.

 

[Commodore_Sailracer]

Dennez, on 01 January 2022 - 03:39 AM, said:

so if i condense all this down,i think you agree with me-its the player not the plane.

in just about any plane a pilots win rate in it will pretty much mirror their overall win rate give or take a little either way.

 

Yes and no.  I don't think the issue is as black and white as anyone on this thread has made it out to be.  When you look at the data, there are some pilots who far exceed their win rate when flying the goose (see the 43% player who has a 75% win rate in the goose.  On the other hand, there's the 66% win rate player with a 50% win rate in the goose.  Your individual opinion of the plane will be determined by how often you face off against pilots of either the first type or the second type.  

 

On the other hand the data shows a ~5% improvement overall for the Goose.  What I can't tell from this data is how that compares to every other plane in the game.  Is that 5%, on average, hugely significant or is it in line with other planes?  It is higher than the Me-410 which many people regard as strong.

 

 

 

Edited by Commodore_Sailracer, 01 January 2022 - 12:45 PM.

[Commodore_Sailracer]

Trauglodyte, on 31 December 2021 - 10:31 PM, said:

I'm curious, when you factored the win rates, did you do so by overall (aka. "with") versus "without"?  You wrote "I took the player's average win rate and their win rate in the goose (as well as their win rate in the Me-410 if they also had it)." but I can't tell if that is to mean that you took their overall average and compared it to that of the Goose or you took their overall average and backed the Goose impact out of it.  Essentially what I'm asking is if you calculated lift.  Because ideally what you want to look at is a player's overall play with, their play without, and then compare Goose win rate versus the win rate, of other tier 6 HFers of those same players.

 

 

I didn't try to do any correction for the influence of the goose on the player's overall win rate.  In general the number of battles played in the Goose is relatively small compared to a player's total.  However, now that I go back and look at it, the 3 highest win rate pilots in that graph have an astonishing 25.75%, 5.7% and 27.36% of their battles flown in the goose.  Taking out the goose impact on their win rate would definitely steepen the pitch of the fit line.

Trauglodyte, on 31 December 2021 - 10:31 PM, said:

Unfortunately, what you cannot do is factor out the statistical noise, of flights.  Clannies will flight up and there is nothing wrong with that.  But, win rate is greatly impacted, in a positive manner, because of it.  So, if you compare two players, both with 5k games played, and player A flights 70% of the time and player B flights <1% of the time, the noise is going to create a statistically unsound picture.  That isn't your fault but rather a limitation of the data available.

 

There is an inherent assumption that a pilot that flights up the same amount in a goose as in other planes as they do in the goose and their win rates reflect that.  I was also thinking that since the goose is only ~6 months old, players like me have a lot of bad battle baggage from learning to play the game that doesn't get applied to the goose.  Another thought- I completed the missions for the goose during the event with bomber escort and many of my first 40 battles were in that mode where my win rate was significantly lower than conquest mode.  As a result, at 80 battles my win rate in the Goose is lower than my overall win rate.  That is an anomaly that is being corrected with every new battle in it though.  I wonder how many other people have a similar issue?

 

Edited by Commodore_Sailracer, 01 January 2022 - 01:05 PM.

[Commodore_Sailracer]

_Bronze_, on 31 December 2021 - 09:06 PM, said:

 

[pic]

Thanks!  How'd you do that?  I tried putting the link in the image dialog both to the google drive and a 3rd party image hosting service and couldn't get it to take.

[hoom]

Commodore_Sailracer, on 02 January 2022 - 01:08 AM, said:

Thanks!  How'd you do that?  I tried putting the link in the image dialog both to the google drive and a 3rd party image hosting service and couldn't get it to take.

They uploaded it to Imgur then paste the URL without the s in https

 

Apparently I need to play at least another 9 battles in my Goose, I've only got a statistically insignificant 21 battles & winrate of 66.67% well below my 68.18% average winrate.

My 410 has a much more healthy 76.85% winrate off 108 battles.

 

Weird thing is my Heavy playstyle tends to be 'tries to turn unhealthily much' so the XP-54 should really be right up my alley.

Edited by hoom, 01 January 2022 - 02:31 PM.

[qu33kKC]

Had a pair of games yesterday that made me think of this thread.  I'm in my P-38J, and have a GAVCA Swoose Flight in blue.  Flight of Tier V fighters* in red.  I post to game chat something to the effect that this game's MM was fit to start a Forum thread about.  Final score was something like 800:89.

 

I switch to my Swoose, and the Flights have switched sides.  About half way thru, the two Fighter drivers are singing the blues in chat, requesting reports, demanding the Swoose be removed as a threat to the game, etc.  The GAVCA boys were trolling them pretty hard in response.  I at least slowed them down for half the game, but 1 v 2 was far too steep of a climb, and lost something like 360:600, got shot down 5 times. 

 

Rough times getting a Swoose flight back to back when down tiered. 

 

*Emil and XFL.

[Flaming_Fart]

 

 

Wery nice work, collecting and analyzing these data. I also expected higher winrates, as in most games xp54 seem to get almost double the score of the next player, and mostly winning, I would have guessed 3/4 - 4/5 of the times, or a winrate of about 70%, unless there is a goose on the other side as well... and that mess things up statisticly, lol

 

 

 

Dennez, on 01 January 2022 - 08:39 AM, said:

so if i condense all this down,i think you agree with me-its the player not the plane.

in just about any plane a pilots win rate in it will pretty much mirror their overall win rate give or take a little either way.

 

 

All players get better score in the goose, good players 2%, bad players 5%.

2% may sound little, but let us say that the pilots average is 67%, then his winrate in the goose is 69%. That is quite a bit, as the higher winrate you have, the harder it is to improve it, think of all the unwinable games you got, like when your human teamplayers are worse than bots vs enemy humans are even better than you, or if the map is much better suited for another class of planes, or anything else you can not overcome.

 

And 5% is much. (If you ask me.) You go from 50% to 55% winning.

[crzyhawk]

Commodore_Sailracer, on 01 January 2022 - 07:00 AM, said:

Yes and no.  I don't think the issue is as black and white as anyone on this thread has made it out to be.  When you look at the data, there are some pilots who far exceed their win rate when flying the goose (see the 43% player who has a 75% win rate in the goose.  On the other hand, there's the 66% win rate player with a 50% win rate in the goose.  Your individual opinion of the plane will be determined by how often you face off against pilots of either the first type or the second type.  

 

On the other hand the data shows a ~5% improvement overall for the Goose.  What I can't tell from this data is how that compares to every other plane in the game.  Is that 5%, on average, hugely significant or is it in line with other planes?  It is higher than the Me-410 which many people regard as strong.

 

 

 

It's what *I* have been saying.  The goose is easy.  It's got a low skill floor.  Bad pilots can do OK in it against other weak pilots.  Good pilots aren't getting anything out of it that they wouldn't get out of a plane with a higher skill floor...but higher skill ceiling.  It's a Spitfire in heavy fighter.  Good players are going to wreck me regardless of what they fly.  Bad pilots die to me regardless of whether they are in a Goose or not.

 

in bad pilot vs bad pilot, much like the Spitfire, it gives the person driving it an edge.  Where it's noticeable is with a average to decentish pilot, vs bad pilots.  They can run up pretty impressive scores, and don't draw a lot of attention from bots to help whittle them down.  it makes them LOOK stronger than they are.  I lose FAR more games to specialist bombers on the red team than I do to specialist XP-54s.

 

I'm honestly shocked to see so many people with decent win rates complaining about the XP-54.

Edited by crzyhawk, 01 January 2022 - 05:09 PM.

[Commodore_Sailracer]

crzyhawk, on 01 January 2022 - 12:06 PM, said:

 

I'm honestly shocked to see so many people with decent win rates complaining about the XP-54.

 

Perhaps another way to look at it is to see how many losses the goose is dealing out to other players.  It would be ideal if the size of each dots in the plot scaled to show how many battles that particular pilot flew the goose in.  The number of battles some of those high win rate players in the upper right of the graph have mean they are dealing out losses to opponents >80% of the time in a much larger number of battles than the average goose player. Meaning your odds of facing off against an 80% win rate pilot in the goose appear to be much higher at the moment than facing the 45% win rate pilot based on number of battles flown in my sampling.  

 

@corsair4790805 does the api support pulling that kind of data from the clan pages?  It would be great if all this data could be in a big spreadsheet so we could apply more visualisations to it.

 

From my own statistics, I am now up to 6 battles with geese in them since starting tracking.  A goose has been on the winning side 100% of the time and all geese in all battles have had to highest scores of any planes in the battle on both the winning and losing sides.  The only goose losses were to other geese.

[qu33kKC]

Took my Swoose out twice today.

 

Back-to-back Hero, Legend, Mc/Mc/Mar.

 

Sadly, only the second one was an Ace.

 

(big maps, low humans, top tier, no GAVCA Flight in red, pretty much best-case scenario in both games.)

[Corsair4790805]

Commodore_Sailracer, on 01 January 2022 - 10:51 AM, said:

 

@corsair4790805 does the api support pulling that kind of data from the clan pages?  It would be great if all this data could be in a big spreadsheet so we could apply more visualisations to it.

 

 

 

Yes and no.  The API allows pulling a list of aircraft for each player, along with the # of battles, losses, wins (and draws, IIRC).  There's no "master" list of players available through the API, nor is there a way to pull server stats for a particular plane across all pilots.

 

The closest I could come would be to use a similar approach to what I'm working on for the clan rankings.  Get list of clans -> get list of pilots in each clan -> get list of each pilot's aircraft -> collect XP-54 information.

 

There's 10,132 players in 18,331 clans as of last Monday.  I can try and repurpose the code I have to pull the XP-54 stats for clans, if there's interest.

 

[Captain_Underpants53]

Corsair4790805, on 01 January 2022 - 07:34 PM, said:

 

Yes and no.  The API allows pulling a list of aircraft for each player, along with the # of battles, losses, wins (and draws, IIRC).  There's no "master" list of players available through the API, nor is there a way to pull server stats for a particular plane across all pilots.

 

The closest I could come would be to use a similar approach to what I'm working on for the clan rankings.  Get list of clans -> get list of pilots in each clan -> get list of each pilot's aircraft -> collect XP-54 information.

 

There's 10,132 players in 18,331 clans as of last Monday.  I can try and repurpose the code I have to pull the XP-54 stats for clans, if there's interest.

 

Almost twice as many clans as there are players?!?!  WG, this whole clan set up is out of control.

 

Edited by Captain_Underpants53, 02 January 2022 - 01:29 AM.

[Commodore_Sailracer]

Corsair4790805, on 01 January 2022 - 07:34 PM, said:

 

Yes and no.  The API allows pulling a list of aircraft for each player, along with the # of battles, losses, wins (and draws, IIRC).  There's no "master" list of players available through the API, nor is there a way to pull server stats for a particular plane across all pilots.

 

The closest I could come would be to use a similar approach to what I'm working on for the clan rankings.  Get list of clans -> get list of pilots in each clan -> get list of each pilot's aircraft -> collect XP-54 information.

 

There's 10,132 players in 18,331 clans as of last Monday.  I can try and repurpose the code I have to pull the XP-54 stats for clans, if there's interest.

 

Hmm.  Sounds possible but not easy.  I'm not going to ask you to do it, but if you are willing to share your clan ranking code as a starting point and give me some help to get started, I might take a crack at it.  I'll send you a message.

[crzyhawk]

Commodore_Sailracer, on 01 January 2022 - 01:51 PM, said:

 

Perhaps another way to look at it is to see how many losses the goose is dealing out to other players.  It would be ideal if the size of each dots in the plot scaled to show how many battles that particular pilot flew the goose in.  The number of battles some of those high win rate players in the upper right of the graph have mean they are dealing out losses to opponents >80% of the time in a much larger number of battles than the average goose player. Meaning your odds of facing off against an 80% win rate pilot in the goose appear to be much higher at the moment than facing the 45% win rate pilot based on number of battles flown in my sampling.  

 

@corsair4790805 does the api support pulling that kind of data from the clan pages?  It would be great if all this data could be in a big spreadsheet so we could apply more visualisations to it.

 

From my own statistics, I am now up to 6 battles with geese in them since starting tracking.  A goose has been on the winning side 100% of the time and all geese in all battles have had to highest scores of any planes in the battle on both the winning and losing sides.  The only goose losses were to other geese.

It's not dealing the losses...the pilots are.  I dunno dude, I'm not eating these losses to goose pilots.

[Commodore_Sailracer]

crzyhawk, on 02 January 2022 - 12:42 AM, said:

It's not dealing the losses...the pilots are.  I dunno dude, I'm not eating these losses to goose pilots.

When do you fly?  I'm mostly flying 6-8PM Eastern time during the week and sporadic times during the day on weekends.  Maybe that lines up with the good pilots flying the goose?

[Flaming_Fart]

well, I think bad players can get very very god stats just by playing very good planes, and rather with a good player in a flight.

I just had a game with an xp54 player who was VERY bad, still he ended up at top of the score on the losers scoreboard. He repeatedly dogfighted with Light Fighters between bases. Not capping, and not stopping us from capping. Next round he was on my team, and red team also had an xp54 in a flight with a p-51a. The red xp54 was not a good player (to me a good player is better at helping team win by taking and holding bases).

 

But when the plane is as good as xp54, you can play bad and still make team win, just by annihilating everything that happen to be in front you. And with that speed, you dont have to wait long before something is.

 

Here a pic with the rounds scoreboard, and the two players stats. They would have had much smaller score in any othe plane the way they p