Stats, ENY, and Perks

[BnZs]

So, what you're saying is that your personal opinions about aircraft performance and ability are more important and authoritative than the judgment of the masses, especially your outlandish opinions about the P-51D.

Yes, my judgment about airplanes is likely superior to that of the masses, as is that of anyone who has been flying awhile and who is a performance stats junkie. Understand, the "masses" we speak of are not necessarily choosing the highest performing plane, but a plane they LIKE, many AH players do this or the arena would be ALL La7s, Spit16s, 109Ks and Ki-84s. This is another reason it is unfair to assign ENY values by popularity, not only may the "masses" be ignorant of each plane's potential, but they may be consciously choosing to fly something less formidable.

What is so outlandish about my opinion of the P-51? Your own chart shows it having a lower k/d than many planes with much higher ENY. As modeled in AHII, fighting at typical MA altitudes it simply is not all that great compared to many other LW fighters. The P-47N, La-7, D9, Typhoon, and 109K have their strengths and weaknesses relative the Mustang, but they all have strengths enough to make them at least as good as the Pony in the MA bnz niche.

You can either be results-oriented and make decisions based on your personal judgment of a fair outcome, or you can make a decision based on a rational principle and let the chips fall where they may.  The former typically uses an unjustifiable procedure to reach its conclusion and therefore undermines its conclusion.  That latter uses a justifiable procedure to reach results that people don't like, but which are completely fair based on the operating principle.  You'll almost always find me in the latter camp.

Oh bloody hell Anax, rolling dice to determine ENY would be completely "fair". Making sense is more important than being fair. You can point out the flaws in a performance based approach, but they seem less severe than those of the popularity contest approach.

And if you want to continue to try a  MA stats based approach like this, drop usage entirely, and add kills/sortie and kills/time to the kills/death. If there was some way to parse fighter sorties away from attack sorties, that would be good too. I'd also be in favor of using only stats from players who rank over 1000 in fighter, if that is possible.

[BnZs]

Why do you think superiority as a dogfighter should weigh so heavily?  The 190D-9 has a better K/D ratio than the 109K-4, and a bigger share of kills in the arena.  Doesn't that make it the bigger threat?  You seem to be insisting that we evaluate ENY on the basis of something that doesn't yield the results you think it should.

Every single A/C in the game has at least one edge on the D9, due to its lousy turning abilities. The same can ot be said about the K-4.

Although, going your system, the D9 has the second highest k/d and thus should have the second lowest ENY. Right behind the P-38J 

No, the D9 is NOT the bigger threat, it is the more popular choice for the pure cherrypick/vulch/run by which practically anything fast can acquire a high k/d. A 109K flying purely by Hartmann's dicta could rack up similar numbers.

 

[Boozeman]

Although, going your system, the D9 has the second highest k/d and thus should have the second lowest ENY. Right behind the P-38J 
 

Having the 3rd highest K/D ratio means the D9 does, all things considered, on average the third best in the arena right after the 38-J and the 152. Thus, the 3 highest ENY seems reasonable.

In contrast The Spit16 may be one of the best planes on paper, but being poorly utillized as one of the noob rides most of the time, the 16 is much less of a threat on average than the D-9.

I'd really like to see ENY being based on K/D alone, updated on a daily basis.

   

[BnZs]

Having the 3rd highest K/D ratio means the D9 does, all things considered, on average the third best in the arena right after the 38-J and the 152. Thus, the 3 highest ENY seems reasonable.

Ah Boozeman, I did not in fact notice that the 152 had the second highest k/d in the list, my bad. That said, you will notice that D9 is ranked 7th in this system, despite its high K/D, and the Ta-152, with the second highest K/D, is ranked 29th!!!! Another nail in the coffin of the idea that "popularity contest" ENY ratings will have any kind of sensible result.

I must ask if vehicle kills are considered in the k/d results shown. If so, the very high k/d of the P-38 makes more sense, instead of being a product of SAPPer skill alone.

In contrast The Spit16 may be one of the best planes on paper, but being poorly utilized as one of the noob rides most of the time, the 16 is much less of a threat on average than the D-9.

Incorrect. The fact that the Spit16 is commonly used by noobs and in desperate base defense does NOT mean it is less of a threat than the D9. The relative k/d ratios mean only that the 190 is flown in a "safer" manner by the majority who fly it.

I'd really like to see ENY being based on K/D alone, updated on a daily basis.

   
So let us say we have a P-38J encountering a Spit16. At MA alts, for all intents and purposes the Spit16 is as fast, turns alot better, climbs alot better, and rolls alot better. But the P-38J apparently has had a higher k/d lately, almost certainly because it is flown mostly by experts. Thus, you argue it should have a lower ENY. So you must think the guy in what amounts to the double-inferior aircraft (P-38) should get *fewer* perk points if he shoots down the Spit16 than a Spit16 driver should get for winning a fight while flying a double-superior aircraft. Madness!

At the very least kills per hour should be factored in, if one is going to go that route.

I also don't buy the idea that people looking for a perk farming bargain will shift the ENYs around much once any system like this is established. If many people are like me, they don't much care for perk rides and perk farming isn't really on the radar. No, I predict if such a system is implemented, the P-51D will retain the "popularity" title thus artificially driving a lower ENY despite its mediocrity compared to many other rides, the Spixteen, N1K2 and La will continue to be upped by noobs and lemmings, artificially raising their ENYs, the D9 will continue to be flown by dedicated cherrypickers making it look better than it really is, and the Ta-152 will continue to be an unknown quantity to most, thus artificially raising its ENY to absurd P-40esque numbers. The P-38J will continue to be flown by experts. The reward to these experts for learning their plane and flying it well? Their Mid-War plane will have one the lowest ENY numbers in the game and they will get fewer perk points for out-flying and killing hyper-modeled LW rides. I say again, MADNESS!

The ENY system we've got has oddities, but it is far and away better than the OP's proposal

[Lusche]

Having the 3rd highest K/D ratio means the D9 does, all things considered, on average the third best in the arena right after the 38-J and the 152. Thus, the 3 highest ENY seems reasonable.

In contrast The Spit16 may be one of the best planes on paper, but being poorly utillized as one of the noob rides most of the time, the 16 is much less of a threat on average than the D-9.

I'd really like to see ENY being based on K/D alone, updated on a daily basis.

   

K/D alone as the single determining factor for ENY is as flawed as pure "usage".
There are many planes out there that aren't used much, but have a very small number of very dedicated and/or good stick artificially boosting a plane's overall K/D. See the P-38J, which has the same performance as the P-38L. But a much higher K/D

Or see the TA 152 you are giving as an example. A single player can boost the K/D significantly and thus "wrek" the ENY (and perk point gain) for other players.
In Dezember, the 152 had a K/D of 1.5. Substract Snailmans stats from this, and the K/D suddenly drops to 1.34 
Two players alone (Snailman & a1a1a1) are having more than 20% of all Ta152h kills, without them the K/D drops to 1.24 When M00t comes back, it will be even more drastic.
In my opinion, a single player shouldn't have such an impact on ENY that would result from this.

[442w30]

As a classically trained statistician their are problems with using statistics to determine eny values. Some of those problems are data that has no meaning because we cannot determine an apples to apples comparison value, i.e. Pilot skills as it relates to usage, data issues based on a/c usage, i.e. Do we use the same measurement for a 110G-2 as we do for a Yak-9?  One carries a boatload of ords, the other none.  An extension of that is how k/d relates to how the deaths occurred.  Do a lot of 110 kills come via Wirbelwind compared to Yak9...  yes I'd reckon they do. Sample size is also a problem on its own- see last paragraph.

All of those issues have been touched on in certain ways by others here.

It is possible that there is another problem. Unless it adds bails and crashes into the deaths tally, it is missing a lot of "deaths".  I feel that this would become statistically even more relevant the more an aircraft is used. This is arguably because new guys are going to log on and fly the Spit, Pony,and maybe the 109 because they have heard of those. What new guy is going to log on and up a Yak, La, C2, KI, etc? So they die again and again, many of them crashes and bails.  

I like the intent of this discussion because I feel the eny system is "off".  Several months ago I assembled a spreadsheet that lists most of the fighters in the game are rates them one 10 different statistical components. The data is from the docgonzo tests and does not include the P39s or the 262 and 163.

The data is then translated into z-scores which serve to make apples to oranges comparisons into legitimate apples to apples comparisons called Standard Scores.  The ten data fields are: Speed ASL, <10k, >10k, Climb rates ASL, <10k, >10k, Turn radius with and without flaps, 200-300 mph acceleration, and lethality.  Then those figures are added and averaged to give each aircraft a single overall score.  

Using Z-scores, 0.0 is absolute mediocrity. It is also a bell curve so the further you go from zero the greater the score difference is. For example a score of .1 is not much better than 0.0, but a score of 1.1 is actually better than an increase of .1 of a point.  A better way to rank them is by percentage.  The sheet also does customized rankings.
Bored to tears yet? 

Anax, if you are interested in seeing this and bouncing statistical ideas off me, pm me with an email addy and I will send you the sheet and instructions on how to use it.  I applaud your efforts and encourage you to continue to work at it and am more than happy to help you with the statistical side of it.  (as applies to rules and laws)

While I was typing this, Lusche posted another truism.  The data is also thrown out by lack of popularity where a certain player(s) can skew the data.   

[Anaxogoras]

Anax, if you are interested in seeing this and bouncing statistical ideas off me, pm me with an email addy and I will send you the sheet and instructions on how to use it.  I applaud your efforts and encourage you to continue to work at it and am more than happy to help you with the statistical side of it.  (as applies to rules and laws)

Thank you, I'd be happy see what you have!

All I want is improvement, and yes, I admit I can't do it alone.

[Anaxogoras]

Here's the same thing but over 6 months, from July through December:

[BnZs]

Not an improvement.

The F6F is better than any unperked F4U? 

The Typhoon is better than the La7? 

The Seafire is better than fellow carrier planes the A6M5b, and F4U-1A? 

The P-38J is better than the F4U-1A, Ki-84, and SpitVIII? Not to mention the P-38L? 

The Fw-190A8 is better than the SpitIX, 109 K-4, and P-38L? 

The Fw-190A5 is better than the P-47N, P51B, Ta-152, 109 G-2? 

Speaking of Jugs, the P-47D-40 is better than the P-47N? 

The FM2 is better than a F4U-1, SpitV, La5Fn? 

The P-40E is better than the P-38G, and both P-39s? 

The P-40B is better than the SpitI and the 109 E-4? 

Come on Anax, you're a smart guy. You can't seriously believe that these results you are getting are anything  but nonsense, you must see by now why I have always so strongly rejected any method that involves MA popularity.

[Guppy35]

Are you seriously considering dropping the P38J's ENY to 7? 

Few fly the G so it is a non issue.

The L is often flown by ground pounders/base takers and suffers a higher loss rate.

The majority of the P38 sticks fly the J making it appear it is much better than it really is.

Bah, it doesn't matter anyway... Vanscrew has a better chance of losing his virginity than this ENY system has of being implemented.

I guess that means I'm one of the "few".

Has an RAF, Battle of Britain ring to it doesn't it...

Never have so few been shot down by so many......over and over again!

[Delirium]

Maybe it is time for SAPP to fly Ta152s for a month, you'd see how flawed your setup is.

Several things should factor in when assigning ENY;

1. K/D

2. Usage

3. How the a/c fits into the MA planeset at usual MA altitude.

4. If the a/c is used properly (or improperly) in the MA.

5. If the a/c is a 'gateway' tool for those that are fairly new in AH.
__________

That said, you can move my P38J down to 5 ENY, it doesn't matter as I fly for the lowest team most of the time anyway.

[humble]

hmmmm....

The A-20 and SBD are missing? How can it be accurate with 2 of the most under rated killing machines in the game?

[Anaxogoras]

Not an improvement.

The F6F is better than any unperked F4U? 

BnZs, it's a mistake to assume that an ENY system must represent what is "better."  It might also be a mistake to assume that representing what is "better" with ENY is even desirable.

[Anaxogoras]

That said, you can move my P38J down to 5 ENY, it doesn't matter as I fly for the lowest team most of the time anyway.

You guys should be damn proud of skewing my results!

[Lusche]

BnZs, it's a mistake to assume that an ENY system must represent what is "better."  It might also be a mistake to assume that representing what is "better" with ENY is even desirable.

What is the ENY system for in your opinion? I'm not sure if your view on that matter is the same as HTC's or many other player's...

I think before getting an agreement about specific ENY numbers and how they are derived, we should have an agreemant about what ENY is for in the first place