Jump to content


The new efficiency rating explained...

New Efficiency Rating Formula Calculations

  • This topic is locked This topic is locked
48 replies to this topic

bushwacker001 #41 Posted 17 October 2012 - 03:22 PM

    Major

  • Player
  • 27272 battles
  • 2,987
  • Member since:
    08-11-2011

View PostGoldStar, on 27 May 2012 - 12:02 PM, said:

Who makes up shit like this? You must have absolutely nothing else to do make before coming up with a E-peen formula with numbers like 150.37593984962 and 153.84615384615.!!

With a job, a wife and 2 kids, I estimate the chance I come up with this shit same the same as winning the big one in the lottery.  


I have a wife, two kids, a dog and cat, some chickens, koi and tropical fish and ladbrookes to look after and I came up with this one  254.0340205803 and this one 523.4391120543 in less than 5 seconds  :Smile-hiding:

AliceUnchained #42 Posted 17 October 2012 - 03:43 PM

    Lieutenant General

  • Player
  • 35931 battles
  • 7,457
  • [322] 322
  • Member since:
    10-18-2011

View PostDerRizerPin, on 17 October 2012 - 02:44 PM, said:

And you think kids and a wife are good things ? :D.

He did well with his estimations of the new formula.

Well they're not estimations but 99% accurate numbers really ;-) As I explained, they're divisions for sure but at the time just couldn't figure out the actual numbers. And for the purpose of providing the calculations involved more than adequate. After staring at the decimals for like 20 minutes I decided I'd rather spent my time doing something else. A simple example for GoldStar: One can write down 0.6666666666666666666667 while it simply is a division: 2/3

And just to make it clear: I did not come up with the new efficiency formula. I simply provided the calculations used (as did others).

Discussing whether efficiency is a good way to determine skill and whether people farm efficiency to look good or whether this new formula makes sense (as I said in original post I consider it a joke) was never the intention of this thread. I simply like math and messing with numbers.

CaptainMastiff #43 Posted 17 October 2012 - 03:50 PM

    Lieutenant

  • Player
  • 20114 battles
  • 1,999
  • [T4KE] T4KE
  • Member since:
    06-08-2011

View Postsharpneli, on 17 October 2012 - 03:18 PM, said:

I have to agree with Dushan here. I haven't really seen anyone with 1600 eff or so intentionally boosting it, thus it's not really likely based on my observations. It's the higher effs which are usually skewed one way or another (e.g seen someone with over 2k eff with 1.3k battles with the old Könich VK3601H). Though I also know a lot of players with high eff that are simply just that awesome.


I'll give another example where I think Eff rating is a bit unfair... Compare T32/IS3/KT/AMX100, the BL-9 has the highest average damage. It was my 2nd Tier 8 after King Tiger (but I only played the KT briefly before the IS-3) I completed the grind to IS-4 and then subsequently did the T32/KT shortly afterwards. I felt like I knew a lot more about the game and was a much better player whilst playing the T32 and KT because of the IS-3 grind... however statistically the IS-3 was by far a much better tank Eff Rating/WR/AVD. To me that isn't because I was a better player but because the tank isn't balanced when it comes to Eff rating.

Its another example of a tank that you can easily play to grind up the eff rating, with regards to the VK3601 last I knew that if you use the Gold gun (7.5 Konisch) that thing is an absolute beast, it wouldn't surprise me if was using gold with that gun but there is equally the chance he was just a damn good player.

It doesn't matter how you look at it the efficiency rating is so exploitable. The problem I see is that some people might take it as a guideline on how to be a better player, capping the base isn't always the best option. I've seen so many times when people have tried to cap but been sniped to death when really if they had just hunted the last 2 down they would of won with ease or vice versa.

Edit: I'm not saying all with 1,600 do boost it but just take a sample amount and look through. Look at certains tanks and compare them to their equivalent tier. Shoot me now if you want but I'm going to say I find more people who have played the Russian line have a higher Eff rating than those who have played mixed/other lines (I'm talking with people like 1,000+ not those guys with like 34% win ratios and shit). I honestly think having those high alpha damage guns just ruins the balance of eff rating. We can see average damage has a direct effect on your efficiency rating also and in a capable player it is quite easy to attain a higher average damage in the ruskie tanks than other nations.

Edited by CaptainMastiff, 17 October 2012 - 03:52 PM.


mondog #44 Posted 17 October 2012 - 03:58 PM

    Lieutenant

  • Player
  • 14590 battles
  • 1,931
  • [-MM] -MM
  • Member since:
    03-17-2011
It's flawed because it does it by class based on one dudes opinion. It doesn't take into account tier x mediums are damage dealers and that different mediums at different tiers and tech trees have purposes outside of what he thinks a medium should be and thus gains efficiency. Same applies to arty, my Su26 stats are ludicrously good but my efficiency is only 1300. Same for my 105AM.

The same applies to heavies getting cap points. A heavy getting cap points should get a ton of efficiency. It means a successful push! But that isn't how the guy who does efficiency interprets it.

The efficiency system thus is flawed so take it all with a pinch of salt.

Xensation #45 Posted 17 October 2012 - 04:05 PM

    Lieutenant General

  • Clan Commander
  • 15157 battles
  • 7,703
  • [IDEAL] IDEAL
  • Member since:
    06-08-2012
http://forum.worldof...not-this-again/
First part is about the "new" efficiency rating formula, which is not up to date anymore. There will be a new one working with logarithms to cap the amount of efficiency rating one can gain with capturing, defending and spotting. And I know that for sure, because the developer of the efficiency rating formula has told me that.

The relevant part:

View PostXensation, on 16 September 2012 - 05:09 PM, said:

Things that are wrong with ER2:
There are three problems with the ER2 caused by the categorization of tank classes.

Problem #1:
How do you reward the different tank classes? Should you get more ER for things you are supposed to do with your tank? Or should you get rewarded for things you are not suppsosed to do, but you still do well, because you have to make up for your team's failure?
Let's say you are a light tank driver, which means spotting the enemies is your main job. Now here's the problem. If the formula rewards you for spotting only there is no motivation to do the job your teammates are supposed to do. E.g. a formula rewarding scouts for their average spottings per battle means they gain much less ER for the damage they deal and the tanks they destroy, which does not make sense at all, because dealing thousands of damage and getting lots of kills with your light tank would gain you less ER than simply rushing in and spotting all 15 enemies.
However, if you do it the other way around and make players gain more ER for things they are not supposed to do with their tank class, it's rewarding bad gameplay, too. For instance, an IS-7 player who rushes straight into the enemies and spots 15 tanks would get a higher ER than a scout doing exactly the same thing. As you can see, both versions are totally flawed and can not be implemented.

Problem #2:
Not every tank in the same category can do the same job. A perfect example is the Super Pershing. It is categorized as a medium tank, even though it plays like a heavy tank. It can't spot, it can't flank, it can't do anything mediums are supposed to do. However, ER2 can't tell the difference between a super pershing and any other medium tank, hence playing the super pershing would result in a terrible rating even if it's played like it's supposed to be played.
Another good example is the AMX 13 90. It is a light tank, even though it's no typical scout like the T-50-2. It is rather a support unit and a very fast and small medium tank, thus it should get rewarded for other things than spotting, too. However, ER2 can't tell the difference between the AMX 13 90 and any other LT, thus it will reward players playing it wrong (= like a T-50-2) instead of players playing it right (= like a fast support unit/medium tank).

Problem #3:
The calculator does not know which statistics are from which tank.
The easiest way to explain this is a simple example.
Let's say we have two players. Player1 rushes with his tank destroyer and spots 10 players. In the next battle, he deals 1000 damage with his light tank. The current formula for ER2 would rate the light tank with 396 and the tank destroyer with 200, because neither did what it is supposed to do. However, the players profile shows the overall statistics only and not the statistics for each tank, which means the ER2 calculator will use the values 500 for average damage dealt, 5 for spotted and 50% for tank destroyers and light tanks played, which results in a ER of 887. This means a player gets a much higher rating than he deserves, because he failed to do his job.
In contrast to that, let's look at player2 who spots 10 players with his light tank and deals 1000 damage with his tank destroyer. The light tank that spots 10 enemies gets an individual rating of 2200 and the tank destroyer gets an individual rating of 751 (average tier of 8 has been used for the calculation, thus 1000 damage is not very rewarding).
As you can see player2 should get a much better rating than player1, because he does what he is supposed to do and deserves a higher rating. However, the calculator gives both players an overall result of 887, even though player1 failed horribly and player2 did well!

A quote related to the problems explained above:

View PostDewirix, on 13 September 2012 - 09:33 AM, said:

The "new" formula is based on the idea that tanks have specific roles and tries to award points for playing those roles. Unfortunately, it can't distinguish which tanks you got your spots, damage, kills and cap points in, so it just takes the percentage of battles you've had in each and multiplies its class-specific formula by this.
For example, if you play a Chaffee and a StuG III for 50 games each, getting 200 spots in the Chaffee and doing 50,000 damage in the StuG, you'll get penalised because the formula will assign half of the spots to the TD part of the formula (where spots are worth 20 points) rather than the light tank part of the formula (where spots are worth 220 points).
Simliarly, half the damage you did in the StuG is assigned to the light tank part of the formula, (damage worth 0.63/point at tier 5) rather than the TD part (damage worth 1.2/point at tier 5).
In the example above, the "new" formula gives an ER of 699, but the ER for the separate tanks is 880 (Chaffee) and 1201.75 (StuG). Thus, the average ER is 1041: considerably higher than the overall ER. The equivalent figures for the normal ER would be a global rating of 650; 800 (Chaffee); and 500 (StuG).
In short, the new formula rewards consistency over breadth of experience - if you play mostly a single tank class then you'll get a rating that better reflects your performance than had you played a range of different classes.

To sum it up: There are three simple reasons why the categorization in the ER formula does not make sense. Firstly not every tank is typical for its class, secondly there's a conflict in the reward system and thirdly wargaming does not shot stastics of individual tanks, which means a categorization like in the ER2 can't work.


Hippopotamus_Rex #46 Posted 17 October 2012 - 04:08 PM

    Second Lieutenant

  • Player
  • 20607 battles
  • 1,377
  • [ETOH] ETOH
  • Member since:
    04-10-2011
I think the whole philosophy of awarding/penalizing some tank categories for different things, i.e. awarding light tanks for scouting more then the rest is flawed.

Light tanks should have exactly the same efficiency for spotting like anyone else. They will have more spots anyway because of what they are, so they will be more efficient on that field. They should be awarded the same for damaging enemies like any others - they will usually do less damage anyway, because what they are. No additional bonuses for what your tank is "intended to do" are needed, because you will get more efficiency for this anyway, you are diesigned for it, as much you will get less than others from the activities you are not designed for. The same goes for TD's - they are intended to do damage, so they will be effcient on that field anyway, no additional bonus needed.

Such role assignment is flawed because of many reasons: first, there is problem with hybrid tanks like IS-8 - med or heavy? Or speedy American TD's? The other is that sometimes you just have to do something you are not intended to do to save the day. You should not be penalized because you are flexible (and in effect efficient as player) and can fulfill the role your tank was not supposed to serve.

Also, sheer spotting should be less important than damage to spotted enemies, if it could be incorporated in any way.

And I would like capturing have much less weight in the formula - many capping points are not really necessary to win a match and are really easy to exploit, while most defense points can be crucial.

The way I see it damage should have most weight, both your and while spotting, after that spotting, kills and defense, capturing as the least important in the formula.

Dan_Abnormal #47 Posted 17 October 2012 - 04:12 PM

    Brigadier

  • Player
  • 31886 battles
  • 4,415
  • Member since:
    10-14-2011

View PostCaptainMastiff, on 17 October 2012 - 03:50 PM, said:

I'll give another example where I think Eff rating is a bit unfair... Compare T32/IS3/KT/AMX100, the BL-9 has the highest average damage. It was my 2nd Tier 8 after King Tiger (but I only played the KT briefly before the IS-3) I completed the grind to IS-4 and then subsequently did the T32/KT shortly afterwards. I felt like I knew a lot more about the game and was a much better player whilst playing the T32 and KT because of the IS-3 grind... however statistically the IS-3 was by far a much better tank Eff Rating/WR/AVD. To me that isn't because I was a better player but because the tank isn't balanced when it comes to Eff rating.

My eff with IS-3, 1245. My eff with KT, 2150. I don't think the IS-3 is more or less able to farm eff than any other tier 8 heavy.

Armo #48 Posted 17 October 2012 - 04:50 PM

    Warrant Officer

  • Beta Tester
  • 9599 battles
  • 573
  • Member since:
    09-12-2010

View PostCaptainMastiff, on 17 October 2012 - 01:55 PM, said:




Too many ways to farm it easily and when I come across anyone with ~1,000 rating that is when I consider that there is a chance of them being a good player, 1200+ I think they are a good player. 1600+ and they've likely farmed it intentionally but there are people who are naturally that good

I also believe your opinion on this is flawed. I took a 6 month break from wot, when I returned the efficiency rating had been devised. I found mine, it was 1910.

How could I farm it without any prior knowledge of it?

My RL friend whom I platoon with something like 60% of the time had a rating of something like 1680 when he came back from his break. He's just a good player.

Playing certain tanks does increase it though, my 300 ISU games came in a relatively short space of time and added +20 to my eff without thinking about it.

Hippopotamus_Rex #49 Posted 17 October 2012 - 05:52 PM

    Second Lieutenant

  • Player
  • 20607 battles
  • 1,377
  • [ETOH] ETOH
  • Member since:
    04-10-2011

View PostArmo, on 17 October 2012 - 04:50 PM, said:

Playing certain tanks does increase it though, my 300 ISU games came in a relatively short space of time and added +20 to my eff without thinking about it.

That's so true. It's very individual both on player and vehicle level. (but please note, that it works both way, playing something you don't like/stock/shit MM can reduce your eff as well).

Some vehicles have favorable match making, others just fits your style of play better. In my case ISU did not make such impact, while in the same time I think it is very good vehicle - lower than my average efficiency (1303eff/375 battles) on this is only my own fault. Individual experience with particular vehicle is also essential.

In the same time I really like E75, it can shine and quite often can be top tier - 1881eff/821 battles. I bet most Type59 have significantly higher eff than their drivers average (maybe with the exception of those good players, who are efficient with just everything they drive :))

It is even more seen on artillery, when spotting and capping is significantly less important:

GWPanther - 1339eff/658 battles
GWTiger - 1021eff/378 battles

Both mean that I am not so good artillery player, but difference between them is significant, and it perfectly matches my, hm, "gameplay experience". I really liked GWPanther, it's mobility, it's good aiming capabilities with this semi-tower, while playing immobile, huge, visible from afar and pain-to-aim GWTiger in randoms is horror for me, and huge damage cannot compensate this.

All in all, I think there is much more significance behind these funny numbers and room for interpretation, than most of users are aware of. :)

Edited by Hippopotamus_Rex, 17 October 2012 - 10:48 PM.





1 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users