Cost Of Accuracy

tehmoriz

Respected

(2,523)

Posts: 1,487

Date of registration
: Jul 20th 2012

Platform: PC

Battlelog:

Reputation modifier: 14

1

Wednesday, August 13th 2014, 11:39pm

In BF4, faster shooting guns are less accurate, while slower shooting guns are more accurate. The transaction is then clearly that by selecting a slower shooting gun, we trade damage for accuracy. The amount of damage being traded is easy to calculate; the amount of accuracy we gain is not.

For this trade to make sense, this MUST be true:

“The amount of accuracy we gain should translate to a big enough advantage in TTK at long range, that perfectly offsets the amount of TTK we give up in CQB.”

The Test Subjects:

I’m going to consider two non-bullpup carbines, each on the opposite side of the scale: the G36C and ACWR. The G36C is the most accurate carbine (and likely the most accurate automatic weapon), whereas the ACWR is one of the fastest shooting carbines with clearly CQB oriented stats.

CQB performance:

Within 8 meters, both weapons have guaranteed 100% accuracy to the torso with a 5 round burst. As such, TTK has no accuracy component.

ACWR: 4*60/880 = 272 ms
G36C: 4*60/650 = 369 ms

Difference: 97 ms in favor of ACWR

Ranged (50m) performance:

At 50 meters, the two weapons have clear accuracy differences. I’m going to simulate weapon accuracy using hitrater (Hitrate simulator)

with a 2 round burst to the body, the G36C can expect an average of 1.932 hits, whereas the ACWR can expect 1.742 hits. If we allow spread to completely settle between bursts, this means that with the G36C, we can expect to need 7/1.932*2 = 7.25 bullets to kill at 50 meters, and the ACWR 7/1.742*2 = 8.04 bullets to kill at 50 meters.

7.25 bullets is a bit more complicated to calculate for. But what it does say, is that ¾ of the time, we need 7 bullets to kill, and ¼ of the time, we need 8. I’m just going to use 8 for the ACWR because of how close it is.
NOTE: I’m going to assume perfect shooting, since imperfect shooting is really hard to simulate. There’s a clear ease of use difference between the two guns.

So the accuracy-weighted TTKs of the two weapons look like this:

G36C: ¾*(6*60/650) + ¼*(7*60/650) + 3*(2*0.085/15) = 611 ms
ACWR: 7*60/880 + 3*(2*0.114/15) = 523 ms

Difference: 88 ms in favor of ACWR

Conclusion:

The G36C’s accuracy advantage ISN’T strong enough to make it kill faster at long range. In fact, its accuracy advantage is basically meaningless.

EDIT: corrected mistake in formula about spread reset. was resetting to 0.0, where it should be resetting to 0.3. derp.

This post has been edited 2 times, last edit by "tehmoriz" (Aug 19th 2014, 4:40pm) with the following reason: forgot that spread recovery needs to cover two bullets, not just one.

Rezal

Rezmer

(10,393)

Posts: 4,259

Date of registration
: Apr 6th 2012

Platform: PC

Location: From the heart of Europe.

Battlelog:

Reputation modifier: 17

2

Wednesday, August 13th 2014, 11:44pm

Conclusion 2: Shoot longer bursts

[Aristocrat's Shoes]

Quoted from "Darktan13"

TLDR -
Teamwork is where players function by themselves, but their effectiveness is multiplied when they work together. Not a checklist of "did we bring a healer so we can start playing?"

tehmoriz

Respected

(2,523)

Posts: 1,487

Date of registration
: Jul 20th 2012

Platform: PC

Battlelog:

Reputation modifier: 14

3

Wednesday, August 13th 2014, 11:55pm

Quoted from "Rezal"

Conclusion 2: Shoot longer bursts

2 round burst is the most accurate way to shoot both weapons. though if we use longer burst lengths, the ACWR is more negatively affected, to the point where a G36C can kill at 50 meters with a single 12 rounds magdump, while the ACWR needs 28 rounds.

Rezal

Rezmer

(10,393)

Posts: 4,259

Date of registration
: Apr 6th 2012

Platform: PC

Location: From the heart of Europe.

Battlelog:

Reputation modifier: 17

4

Wednesday, August 13th 2014, 11:58pm

Well, my G36C kills in 669 ms at 60 meters with 4rb...

But I may be calculating differently.

[Aristocrat's Shoes]

Quoted from "Darktan13"

TLDR -
Teamwork is where players function by themselves, but their effectiveness is multiplied when they work together. Not a checklist of "did we bring a healer so we can start playing?"

This post has been edited 1 times, last edit by "Rezal" (Aug 14th 2014, 12:04am)

JSLICE20

(2,444)

Posts: 3,291

Date of registration
: Apr 26th 2013

Platform: PS4

Location: Arizona, USA

Reputation modifier: 15

5

Thursday, August 14th 2014, 12:22am

Brute-force approach to BF4 gun balance & optimal attachments, burst sizes, and aim targets

According to these charts, with perfect compensation and an inter-burst interval of 100ms, the G36C is most optimally fired in 2-shot bursts at 60m.

To Aim Assist or not to Aim Assist, that is the question.

Nope. Aim Assist or bust; here's why:

Default Aim Assist Data

Source code

AccelerationInputThreshold 0.98
    AccelerationMultiplier 5.0
    AccelerationDamping 4.0
    AccelerationTimeThreshold 0.15
    SquaredAcceleration 0.0
    MaxAcceleration::Vec2
        x 2.0
        y 2.0
    YawSpeedStrength 1.0
    PitchSpeedStrength 1.0
    AttractDistanceFallOffs::Vec2
        x 1.0
        y 1.2
    AttractSoftZone 0.75
    AttractUserInputMultiplier 0.45
    AttractUserInputMultiplier_NoZoom 0.5
    AttractOwnSpeedInfluence 0.0
    AttractTargetSpeedInfluence 0.85
    AttractOwnRequiredMovementForMaximumAttract 0.0
    AttractStartInputThreshold 0.1
    AttractMoveInputCap 0.0
    AttractYawStrength 1.0
    AttractPitchStrength 0.34
    MaxToTargetAngle 45.0
    MaxToTargetXZAngle 45.0
    ViewObstructedKeepTime 0.0
    SnapZoomLateralSpeedLimit 1000.0
    SnapZoomTime 0.2
    SnapZoomPostTimeNoInput 0.2
    SnapZoomPostTime 0.2
    SnapZoomReticlePointPriority 999
    SnapZoomAutoEngageTime 0.0
    SnapZoomBreakTimeAtMaxInput 0.2
    SnapZoomBreakMaxInput 0.2
    SnapZoomBreakMinAngle 90.0
    SnapZoomSpamGuardTime 1.2
    SoldierBackupSkeletonCollisionData *nullGuid*
    CheckBoneCenterOnlyDistance 40.0

No Slowdown Data

Source code

AccelerationInputThreshold 0.98
    AccelerationMultiplier 5.0
    AccelerationDamping 4.0
    AccelerationTimeThreshold 0.15
    SquaredAcceleration 0.0
    MaxAcceleration::Vec2
        x 2.0
        y 2.0
    YawSpeedStrength 1.0
    PitchSpeedStrength 1.0
    AttractDistanceFallOffs::Vec2
        x 1.0
        y 1.2
    AttractSoftZone 0.0
    AttractUserInputMultiplier 1.0
    AttractUserInputMultiplier_NoZoom -1.0
    AttractOwnSpeedInfluence 0.0
    AttractTargetSpeedInfluence 0.0
    AttractOwnRequiredMovementForMaximumAttract 0.0
    AttractStartInputThreshold 0.0
    AttractMoveInputCap 0.0
    AttractYawStrength 0.0
    AttractPitchStrength 0.0
    MaxToTargetAngle 45.0
    MaxToTargetXZAngle 45.0
    ViewObstructedKeepTime 0.0
    SnapZoomLateralSpeedLimit 1000.0
    SnapZoomTime 0.2
    SnapZoomPostTimeNoInput 0.0
    SnapZoomPostTime 0.0
    SnapZoomReticlePointPriority 999
    SnapZoomAutoEngageTime 0.0
    SnapZoomBreakTimeAtMaxInput -1.0
    SnapZoomBreakMaxInput 0.2
    SnapZoomBreakMinAngle 90.0
    SnapZoomSpamGuardTime 0.5
    SoldierBackupSkeletonCollisionData *nullGuid*
    CheckBoneCenterOnlyDistance 40.0
    DisableForcedTargetRecalcDistance 7.0

No Auto Rotation Data

Source code

 AccelerationInputThreshold 0.98
    AccelerationMultiplier 5.0
    AccelerationDamping 4.0
    AccelerationTimeThreshold 0.15
    SquaredAcceleration 0.0
    MaxAcceleration::Vec2
        x 2.0
        y 2.0
    YawSpeedStrength 1.0
    PitchSpeedStrength 1.0
    AttractDistanceFallOffs::Vec2
        x 1.0
        y 1.2
    AttractSoftZone 0.75
    AttractUserInputMultiplier 0.45
    AttractUserInputMultiplier_NoZoom 0.5
    AttractOwnSpeedInfluence 0.0
    AttractTargetSpeedInfluence 0.85
    AttractOwnRequiredMovementForMaximumAttract 0.0
    AttractStartInputThreshold 0.1
    AttractMoveInputCap 0.0
    AttractYawStrength 1.0
    AttractPitchStrength 0.34
    MaxToTargetAngle 45.0
    MaxToTargetXZAngle 45.0
    ViewObstructedKeepTime 0.0
    SnapZoomLateralSpeedLimit 1000.0
    SnapZoomTime 0.2
    SnapZoomPostTimeNoInput 0.0
    SnapZoomPostTime 0.0
    SnapZoomReticlePointPriority 999
    SnapZoomAutoEngageTime 0.0
    SnapZoomBreakTimeAtMaxInput -1.0
    SnapZoomBreakMaxInput 0.2
    SnapZoomBreakMinAngle 90.0
    SnapZoomSpamGuardTime 0.5
    SoldierBackupSkeletonCollisionData *nullGuid*
    CheckBoneCenterOnlyDistance 40.0
    DisableForcedTargetRecalcDistance 7.0

Prepare your laughbox

Quoted from "Zer0Cod3x"

the Sebstalder is quiet good since it can 3hit kill at any distanc ,but In my opinion i actually thikn the sweeper is better, its got a really really fast firerate that can beat alll those Noobmaticos, Helregall adn shitguns in close quarters , and its also really accurate out to like l;ong range,. overall great allround gun, jsut my 2$ tho

My "Contributions"

Gun Sway Modifiers (Hidden Attachments in Variants)
Gadget, Stationary, and Vehicle Weaponry Stats (Pre-Official)

tehmoriz

Respected

(2,523)

Posts: 1,487

Date of registration
: Jul 20th 2012

Platform: PC

Battlelog:

Reputation modifier: 14

6

Thursday, August 14th 2014, 12:25am

i just fixed an error in my formula, which lowered the TTK values so they're aligned with rezal's. however, the difference in TTK between the two weapons didn't change.

either way, the point is, that the G36C should be able to kill faster than the ACWR. however, even under the perfect circumstances, it doesn't kill any faster.

ToTheSun!

Be Creative.

(5,050)

Posts: 7,811

Date of registration
: Mar 9th 2012

Platform: PC

Location: Portugal

Reputation modifier: 19

7

Thursday, August 14th 2014, 12:26am

Most of the players kind of intuitively reached that conclusion all by themselves.

Zormau

(851)

Posts: 1,400

Date of registration
: Jun 23rd 2012

Platform: PC

Location: Germany, Bavaria

Battlelog:

Reputation modifier: 12

8

Thursday, August 14th 2014, 12:31am

Quoted from "ToTheSun!"

Most of the players kind of intuitively reached that conclusion all by themselves.

Intuition is worth nothing compared to stats on symthic, as you are most likely well aware.

Zormau - Battlelog / Battlefield 4

Memorable quotes not taken yet:

Quoted from "NoctyrneSAGA"

Quoted from "Nick 30075"

Of course, this ignores the non-constant cross-sectional first moment of area across the chest as well as non-constant material properties of the boob; it would be difficult to perform a more detailed analysis (as in, I'd have to have a shape function AND I'd need to derive a function for elastic modulus as a function of lateral breast coordinate) but whatever. It's 2am and I'm lazy.

I always believed science should be very hands on experience.

Quoted from "NoctyrneSAGA"

You should also answer this question I had posed in that thread: Would you be willing to pay your surgeon more if he was going to use a chainsaw for the opening incision of surgery? Clearly using a chainsaw isn't truly suited for surgery but that doesn't really matter. If he's "skilled" enough to be able to use the wrong tools of the trade, he should be rewarded for that skill right?

Maveco

You can't stop the Smooze!

(610)

Posts: 2,351

Date of registration
: Mar 16th 2013

Platform: 360

Location: ~/

Battlelog:

Reputation modifier: 12

9

Thursday, August 14th 2014, 12:31am

For n = 2 is done, what about the other burst lengths?

ToTheSun!

Be Creative.

(5,050)

Posts: 7,811

Date of registration
: Mar 9th 2012

Platform: PC

Location: Portugal

Reputation modifier: 19

10

Thursday, August 14th 2014, 12:37am

Quoted from "Zormau"

Intuition is worth nothing compared to stats on symthic, as you are most likely well aware.

Intuition is worth a lot more than one would have guessed, had one been blinded by the thought of how wholly meaningful math is.

In fact, the math has always supported such intuition, but the conclusion came faster to those more accepting of mother experience.

Cost Of Accuracy

Nope. Aim Assist or bust; here's why:

Default Aim Assist Data

Source code

No Slowdown Data

Source code

No Auto Rotation Data

Source code

Prepare your laughbox

My "Contributions"

Memorable quotes not taken yet:

Similar threads