Welcome to symthic forums! We would love if you'd register!
You don't have to be expert in bit baking, everyone is more than welcome to join our community.

You are not logged in.

Hey! If this is your first visit on symthic.com, also check out our weapon damage charts.
Currently we have charts for Battlefield 3, Call of Duty: Black Ops 2, Medal of Honor: Warfighter and Call of Duty: Modern Warfare 3

  • "Veritable" started this thread

Posts: 856

Date of registration
: Dec 8th 2013

Platform: PC

Battlelog:

Reputation modifier: 12

  • Send private message

91

Saturday, March 18th 2017, 8:59am

I would be interested in how the 2 compare with TTK 2-3 targets back to back.

If 100% hitrate and max damage / min BTK....

2 kills: RSC (4 shots) - 1.1s vs. Auto-8 (6 shots) - 0.8333333333s

3 kills: RSC (6 shots) - 1.833333333s vs. Auto-8 (9 shots) 1.333333333s

If you want to include SIPS and H-Recoil, then my Rankings are already for dealing 200 damage, so just look at the Shoot + Recover charts.

Posts: 72

Date of registration
: Mar 31st 2015

Platform: Xbox One

Reputation modifier: 5

  • Send private message

92

Saturday, March 18th 2017, 2:24pm

I would be interested in how the 2 compare with TTK 2-3 targets back to back.

If 100% hitrate and max damage / min BTK....

2 kills: RSC (4 shots) - 1.1s vs. Auto-8 (6 shots) - 0.8333333333s

3 kills: RSC (6 shots) - 1.833333333s vs. Auto-8 (9 shots) 1.333333333s

If you want to include SIPS and H-Recoil, then my Rankings are already for dealing 200 damage, so just look at the Shoot + Recover charts.


1) Does that include the necessary reload of the auto 8 after the first kill?

Two enemies: 3 bullets + reload + 3 bullets is 2.167 s to kill two targets.
Three enemies: 5 bullets + reload + 4 bullets is 2.737 s to kill three targets.

RSC beats it to two and therefore 3 kills in exchange for requiring better gunskill (penalizes misses heavily).

2) Alternatively, you could do the math as: mag dump the auto 8, switch to Mle 1903, and fire 1 bullet, which is much faster than 3-R-3: 1.1167 s. The RSC remains at 1.0975, which is ever so slightly faster than this new TTK for 2 enemies.

This post has been edited 5 times, last edit by "InterimAegis" (Mar 18th 2017, 3:08pm)


  • "Veritable" started this thread

Posts: 856

Date of registration
: Dec 8th 2013

Platform: PC

Battlelog:

Reputation modifier: 12

  • Send private message

93

Saturday, March 18th 2017, 6:05pm

1) Does that include the necessary reload of the auto 8 after the first kill?

Two enemies: 3 bullets + reload + 3 bullets is 2.167 s to kill two targets.
Three enemies: 5 bullets + reload + 4 bullets is 2.737 s to kill three targets.

RSC beats it to two and therefore 3 kills in exchange for requiring better gunskill (penalizes misses heavily).

2) Alternatively, you could do the math as: mag dump the auto 8, switch to Mle 1903, and fire 1 bullet, which is much faster than 3-R-3: 1.1167 s. The RSC remains at 1.0975, which is ever so slightly faster than this new TTK for 2 enemies.

Oh, you mean... let's do Shooting Times + Reload Times? Why not additionally add Recovery Times? Maybe introduce some Useability scores as well? Gee, if only someone included all those things? :D

PS since those Shooting Times you quoted for the RSC require shooting all 6 rounds consecutively and at max-RPM, whether or not all 6 rounds hit is no longer determined by "better gunskill," because you are not only dealing with visible H-Recoil (which is randomly L/R and thus impossible to predict), you have INvisible SIPS causing miss probability despite still being on target. Hence Hitrater. Even at range that is as close as 5m and ADS - Not Moving, Hitrater says you have probability-to-miss starting at the 3rd round fired.

1st shot: 100%
2nd shot: 100%
3rd shot: 100%
4th shot: 99.48192%
5th shot: 97.48704%
6th shot: 95.0616%

As opposed to the Auto-8 Marksman, which guarantees all 5 rounds on target 100% of the time, the only meaningful difference between their respective H-Recoil values.

For pure "gunskill" shooting, you need "optimal clickrate," which is 1-and-recover-SIPS, which gets rid of the INVISIBLE SIPS effect and only keeps the VISIBLE Recoil. For that, shoot the RSC at 0.4s per shot (150RPM) while shoot the Auto-8 at 0.2s per shot (300RPM).

-----

Well, I guess the whole point of this exercise is to analyse weapon stats and discuss how to best use them, without having to learn these (otherwise painful) lessons on the Battlefield. You see the advantages of a gun that you like, and devise scenarios where it succeeds, while others provide scenarios where it falls short. Now you can go into games, and play to its strengths while avoiding its weaknesses. Just as I have when I championed the Auto-8 Marksman way back in Origin Access.

I suppose my work here is done! 8)

Posts: 72

Date of registration
: Mar 31st 2015

Platform: Xbox One

Reputation modifier: 5

  • Send private message

94

Saturday, March 18th 2017, 6:41pm

Oh, you mean... let's do Shooting Times + Reload Times? Why not additionally add Recovery Times? Maybe introduce some Useability scores as well? Gee, if only someone included all those things?


Sure, I am well aware of and enjoy your rankings very much - but what players in this thread are reporting is that the RSC performs better than your scoring metrics would suggest. So, either you are missing some criteria (e.g. recoil decrease scale), or are unfairly biasing your results in some way (I am not sure what that would be). Accuracy is higher and KPM is higher - certainly less downtime and fewer misses, resulting more quick kills. So while I generally think your methodologies are fantastic, there is something missing here, because the RSC is most definitely not bottom of the barrel.

This post has been edited 1 times, last edit by "InterimAegis" (Mar 18th 2017, 6:52pm)


Posts: 104

Date of registration
: Oct 28th 2016

Platform: PS4

Location: California

Battlelog:

Reputation modifier: 3

  • Send private message

95

Saturday, March 18th 2017, 7:07pm

Oh, you mean... let's do Shooting Times + Reload Times? Why not additionally add Recovery Times? Maybe introduce some Useability scores as well? Gee, if only someone included all those things?


Sure, I am well aware of and enjoy your rankings very much - but what players in this thread are reporting is that the RSC performs better than your scoring metrics would suggest. So, either you are missing some criteria (e.g. recoil decrease scale), or are unfairly biasing your results in some way (I am not sure what that would be). Accuracy is higher and KPM is higher - certainly less downtime and fewer misses, resulting more quick kills. So while I generally think your methodologies are fantastic, there is something missing here, because the RSC is most definitely not bottom of the barrel.
Don't use the opinions of others to confirm your own biases. Veritable has completed the best statistical analysis of BF1's weapons that we've got. The numbers do not lie.

Of course, if he is truly missing a metric, then he will update and re-review his analysis in the future. In my subjective personal opinion however, I find that unlikely.

The RSC's successful usage from other players simply means that they've found a weapon with which they are comfortable with and can use effectively. But popular sentiments are opinion, not fact.

Posts: 72

Date of registration
: Mar 31st 2015

Platform: Xbox One

Reputation modifier: 5

  • Send private message

96

Saturday, March 18th 2017, 7:17pm

The RSC's successful usage from other players simply means that they've found a weapon with which they are comfortable with and can use effectively. But popular sentiments are opinion, not fact.


Data from weapon usage are facts. Subjective methods of scoring weapons are not facts, as good as Veritable is. His work are models designed to describe the performance capabilities of weapons. For them to be good models, they should describe the data we observe in-game. I am not referencing sentiments and opinions. I am ONLY referencing Accuracy numbers (in-game and hit rater), KPM, KPR, and TTK.

  • "Veritable" started this thread

Posts: 856

Date of registration
: Dec 8th 2013

Platform: PC

Battlelog:

Reputation modifier: 12

  • Send private message

97

Saturday, March 18th 2017, 8:39pm

Accuracy numbers (... hit rater)

Here is something you can quibble with. According to the latest version of Hitrater that I have access to, here are the 5RB 40m ADS - Not Moving hit rates for the RSC 1917 Factory vs. Auto-8 .35 Factory.

1st bullet: 100% vs. 100%
2nd bullet: 32.4926% vs. 89.3704%
3rd bullet: 28.2166% vs. 68.1331%
4th bullet: 21.2584% vs. 45.5554%
5th bullet: 15.6207% vs. 29.4876%

Average 5RB damage output: 104.721799dmg vs. 126.36767dmg

-----

Are they what Hitrater gave me after I plugged in the very data that Symthic has extracted from the BF1 files? Yes. Do I think the results are low for the RSC? Also yes. Finally, is that due to some shenanigans with Recoil Decrease During Firing that causes H-Recoil to not be simulated properly? Possibly yes, but then why was the Luger 1906 H-Recoil untouched, and Auto-8 H-Recoil BUFFED, while all other weapon classes were also not touched?

So, get me better numbers, and I'll gladly enter those into my worksheet and make new rankings. Until then, I can only work with what I got.

Posts: 72

Date of registration
: Mar 31st 2015

Platform: Xbox One

Reputation modifier: 5

  • Send private message

98

Saturday, March 18th 2017, 8:40pm

For transparency, here are my statistics. Everything is generated using hit rater and 10^6 simulations in python. This takes into account: expected accuracy as a function of range, rate of fire, bullet damage, bullet capacity, and projectile speed for 8 weapons at 4-6 different ranges. I have now completed the analysis and updated the plots.

Top 3 weapons by my analysis thus far are:
10 m: 1) Auto 8 Ext 2) 1907 F 3) RSC
20 m: 1) 1907 F 2) RSC 3) Auto 8
30 m: 1) Auto 8 and C-R 3) RSC
40 m: 1) 1906 2) Auto 8 3) RSC
50, 70+ m: 1906 is king, nothing else has a positive score

This leaves the RSC in an excellent position at all ranges under typical engagement ranges for infantry-heavy maps and modes (top 3 at every range sampled under 40 m, never #1). The Auto 8 is VERY well represented by this analysis (top 3 in 3 ranges, #1 at 30 m and 40 m), consistent with our general consensus on Symthic. The 1906 is king at long ranges (obviously). I think we as a community underrate the Cei Rigotti in CQC, and if you really don't like the above options, the Mondragon is acceptable at long range. Autoloaded 8 Extended is kind of garbage unless you are at point blank range. 1907 is very good in CQC, clearly.

The metrics are: k_eff is effective kills per reload (higher better), t_eff is effective TTK at a given range (lower better), and score is a scaled and equal combination of both. Some weapons have more data than others, I just haven't quiiiiite finished yet, but this seemed like a good time to show my work. Enjoy and rip me apart Symthic.

Here is something you can quibble with. According to the latest version of Hitrater that I have access to, here are the 5RB 40m ADS - Not Moving hit rates for the RSC 1917 Factory vs. Auto-8 .35 Factory.


Hit rater doesn't include the recoil decrease scaling between shots. This scaling helps the RSC *immensely* because it has such a low rate of fire. In order to compare the RSC, you have to estimate what the H-Rec would have been pre-patch, or otherwise you would have to recode the hit rater.

Are they what Hitrater gave me after I plugged in the very data that Symthic has extracted from the BF1 files? Yes. Do I think the results are low for the RSC? Also yes. Finally, is that due to some shenanigans with Recoil Decrease During Firing that causes H-Recoil to not be simulated properly? Possibly yes, but then why was the Luger 1906 H-Recoil untouched, and Auto-8 H-Recoil BUFFED, while all other weapon classes were also not touched?

So, get me better numbers, and I'll gladly enter those into my worksheet and make new rankings. Until then, I can only work with what I got.


Here is what I get when I estimate pre-patch recoil numbers for the RSC. The number indicates the range (ADS, not moving):
acc10 = np.array([100, 100, 100, 98.75, 96, 92.4])/100.0
acc20 = np.array([100, 99.145, 84.81, 74.86, 67.675, 61.47])/100.0
acc30 = np.array([100, 82.2, 64.4, 54.7, 49.3, 43.9])/100.0
acc40 = np.array([100, 65, 51.5, 43, 37.5, 31.4])/100.0
acc50 = np.array([100, 52.145, 42.37, 35, 27.15, 21.25])/100.0
acc70 = np.array([100, 36.9, 31, 21, 15.77, 11.5])/100.0
InterimAegis has attached the following files:
  • Unknown-2.png (69.45 kB - 3 times downloaded - latest: May 15th 2017, 6:14am)
  • Unknown.png (39.77 kB - 5 times downloaded - latest: May 15th 2017, 6:13am)

This post has been edited 13 times, last edit by "InterimAegis" (Mar 20th 2017, 3:55pm) with the following reason: Updated with final statistics


This post by "InterimAegis" (Saturday, March 18th 2017, 8:43pm) has been deleted by the author himself (Saturday, March 18th 2017, 8:45pm) with the following reason: Double post

Posts: 22

Date of registration
: Jun 22nd 2012

Platform: PC

Battlelog:

Reputation modifier: 7

  • Send private message

100

Sunday, March 19th 2017, 2:23am

The image for medic_scoring_all_move.png (usability) shows the hipfire image instead.