Carbineer Archive
Thread: The new speed formula:
TAfirehawk wrote:
ROFLMAO, of course I estimated WHAT IT SHOULD DO, but there is little proof what happens beyond the Examine Window on LIVE right now.
Fails on rounding, don't think so.......you should check your data against playing on LIVE, not TC old data from 3 weeks ago.
Actually I did test Rifle Speed which goes to +135 and you numbers are off, just checked on TC-EP3. We can argue all day, but the fact is that going much above +100 Speed in any prof is a waste of time. Please look at Ben's post again.
I wish you had access to the CU Alpha Forum where this was all properly discussed.
You ESTIMATED, then told me my model was off because it didn't agree with your estimations. BTW you should never extrapolate anything and take it as any kind of proof, you should extrapolate in order to validate your theory against future tests. You are quite good at math but a little too undereducated to really understand what you are doing.
Did you ever try yours over 135? Your model contains real numbers up to 125 so it should still be decentlyaccurate at 135 but with an increasing error, just like in the picture i posted. You will get numbers that are off when trying speeds of 150, trust me.
Did you do a plot of yourresiduals to look for systematic flaws? I did, completely random and between -5e-3 and +5e3, which rounding errors would account for.
Zhorx wrote:
TAfirehawk wrote:
ROFLMAO, of course I estimated WHAT IT SHOULD DO, but there is little proof what happens beyond the Examine Window on LIVE right now.
Fails on rounding, don't think so.......you should check your data against playing on LIVE, not TC old data from 3 weeks ago.
Actually I did test Rifle Speed which goes to +135 and you numbers are off, just checked on TC-EP3. We can argue all day, but the fact is that going much above +100 Speed in any prof is a waste of time. Please look at Ben's post again.
I wish you had access to the CU Alpha Forum where this was all properly discussed.
You ESTIMATED, then told me my model was off because it didn't agree with your estimations. BTW you should never extrapolate anything and take it as any kind of proof, you should extrapolate in order to validate your theory against future tests. You are quite good at math but a little too undereducated to really understand what you are doing.
Did you ever try yours over 135? Your model contains real numbers up to 125 so it should still be decentlyaccurate at 135 but with an increasing error, just like in the picture i posted. You will get numbers that are off when trying speeds of 150, trust me.
Did you do a plot of yourresiduals to look for systematic flaws? I did, completely random and between -5e-3 and +5e3, which rounding errors would account for.
And what data do you have past +135? Like I said, that data you linked to is about as old as dirt. And the +135 line in my data shouldn't be estimated as I got that hard number verified in Rifle Speed once I switched from just looking at the data in Carbineer.
Neither of us has the LIVE test data of actual firing speed to check our numbers, but if your formula is correct then the Devs sure have screwed this up again because the diminishing returns is gone. Let us hope your formula is wrong or the whole Speed Game will be no better than pre-CU.
Yes I looked at R squared and it matches within 99.98%, but as you have pointed out, I had to estimate data since no ACCURATE data is available.......I estimated data based on what the Devs told me, but if you can provide some REAL data I would be happy to update my formula. But like I said, the reason I didn't post this again in the last 2 weeks is TC data is invalid and LIVE data hasn't been gotten yet.
Zhorx wrote:
TAfirehawk wrote:
Like I said in my last post, get some new data as that Swordsman stuff is over 3 weeks old.
Funny story: My formula was made from those data, they fit all your more recent data, and every single data i have found so far.
Your furmula fit data up to 125, then it is off.
Where is your data above +135?
Where is your actual firing speed timed via the in-game timestamp or stopwatch?
Who says the Examine Window is the actual firing speed?
Who says there still isn't a hard cap for speed on a weapon type basis to flatten out the curve for zero speed improvements for those uber high player mods?
Like I said before, we can argue all day around false information (which we both have).....please listen, there is a reason I didn't bring up this topic since the CU went LIVE and why nothing is linked in the FAQ.....nobody has gotten ANY accurate test data for actual firing speed and using the Examine Window doesn't cut the mustard on this one.
TAfirehawk wrote:
And what data do you have past +135? Like I said, that data you linked to is about as old as dirt. And the +135 line in my data shouldn't be estimated as I got that hard number verified in Rifle Speed once I switched from just looking at the data in Carbineer.
The data i have past 135 is the one in the link. All data, including that, have so far followed the formula.
Neither of us has the LIVE test data of actual firing speed to check our numbers, but if your formula is correct then the Devs sure have screwed this up again because the diminishing returns is gone. Let us hope your formula is wrong or the whole Speed Game will be no better than pre-CU.
Diminishing results are by no means gone, look at the formula again and in the polots i provided look at the y-axis. I did not plot down to 0 in order to increase the resolution in the interesting area. With my formula you have diminishing return for all values of the speed. Any polynomial would fail in this aspect.
Yes I looked at R squared and it matches within 99.98%, but as you have pointed out, I had to estimate data since no ACCURATE data is available.......I estimated data based on what the Devs told me, but if you can provide some REAL data I would be happy to update my formula. But like I said, the reason I didn't post this again in the last 2 weeks is TC data is invalid and LIVE data hasn't been gotten yet.
Looking at R squared is no good way of validating a model. You need to make a residual plot and a cross reference plot for variables in case they describe the same effect.I did both and found no significant correlations between my modelling variables or any systematic error in the residuals.
Again, please post ANY data from LIVE above +135.....and remember even with that there is no indication of any REAL effect above +100/+125. For all we know the Character Sheet and Examine Window are using different formula and SEA's/Buffs/FS Mods display but have no real effect. It is a lot more than a numbers game, it requires LIVE data and if you base all of your info on that old Swordsman post, well it has no more validity than my data.
No, the difference is my model so far never failed. Yours did on the old data. Live is a different story and the actual use you have of the weapon speed is quite unclear to me.
All I claim is the Speed Formula is SUPPOSED to be the shape of my graph (per the Devs), not yours.....but I am still waiting for ANYBODY to provide REAL data from LIVE.
Yours has diminishing results up to about 200, then you get increased returns. My modelhas diminishing results forall values of the speed. You can validate this by calculating the secondderivative of the inverse speed (proportional to the DPS). It will be negative for all numbers in my case, yours will be positive after 200something.
This is funny, I am on Chilastra right now and BOTH OF OUR FORMULA are wrong.
+120 Speed with a 2.25 Laser Carbine comes up with 1.37 on the Examine Window.....my formula says 1.362 and yours is 1.353
All this discussion and asI figured all along, we are both wrong. Of course I still maintain the diminishing returns 'feels' like it happens below +125 and thus follows my graph, but obviously neither of us has accurate data ROFLMAO.
TAfirehawk wrote:
This is funny, I am on Chilastra right now and BOTH OF OUR FORMULA are wrong.
+120 Speed with a 2.25 Laser Carbine comes up with 1.37 on the Examine Window.....my formula says 1.362 and yours is 1.353
All this discussion and asI figured all along, we are both wrong. Of course I still maintain the diminishing returns 'feels' like it happens below +125 and thus follows my graph, but obviously neither of us has accurate data ROFLMAO.
+120 Speed with a 2.25 Laser Carbine comes up with 1.37 on the Examine Window.....my formula says 1.362 and yours is 1.353
+123 Speed with a 2.25 Laser Carbine comes up with 1.37 on the Examine Window.....my formula says 1.357 and yours is 1.345
TAfirehawk wrote:
This is funny, I am on Chilastra right now and BOTH OF OUR FORMULA are wrong.
+120 Speed with a 2.25 Laser Carbine comes up with 1.37 on the Examine Window.....my formula says 1.362 and yours is 1.353
All this discussion and asI figured all along, we are both wrong. Of course I still maintain the diminishing returns 'feels' like it happens below +125 and thus follows my graph, but obviously neither of us has accurate data ROFLMAO.
Sure it wasn't +115 speed?
Doc Buff for Speed doesn't change anything on the Character Sheet or Examine Window of the weapon.....but it speeds you up is the report so no way to track it....
And the difference in Examine Windows numbers up there is probably from being over the +25 SEA cap. The Character Sheet just like the Skill Sheet pre-CU shows more than +25 SEA gains when in reality it has no effect.
Anyway, as the data shows, both formula are wrong 
Zhorx wrote:
TAfirehawk wrote:
Doc Buff for Speed doesn't change anything on the Character Sheet or Examine Window of the weapon.....but it speeds you up is the report so no way to track it....
And the difference in Examine Windows numbers up there is probably from being over the +25 SEA cap. The Character Sheet just like the Skill Sheet pre-CU shows more than +25 SEA gains when in reality it has no effect.
Anyway, as the data shows, both formula are wrong
*Bangs hishead against the keyboard and realized he just spent valuable time of his life discussing with a *edit**
You can't use data like this...ever figured the reason why the numbers are off seems to be because YOU ARE OVER THE+25 CAP!!! For gods sake...
Like MAGIC when you used the speed stim the speed decreased, not by adding additional SEAs beyond the +25 cap. I can't fathom why. Can you?
Now, that 120 number. That is with +30 in SEAs right? So the actual number is 115...
Actually I was below the +25 on most of those I think but as I posted after I was done and logged out, there might be problems with my data (BUT AT LEAST I TRIED TO PROVIDE DATA FROM LIVE). Once again I ask for in-game data to verify your findings, if somebody could come up with that then we might prove your formula but at this point it is just made up THEORY like mine.
And at the end of the day, what does your formula do that the Examine Window doesn't? And what does it really say about the ACTUAL FIRING SPEED of a specific person with a specific weapon using a specific attack against a specific target? We can all make BS statements and arguements about totally useless data and formula......which AGAIN is why I didn't feel it necessary to publish the data/formula I had until you brought this up. And your formula doesn't reflect the actual play experience either. So the bottom line is neither formula works or tells us anything beyond what the Examine Window already gives players....and hell that is quite possibly wrong as many display things are wrong on LIVE right now.
Quite simply you have given us no more proof about your formula than I have about mine, except I admitted 3 weeks ago when I came up with this that it had limitations and assumptions from TC.
I just re-ran the numbers in-game with just +25 in SEA's this time for sure....up to 140 your formula is correct per the Examine Window. My formula breaks down and is wrong......I should have verified it with LIVE data instead of recent TC data I have.
Of course making a formula to just prove the Examine Window shows us little....when I started the formula I wanted to include real, actual firing speeds but the changes to Specials make it more difficult. The actual in-game experience also doesn't seem to see improvements in speeds over +125 but your formula does....which is why I was using ESTIMATED points to provide the proper curve based on the info from the Devs. Maybe we have a hard-cap or some other limit that the Examine Window doesn't show.....but clearly the Examine Window gives us only a portion of the Speed Formula.
I would still like to hear what this does for usyou have just developed a formula to display a number already given to us by the Devs.....if you plan on expanding it to include the whole picture, then GREAT JOB I am happy to see an accurate formula but there is much work to be done to get to Actual Firing Speed and thus Actual Damage.
TAfirehawk wrote:
Doc Buff for Speed doesn't change anything on the Character Sheet or Examine Window of the weapon.....but it speeds you up is the report so no way to track it....
And the difference in Examine Windows numbers up there is probably from being over the +25 SEA cap. The Character Sheet just like the Skill Sheet pre-CU shows more than +25 SEA gains when in reality it has no effect.
Anyway, as the data shows, both formula are wrong
*Bangs hishead against the keyboard and realized he just spent valuable time of his life discussing with a *edit**
You can't use data like this...ever figured the reason why the numbers are off seems to be because YOU ARE OVER THE+25 CAP!!! For gods sake...
Like MAGIC when you used the speed stim the speed decreased, not by adding additional SEAs beyond the +25 cap. I can't fathom why. Can you?
Now, that 120 number. That is with +30 in SEAs right? So the actual number is 115...