The efficiency of scopes/lasers

[Vagrant]

Thanks Recoda for actualy proving what i have been saying for 3years

 

[e-lite]

Skippie, it hasn't been proved yet: only Recoda himself has been able to see any difference. The two other testers since then have only checked maxed weapons and seen no obvious difference.

Anyone wondering about this should download Recoda's little program and start collecting data themselves. It will take a massive amount of very carefully collected data to "prove" this to any degree of certainty, much less work out what exactly the attachments do. For example, even if the full mod % is applied to the HA (like what you describe Skippie), but there is no modification at 10/10, at what point does one fade into the other?

BTW that program is useful for keeping count if you're testing evade, too.

Naa, I backup recoda 100% on this. I don't have any fancy number to show for it anymore since it was several years ago since I tested it. But back then we came to the same conclusion. I am convinced that Recoda's data is correct. Besides, Recoda is a wiz with statistics and data

 

[Doer]

Besides, Recoda is a wiz with statistics and data

These tests were done with 1000 shoots in total. I think I'm going to try to do at least 10,000 shoots later on to get more accurate results.

No amount of statistical wizardry can make a proof of this out of 1000 shots. I did several runs of 1000 shots each and saw an experimentaly relevant variation even without changing the parameters. Sorry. I'm not saying it is wrong, just that it hasn't been proven by any stretch of the imagination.

If you have data of your own please share it here for us to evaluate.

 

[Gingko]

Influence of mob??

Hi,

impressive work, I will spread some +rep around.

One thing bugs me though -- I have the strong feeling (and I may test that later using the program) that the mob itself does also affect my hitrate. I have noticed several times that hunting mobs that are near my limit (i.e. that I can take down alone barely) comes with a crappy hitrate. E.g. I hit Corn Mature almost every time, on a Corn Stalker, I can easily miss three shots in a row with the same weapon (as they are bigger, I should miss lest often, but hey, that would be realistic ). Meaning: Each mob may have it's own Evade/Dodge skill. If this turns out to be true, the research must be limited to 1 mob, 1 maturity .

Anyone have proof against his (making things a bit easier)? Otherwise, I shall test this.

 

[Doer]

Hi,

impressive work, I will spread some +rep around.

One thing bugs me though -- I have the strong feeling (and I may test that later using the program) that the mob itself does also affect my hitrate. I have noticed several times that hunting mobs that are near my limit (i.e. that I can take down alone barely) comes with a crappy hitrate. E.g. I hit Corn Mature almost every time, on a Corn Stalker, I can easily miss three shots in a row with the same weapon (as they are bigger, I should miss lest often, but hey, that would be realistic ). Meaning: Each mob may have it's own Evade/Dodge skill. If this turns out to be true, the research must be limited to 1 mob, 1 maturity .

Anyone have proof against his (making things a bit easier)? Otherwise, I shall test this.

No proof on this one way or another but i have noticed that if i frantically hit the fire button i miss more. Meeting up with a mob at the edge of your abilities is not conducive to calm and careful aim-taking, so your observation doesn't surprise me too much. Seriously, though, check first to see if it isn't how fast you hit the button. I haven't actually measured but it seems that if i pause for just a fraction of a second after reload before shooting, i miss less. (Yeah it doesn't make a lot of sense from a programming perspective.) Maybe it is simply a matter of not aiming as carefully on my part...

 

[Vedder]

So, to recap the most popular conclusions from this thread:

Full conditioned lasers/scopes, reduces your misses and changes your skill gains, which leads us to believe that they actually increase your effective HA.

Low conditioned lasers/scopes, increases your misses.



Now... Based on the above, wouldn't the obvious assumption be that low conditioned lasers/scopes lowers your effective HA?

And if so, is it then worth to use low conditioned lasers/scopes if your on 6.2 HA, to get skills like you did at 5.8? (Just example numbers, ofc)

Hmmm... Interesting.
I agree with Demoniac that it's about time, MA visualized the effects of the amps.
Who knows... Maybe damage amps also affetcs HA? It's another popular belief that damage amps affects you skill gains too...
Hmm... Now my head hurts again

 

[Konve]

Now... Based on the above, wouldn't the obvious assumption be that low conditioned lasers/scopes lowers your effective HA?

And if so, is it then worth to use low conditioned lasers/scopes if your on 6.2 HA, to get skills like you did at 5.8? (Just example numbers, ofc)

Hmmm... Interesting.

I didn't consider low TT attachments, but it would fit:

Edit2: I wonder if there will be laser and scopes with Negative figgures in the future. Say you have really high skills, your have a hidden HA figgure of 15. But the gun is maxed at 10 so it doesn't show more then that. But still, from hunting you can tell that you gain almost no skills at all, because they are already so high - all mobs are below your level.

So if you attach some -50% to your weapon, and you now have a hidden figgure of 10, and your shown HA is still 10. So your stats were not modified, however you now gain skills a lot faster because the mobs are at your level (again).

I agree with Demoniac that it's about time, MA visualized the effects of the amps.
Who knows... Maybe damage amps also affetcs HA? It's another popular belief that damage amps affects you skill gains too...
Hmm... Now my head hurts again

Well a higher dmg weapon would require you to be at a higher level in order to utilize it properly (be at your level), as well as the skillgains from diffrent mobs should be different because of the level of your weapon.

 

[Witte]

I have added scopes and lasers to the weaponchart at pw-wiki. It simply sums up the skillmod of all attachments you select, and increases your HA with that percentage (untill it reaches 10). The decay of the attachments is added to the total decay so also calculated in the "eco" figure you get. You will notice that attachments are not beneficial in all situations. Basicly, the more damage the gun does, the more it will benefit (logically).

see http://www.pe-wiki.info/Chart.aspx?chart=Weapon
To make it work you need to enable the personalized stats.

PS, for allot of attachments the decay isnt filled in yet. So if you have any decay data you can fill it in here: http://www.pe-wiki.info/Chart.aspx?chart=Attachment

 

[Doer]

Nice work on the wiki Witte. I'm glad to see there's a bold disclaimer that attachments aren't fully understood yet. I notice that it doesn't remove non-ranged weapons from the list when sights and scopes are added, is that intended behavior? Also, it appears from testing here that there is no affect over 10 HA. Is that currently considered for SIB weapons in the wiki stats?

Edit: actually, there is very little change in any weapon eco on wiki when adding attachments. What formula did you use?

 

[Konve]

Very cool thing Witte. But might I ask, does it calculate your eco on the gun using both average damage and missed shots accoring to your HA?

Soon there will only be critical hits left to deal with.

(And of course the whole loot pool thing. )

 

[Witte]

Nice work on the wiki Witte. I'm glad to see there's a bold disclaimer that attachments aren't fully understood yet. I notice that it doesn't remove non-ranged weapons from the list when sights and scopes are added, is that intended behavior? Also, it appears from testing here that there is no affect over 10 HA. Is that currently considered for SIB weapons in the wiki stats?

Edit: actually, there is very little change in any weapon eco on wiki when adding attachments. What formula did you use?

I took a look at the code again, and discovered some mistakes. I fixed it and it should work correct now.

Very cool thing Witte. But might I ask, does it calculate your eco on the gun using both average damage and missed shots accoring to your HA?

Soon there will only be critical hits left to deal with.

(And of course the whole loot pool thing. )

For info on how its calculated you can click on help behind the checkbox and then see the discussion tab.

 

[Gingko]

Did a small test with Breer M3a + A102 (no other attachments) on Merp Prowler/Stalker to start with. I have this gun maxed out, my "Hit" Profession is at 25. I have been very careful on this (only stopped mob, shot with patience, ...). Did it the old fashioned paper-way as I needed to record mob maturity also.

Results:
1128 shots
108 misses (9.57%; 95% CI: 7.92% ... 11.44%)
22 critical hits (1.95%; 95% CI: 1.23% ... 2.94%)

The problem I see overall is the pretty wide 95% Confidence Interval (stretching across 3.5%), therefore many more shots than just 1000 will be needed to really see the effects we're after. If we want to narrow the corridor (95% CI) to 1%, we'd need approx. 10000 shots...

 

[Rival]

I invested heavily in the old-style lasers and sights 6 months ago because they decayed at such a lower rate than the newer ones for the same increases. Perhaps when MA finally lets the cat out of the bag I will be a gazillionaire

 

[Doer]

I did a small test with an unmaxed weapon on ambus today and updated my earlier post with the actual data. Lots of uncertainty, but for what it's worth here is a chart with the summary so far of all my tests.

I'm still seeing a lot of variation during a test, and several misses in a row or on one mob. I was much less careful in this test than the previous: i tagged the ambu without counting the shot(s) to discount lagged mobs, and once it was headed toward me i started counting.

I tried changing the conventionally accepted hit % range (80-92) to 80-91 because i always got around 9% misses with the maxed weapon, but the unmaxed values fit much closer if i kept the old range.

This uses a range from 81 to 91 hit%. While the higher effective HAs come closer, the lowest is farther away so i'm not sure this is the right way to do it, either. I don't consider my data accurate or confident enough to justify either of them

Results seem favorable for attachments raising the effective hit ability on unmaxed weapons, but i'm not fully convinced. Call it 50:50. I don't see any way to include the tt of the attachments (Recodo's initial post suggested that the condition of the attachment plays a role) and get sensible results though. I'd guess the attachments contribute full mod % until broken completely. Having not broken an attachment i don't know if you can even put it on the weapon when broken, anyway.

It seems pretty clear they don't do anything significant on maxed SIB weapons. I guess my earlier observation was a coincidence after all. Eerie.

 

[Jimmy B]

Nice work Doer...

I think I'll try not to think too hard about all those PED I've wasted in the last 12 months from using lasers with SIB guns!

 

[Tamn]

Wow, can't believe I didn't see this post earlier, awsome data!!!!!
great work, wow, this is great stuff

 

[Witte]

I tried changing the conventionally accepted hit % range (80-92) to 80-91 but the unmaxed values fit much closer if i kept the old range.

What about 81 to 91? What is your HA in this test?

 

[Doer]

What about 81 to 91? What is your HA in this test?

Nearer to the majority of the measured results, but farther from the lowest one. I don't consider my results accurate enough to justify one or the other.


base HA 4.65

This would be much easier to conclude if we had hit % at 0 HA.

 

[Konve]

I did a small test with an unmaxed weapon on ambus today...

But how small, is small?

 

[Doer]

But how small, is small?

I did a small test with an unmaxed weapon on ambus today and updated my earlier post with the actual data.

About the same size as Recoda's original tests, with no repetitions to determine the error or confidence level. For what it's worth the hit % fluctuated about +-0.6 during the test after the first 100 shots or so.

Using the counter utility is a big enough departure from my usual hunting controls setup that it makes me feel claustrophobic and generally not want to repeat the tests.

 

[Gingko]

Statistical significance....

Folks,

I have been thinking about the data posted in all the posts and while there is some good data, it is still not enough to draw any conclusion... why? Read on.

There is a nice thing called a statistical test, used to determine if two sets of data are significantly different from another. Significantly means a level of confidence of 95%. In other words: if the test tells you the data sets are different, you can be 95% sure they are. If the test tells you nothing, you really know not much (you cannot really say they are the same).

If you plug the figures we generally deal with in this thread into these tests, you have a certain number of shots and misses in each data set. Assuming the number of shots is the same for each data set, we are looking for a difference in the misses.

So how large does the difference in the misses need to be to be 95% sure that there is one? It depends, mainly on the number of shots fired, but also on the general miss rate. Therefore, I assumed 8% miss rate in the following calculations. To make things more complicated, there are more than one test giving different results. I use the one that assumes the data to be normal distributed, which should be an OK assumption for the data we're dealing with. This test is a bit more "friendly" than the other (Chi²-test), so the results are a bit "optimistic".

Case 1: 1000 shots fired

Assume data set 1 has 1000 shots, 80 misses (8%).
Data set 2 would be significantly different at 106 or more misses (10.6%).
Means: The difference in the miss rate must be at least 2.6% absolute or 32.5% relative.

Tells me: the results in the first post of this thread "look good", but all the differences are not significant, ergo nothing has been proven to date.

Case 2: 3000 shots fired

Set 1: 3000 shots, 240 misses (8%)
Set 2: needs 283 or more misses (9.43%) to be significantly different
Difference needed: 1.43% abs or 17.9% relative

Case 3: 10000 shots fired

Set 1: 10000 shots, 800 misses (8%)
Set 2: needs 877 or more misses (8.77%) to be significantly different
Difference needed: 0.77% abs or 9.6% relative

Here it starts to become interesting...

Conclusion: The only data set we seem to have enough data on is the one with maxed weapons -- here we could potentially throw all data in one pot.
Non-maxed weapons with and without attachments need much more data to be able to really see the difference -- and to calculate the effect on eco without big assumptions.

I do think this thread lists a very important finding, but we all need to collect more data

 

[Konve]

I have been thinking about the data posted in all the posts and while there is some good data, it is still not enough to draw any conclusion...

There is a nice thing called a statistical test, used to determine if two sets of data are significantly different from another. Significantly means a level of confidence of 95%.

As I posted earlier in this thread; I did four runs at 10 HA:

1079 shots fired; 89 misses (8,2%)
1212 shots fired; 106 misses (8,7)
1171 shots fired; 117 misses (9,9%)
1098 shots fired; 98 misses (8,9%)


None of these were significantly different (form the lowest at 8,2%), right?

Or do they have to be exactly the same amount of shots before you're convinced?

 

[Doer]

As I posted earlier in this thread; I did four runs at 10 HA:

1079 shots fired; 89 misses (8,2%)
1212 shots fired; 106 misses (8,7)
1171 shots fired; 117 misses (9,9%)
1098 shots fired; 98 misses (8,9%)


None of these were significantly different (form the lowest at 8,2%), right?

Or do they have to be exactly the same amount of shots before you're convinced?

His point is that for the measurements (eg mine without SIB and Recoda's) to show a significant difference between attachments and no attachments, there must be much more data for each of those cases. Or, as he put it,

Conclusion: The only data set we seem to have enough data on is the one with maxed weapons -- here we could potentially throw all data in one pot.
Non-maxed weapons with and without attachments need much more data to be able to really see the difference -- and to calculate the effect on eco without big assumptions.

It has nothing to do with convincing him of something.

However, i did say i was about 50% convinced that attachments improve the HA on an unmaxed weapon. When we have enough data to have statistical confidence that there is a significant difference reflecting that, i will be fully convinced.

It's a mathematical (statistical) propriety he's talking about and has nothing to do with opinion.

 

[Grounder]

Hello,

I have read the thread and must admit I had a little trouble following it.

Am I correct in stating that the testing to date has shown that there is a decrease in hit rate (and therefore efficiency) when using low TT sights and scopes - but that some think the evidence is not yet conclusive due to the number of potential variants?

Like most of you - I'm just looking for a clear answer.

This sort of thing is why I would really like MA to publish a user guide with stat's and the way things work.

Grounder

 

[Doer]

What has been "proven" so far:
-the average miss rate for a maxed weapon is around 9%

What is mostly proven so far:
-skill modification attachments (sights and scopes) do not improve your hit rate on maxed SIB weapons

What is possibly proven so far:
-skill modification attachments (sights and scopes) improve your hit rate on unmaxed weapons

What is seen to conflict even with the data posted in this thread:
-low tt attachments decrease your hit rate on unmaxed weapons

 

[falkao]

I did an investigation some time ago and didn't find any statistically difference in using or not using scopes/lasers. It seems that lag has more influence on miss rate then anything else.

https://www.planetcalypsoforum.com/forums/showthread.php?t=45360

I was for example able to get the following results:

Unamped Opalo:
Shots: 538
Misses: 45
Miss Rate 8.4%

Fully Amped Opalo with Lasers and scopes
Shots: 533
Misses: 68
Miss Rate 12.8%

The difference between 8.4% and 12.8% is statistically significant p = 0.019. So I did more misses with an amped weapon and I guess it was lag.

 

[Konve]

What is mostly proven so far:
-skill modification attachments (sights and scopes) do not improve your hit rate on maxed SIB weapons

Actually - any and all test so far using an already maxed out weapon (only SIB so far) with scope and lasers has had a worse performance then those without.

 

[Doer]

Actually - any and all test so far using an already maxed out weapon (only SIB so far) with scope and lasers has had a worse performance then those without.

Not significantly worse, though. In my tests it was well within the margin of error. However, due to the extra decay i do feel pretty safe drawing this conclusion even if the attachments do give a small but basically impossible to measure advantage:

Any improvement (if it exists) in the hit rate of maxed weapons due to scopes and sights is more than overcome by the extra decay of those attachments, meaning that putting modification attachments on a maxed SIB weapon is a bad idea when it comes to economy.

 

[Konve]

Any improvement (if it exists) in the hit rate of maxed weapons due to scopes and sights is more than overcome by the extra decay of those attachments, meaning that putting modification attachments on a maxed SIB weapon is a bad idea when it comes to economy.

Unless it will alter your level (for better or worse {by using low TT attachments}) and alater your skillgains to the better - which has not been proved.

 

[Doer]

Unless it will alter your level (for better or worse {by using low TT attachments}) and alater your skillgains to the better - which has not been proved.

Sure but that has nothing directly to do with economy, which was the qualifier in my statement. I won't be putting attachments on an m2a on the slim chance it will get me "some" more skills than without, considering the difference in eco (2.87 vs. 2.83 for some the lowest decay attachments). If the attachments make your effective level higher it will if anything most probably reduce your skill rate, and we still have precious little evidence to suggest that low tt attachments do anything significantly different than full tt attachments.

Recoda, where are you? Any more tests on your end?