Whats your SD (PRS/ELR)

Willie16wild

WillieGunner
Belligerents
Mar 3, 2017
242
17
22
Punxsutawney, PA
I have a simple question.

What standard deviation (SD) are you getting?

I shoot PRS style and I was wondering what your SD is.

I have been averaging a SD of 5. Thats only with 5 rounds per series. Thats out of 100 rounds. I also let my gun cool a little every 10 shoots. My average velocity is 2810 fps.

I am running a 6.5 creedmoor, MPA chassis, stiller action, brux barrel, and silencer co. omega surppressor.
 

sleeplz

Private
Belligerents
May 12, 2017
791
407
69
Sd of 5 and es of 18 with 6mm gt. I shot alabama precision with it and when I held my part and stabilized my reticle, I poked out to 1100 in their long distance stage. They allowed 2 spotter rounds.
 

Ledzep

Chancellor
Belligerents
Jun 9, 2009
2,570
1,705
219
NE
This topic came up in another post in the range reports section. The moral of the story is that if you're using any sort of analysis on a sample size of less than 20-25, you're lying to yourself.

This is .223 data with 5 different charge weights every .2 grains from 23.5-24.3 of Varget. Sorry for not labeling well; it was originally only for personal use. On the left side is SD in fps, on the bottom is the number of samples. It represents a running SD as the number of shots/samples increases over 30-35 rounds.
Std dev 23.5-24.3gr Varget 75gr bthp T2.PNG

If you look at the data from 5 shots it's a complete lie. If you look at the data at 10 shots it's a lie. From 20-30 shots you're kinda getting an idea, and IMHO I should have taken this to 50 shots because I think it would tell me that changing charge weight really doesn't change SD appreciably at all.

Most notably, follow the dark blue and orange lines as sample size increases and think about how your perception of a load would be at 5, 10, 15, 20, etc.. shot sample sizes...

Now with that said, most of these ended up hovering around 20fps. Another rifle (6 creed) I shot a 30 shot sample and it was at 12.5fps SD after 30.

From what I have seen so far I would say a TRUE SD of between 10-15fps is excellent. Any 5-10 shot sampling is a waste of your time, other than potentially grabbing a quick average velocity to plug into your calculator. However, even average velocity bounces around and settles out after 20 samples, too.
 

phlegethon

Gunny Sergeant
Hessian
Online Training Access
Belligerents
Minuteman
Nov 4, 2018
874
666
99
This topic came up in another post in the range reports section. The moral of the story is that if you're using any sort of analysis on a sample size of less than 20-25, you're lying to yourself.

This is .223 data with 5 different charge weights every .2 grains from 23.5-24.3 of Varget. Sorry for not labeling well; it was originally only for personal use. On the left side is SD in fps, on the bottom is the number of samples. It represents a running SD as the number of shots/samples increases over 30-35 rounds.
View attachment 7258249

If you look at the data from 5 shots it's a complete lie. If you look at the data at 10 shots it's a lie. From 20-30 shots you're kinda getting an idea, and IMHO I should have taken this to 50 shots because I think it would tell me that changing charge weight really doesn't change SD appreciably at all.

Most notably, follow the dark blue and orange lines as sample size increases and think about how your perception of a load would be at 5, 10, 15, 20, etc.. shot sample sizes...

Now with that said, most of these ended up hovering around 20fps. Another rifle (6 creed) I shot a 30 shot sample and it was at 12.5fps SD after 30.

From what I have seen so far I would say a TRUE SD of between 10-15fps is excellent. Any 5-10 shot sampling is a waste of your time, other than potentially grabbing a quick average velocity to plug into your calculator. However, even average velocity bounces around and settles out after 20 samples, too.
I couldn't agree more. This is absolutely correct. An SD of five shots is like measuring a two shot group.
 
  • Like
Reactions: _Raining

Willie16wild

WillieGunner
Belligerents
Mar 3, 2017
242
17
22
Punxsutawney, PA
This topic came up in another post in the range reports section. The moral of the story is that if you're using any sort of analysis on a sample size of less than 20-25, you're lying to yourself.

This is .223 data with 5 different charge weights every .2 grains from 23.5-24.3 of Varget. Sorry for not labeling well; it was originally only for personal use. On the left side is SD in fps, on the bottom is the number of samples. It represents a running SD as the number of shots/samples increases over 30-35 rounds.
View attachment 7258249

If you look at the data from 5 shots it's a complete lie. If you look at the data at 10 shots it's a lie. From 20-30 shots you're kinda getting an idea, and IMHO I should have taken this to 50 shots because I think it would tell me that changing charge weight really doesn't change SD appreciably at all.

Most notably, follow the dark blue and orange lines as sample size increases and think about how your perception of a load would be at 5, 10, 15, 20, etc.. shot sample sizes...

Now with that said, most of these ended up hovering around 20fps. Another rifle (6 creed) I shot a 30 shot sample and it was at 12.5fps SD after 30.

From what I have seen so far I would say a TRUE SD of between 10-15fps is excellent. Any 5-10 shot sampling is a waste of your time, other than potentially grabbing a quick average velocity to plug into your calculator. However, even average velocity bounces around and settles out after 20 samples, too.



This is actually new information to me. Really good information too!

Can I ask where you came across this knowledge of information or were you just that smart?

I am not trying to be mean about how I said that. I am being serious.
 

Ledzep

Chancellor
Belligerents
Jun 9, 2009
2,570
1,705
219
NE
Probability and Statistics class in college and curiosity. The results above are from my .223 and a LabRadar.

Standard deviation only means something if the collection of your samples resembles a Normal distribution (bell curve). A LOT of things people sample end up looking like a normal distribution, but it's hard to get there with a sample of size 5.

ETA: This covers the basics pretty well.
 
Last edited:

Willie16wild

WillieGunner
Belligerents
Mar 3, 2017
242
17
22
Punxsutawney, PA
Thank you very much for that.

Now how do you go about cooling of the barrel? Cause I know after 10 or 15 shots with my 6.5 the barrel will be pretty darn warm and my surppressor will be very hot.
 

kthomas

Sergeant
Belligerents
Jun 17, 2009
2,775
3,007
219
32
Tucson, AZ
This topic came up in another post in the range reports section. The moral of the story is that if you're using any sort of analysis on a sample size of less than 20-25, you're lying to yourself.

This is .223 data with 5 different charge weights every .2 grains from 23.5-24.3 of Varget. Sorry for not labeling well; it was originally only for personal use. On the left side is SD in fps, on the bottom is the number of samples. It represents a running SD as the number of shots/samples increases over 30-35 rounds.
View attachment 7258249

If you look at the data from 5 shots it's a complete lie. If you look at the data at 10 shots it's a lie. From 20-30 shots you're kinda getting an idea, and IMHO I should have taken this to 50 shots because I think it would tell me that changing charge weight really doesn't change SD appreciably at all.

Most notably, follow the dark blue and orange lines as sample size increases and think about how your perception of a load would be at 5, 10, 15, 20, etc.. shot sample sizes...

Now with that said, most of these ended up hovering around 20fps. Another rifle (6 creed) I shot a 30 shot sample and it was at 12.5fps SD after 30.

From what I have seen so far I would say a TRUE SD of between 10-15fps is excellent. Any 5-10 shot sampling is a waste of your time, other than potentially grabbing a quick average velocity to plug into your calculator. However, even average velocity bounces around and settles out after 20 samples, too.
So true.

I hate the vanity posts on Facebook where people post a picture of their chronograph after 3-5 shots showing a really low SD. Not truly indicative of what their real SD/ES is, but it looks good in front of your fake internet friends!

All that does is make new people getting into the sport chase unrealistic goals.

I find that SD/ES numbers can vary with the exact same load/ammo can vary by small amounts day to day. A load that may shoot an SD of ~5 for 25+ shots one day can have an SD of ~8 the next day, even with really temperature stable ammo.

People put too much stock in statistically insignificant numbers. By that, I mean that generally very few people use the numbers from a statistically relevant sample size, but I believe a lot of that has to do with vanity because the numbers don't look as good, plus a lack of misunderstanding of how statistics works. It's also why the "Satterlee method" for load development doesn't really work.

Anyways, for all my loads, I aim for an ES of 20 or less for 25+ shots. Depending on the day, my loads usually hover between ~15-20 ES.
 

Winny94

Major Hide Member
Belligerents
Nov 19, 2013
1,515
548
219
I tell people your ES should be 1% of your velocity (Ex. 3000 fps ES 30 "+/- 15 fps"). You can burn a barrel up chasing single digit ES numbers.
For sure. ES is statistically irrelevant (within reason obviously). People waste a whole helluva lotta money, time, and components chasing low ES. By their very definition they are outliers - I saw someone on here once what an SD of 10 looks like on the bell curve then extrapolated that into actual drop/deflection for an average short action round at various ranges. Very enlightening seeing it displayed that way. I have loads that shoot an SD of 3-5, but anymore, unless I get into ELR or BR, theres not a lot of reason to keep chasing anything less than a 10 SD.
 

Steel head

Feral kitten
Belligerents
Aug 3, 2014
7,457
10,191
219
Washington
For sure. ES is statistically irrelevant (within reason obviously). People waste a whole helluva lotta money, time, and components chasing low ES. By their very definition they are outliers - I saw someone on here once what an SD of 10 looks like on the bell curve then extrapolated that into actual drop/deflection for an average short action round at various ranges. Very enlightening seeing it displayed that way. I have loads that shoot an SD of 3-5, but anymore, unless I get into ELR or BR, theres not a lot of reason to keep chasing anything less than a 10 SD.
Lol
Reminds me of the guys chasing perfect groups at 100 or making short range AR fodder making a fuss about SD/ES.

My favorite AR load has an abysmal ES.
 

kthomas

Sergeant
Belligerents
Jun 17, 2009
2,775
3,007
219
32
Tucson, AZ
Lol
Reminds me of the guys chasing perfect groups at 100 or making short range AR fodder making a fuss about SD/ES.

My favorite AR load has an abysmal ES.
People like to "chase ghosts" so to speak.

I'm guilty of it too to some degree, I want my ammo to perform as best as feasibly possible, and ES is one measure of that. But at some point you need to draw a practical limit and be happy if you get to a certain point.

Interestingly, the guys chasing world records (benchrest shooters) don't even know what their SD/ES is for their loads. Most couldn't even tell you what their velocity is.
 

LawnMM

Harbinger of Sarcasm
Belligerents
Jul 5, 2009
3,365
2,953
219
Colorado
Probability and Statistics class in college and curiosity. The results above are from my .223 and a LabRadar.

Standard deviation only means something if the collection of your samples resembles a Normal distribution (bell curve). A LOT of things people sample end up looking like a normal distribution, but it's hard to get there with a sample of size 5.

ETA: This covers the basics pretty well.
Been a while since my statistics class in college...

But! Given that our stages typically run 10 rounds wouldn't that be the more statistically relevant distribution?

You guys talking about measuring SD over 50rds, when was the last stage you shot 50 rounds on?

Maybe we should shoot a 250rd sample so it's more statistically relevant to an entire match. Or 2000rds so it more closely resembles the life of the barrel entirely.

Or...we could spend more time on the range and less printing charts and spreadsheets and arguing about statistics on the internet.

 

phlegethon

Gunny Sergeant
Hessian
Online Training Access
Belligerents
Minuteman
Nov 4, 2018
874
666
99
Been a while since my statistics class in college...

But! Given that our stages typically run 10 rounds wouldn't that be the more statistically relevant distribution?
The issue is that you are trying to make a prediction about the whole population of cartridges that you are loading. You need a certain sample size to be able to do that. This would be true whether you are planning to shoot a ten round stage, a one round stage, or a thousand rounds in a row.
 

phlegethon

Gunny Sergeant
Hessian
Online Training Access
Belligerents
Minuteman
Nov 4, 2018
874
666
99
Lol, why stop there might as well permanently attach a magnetospeed to the gun so you can know what every shot is.
There are ELR shooters who do that, I think, so they can see if the shot hits low it was because it was slower or if they really need to adjust the point of aim.
 

kthomas

Sergeant
Belligerents
Jun 17, 2009
2,775
3,007
219
32
Tucson, AZ
Lol, why stop there might as well permanently attach a magnetospeed to the gun so you can know what every shot is.

I personally dont care that much. I dont chase single digits.
That's kind of my point - people who chase single digit ES or think they are already there, are more or less chasing ghosts.

Ultimately what matters is what the bullet is doing down range. If what you are doing is working, then awesome.

I don't shoot with my magnetospeed attached all the time, nor do I always shoot 25+ round strings with the magnetospeed. That's not my point.

My point is that most people don't really understand what their true SD/ES is, and what the statistics behind it is. But people love to throw out a super low SD number from a 3 shot string they shot once, thinking they are something special.
 
  • Like
Reactions: eldritch

Sig Marine

Sergeant - USMC 1968-1970
Belligerents
Dec 29, 2013
546
298
69
So Cal
The
There are ELR shooters who do that, I think, so they can see if the shot hits low it was because it was slower or if they really need to adjust the point of aim.
This exactly why it’s done; making corrections based solely on POI is futile and without merit at longer distances where ES is your enemy. I am not anal about my reloading yet, it may come as longer distances avail themselves, but this is my last trip out shooting to 2300 yards with the MagnetoSpeed attached.

IMG_0116.jpg
 

_Raining

Private
Online Training Access
Belligerents
Feb 14, 2017
404
178
49
This topic came up in another post in the range reports section. The moral of the story is that if you're using any sort of analysis on a sample size of less than 20-25, you're lying to yourself.

This is .223 data with 5 different charge weights every .2 grains from 23.5-24.3 of Varget. Sorry for not labeling well; it was originally only for personal use. On the left side is SD in fps, on the bottom is the number of samples. It represents a running SD as the number of shots/samples increases over 30-35 rounds.
Std dev 23.5-24.3gr Varget 75gr bthp T2.PNG


If you look at the data from 5 shots it's a complete lie. If you look at the data at 10 shots it's a lie. From 20-30 shots you're kinda getting an idea, and IMHO I should have taken this to 50 shots because I think it would tell me that changing charge weight really doesn't change SD appreciably at all.

Most notably, follow the dark blue and orange lines as sample size increases and think about how your perception of a load would be at 5, 10, 15, 20, etc.. shot sample sizes...

Now with that said, most of these ended up hovering around 20fps. Another rifle (6 creed) I shot a 30 shot sample and it was at 12.5fps SD after 30.

From what I have seen so far I would say a TRUE SD of between 10-15fps is excellent. Any 5-10 shot sampling is a waste of your time, other than potentially grabbing a quick average velocity to plug into your calculator. However, even average velocity bounces around and settles out after 20 samples, too.
Does this make you horny baby?

(Not that it rly matters but this was 40.4 grains of RL16 in 6.5 Creedmoor Lapua SRP Brass , CCI400 Small Rifle Primers, Berger 140 gr Hybrids)

Bunch of 10 shot groups:

no reading, 2814, 2815, 2813, 2805, 2819, 2815, 2813, 2815, 2806
avg: 2813
SD: 4.5
ES: 14
Temp: 74.5

2818, 2808, 2815, 2819, 2817, 2810, 2813, 2810, 2811, 2816
avg: 2814
SD: 3.8
ES: 11
Temp: 75.5

2810, 2816, 2819, 2813, 2807, 2807, 2812, 2814, 2815, 2812
avg: 2813
SD: 3.8
ES: 12
Temp: 76.4

2814, 2814, 2816, 2805, 2816, 2815, 2813, 2814, 2809, 2808
avg: 2812
SD: 3.8
ES: 12
Temp: 76.9

2818, 2815, 2811, 2813, 2822, 2815, 2811, 2814, 2809, 2806
avg: 2813
SD: 4.4
ES: 15
Temp: 77.9

2818, 2808, 2815, 2824, 2809, 2813, 2807, 2805, 2814, 2813
avg: 2813
SD: 5.6
ES: 18
Temp: 77.9

2813, 2812, 2812, 2816, 2813, 2809, 2816, 2817, 2812, 2807
avg: 2813
SD: 3.3
ES: 11
Temp: 77.8

2814, 2812, 2815, 2805, 2814, 2808, 2815, 2816, 2822, 2816
avg: 2814
SD: 4.7
ES: 17
Temp: 78.3

(looks like I was lazy and didn't record the individual values here :/)
avg: 2815
SD: 2.4
ES: 8
Temp: 79.2

Total Sample (79 count)
Avg: 2812.9747
Avg Confidence Level of 95% = 2812.9747 ± 0.911
SD: 4.12
ES: 19
 

b6graham

MMPRL
Belligerents
Jul 29, 2014
4,025
3,216
219
Does this make you horny baby?

(Not that it rly matters but this was 40.4 grains of RL16 in 6.5 Creedmoor Lapua SRP Brass , CCI400 Small Rifle Primers, Berger 140 gr Hybrids)

Bunch of 10 shot groups:

no reading, 2814, 2815, 2813, 2805, 2819, 2815, 2813, 2815, 2806
avg: 2813
SD: 4.5
ES: 14
Temp: 74.5

2818, 2808, 2815, 2819, 2817, 2810, 2813, 2810, 2811, 2816
avg: 2814
SD: 3.8
ES: 11
Temp: 75.5

2810, 2816, 2819, 2813, 2807, 2807, 2812, 2814, 2815, 2812
avg: 2813
SD: 3.8
ES: 12
Temp: 76.4

2814, 2814, 2816, 2805, 2816, 2815, 2813, 2814, 2809, 2808
avg: 2812
SD: 3.8
ES: 12
Temp: 76.9

2818, 2815, 2811, 2813, 2822, 2815, 2811, 2814, 2809, 2806
avg: 2813
SD: 4.4
ES: 15
Temp: 77.9

2818, 2808, 2815, 2824, 2809, 2813, 2807, 2805, 2814, 2813
avg: 2813
SD: 5.6
ES: 18
Temp: 77.9

2813, 2812, 2812, 2816, 2813, 2809, 2816, 2817, 2812, 2807
avg: 2813
SD: 3.3
ES: 11
Temp: 77.8

2814, 2812, 2815, 2805, 2814, 2808, 2815, 2816, 2822, 2816
avg: 2814
SD: 4.7
ES: 17
Temp: 78.3

(looks like I was lazy and didn't record the individual values here :/)
avg: 2815
SD: 2.4
ES: 8
Temp: 79.2

Total Sample (79 count)
Avg: 2812.9747
Avg Confidence Level of 95% = 2812.9747 ± 0.911
SD: 4.12
ES: 19
i'd be curious how this changes at 45 and 95 degrees
 

_Raining

Private
Online Training Access
Belligerents
Feb 14, 2017
404
178
49
i'd be curious how this changes at 45 and 95 degrees
I don't think I have data for colder/hotter temps with this rifle/load and the barrel is now a tomato stake. Sadly my wrists are just getting worse and worse so I don't do any reloading and very little shooting. It is what it is.
 

Ledzep

Chancellor
Belligerents
Jun 9, 2009
2,570
1,705
219
NE
I don't think I have data for colder/hotter temps with this rifle/load and the barrel is now a tomato stake. Sadly my wrists are just getting worse and worse so I don't do any reloading and very little shooting. It is what it is.
Just curious what the details of the rifle were. Barrel make, profile and chamber mostly-- and load prep. That's the best data set for ES/SD that I've seen for a large sample size. Most of my "good" rifles that I've got to testing so far are in the 10-13fps range for SD. I'm not doing any prep, either, though. Just load and shoot.
 

2aBaCa

Major Hide Member
Belligerents
Minuteman
Jan 27, 2019
2,487
2,173
119
Behind enemy lines Northern Commiefornia
That's kind of my point - people who chase single digit ES or think they are already there, are more or less chasing ghosts.

Ultimately what matters is what the bullet is doing down range. If what you are doing is working, then awesome.

I don't shoot with my magnetospeed attached all the time, nor do I always shoot 25+ round strings with the magnetospeed. That's not my point.

My point is that most people don't really understand what their true SD/ES is, and what the statistics behind it is. But people love to throw out a super low SD number from a 3 shot string they shot once, thinking they are something special.
If you look ledzeps graph, the difference between the shot 10 and 30 are minimal.

Aside from the looks of it the yellow is the only normal one and the others trending one way or the other leads me to suspect something else is influencing.
 
  • Like
Reactions: Ksracer

_Raining

Private
Online Training Access
Belligerents
Feb 14, 2017
404
178
49
Just curious what the details of the rifle were. Barrel make, profile and chamber mostly-- and load prep. That's the best data set for ES/SD that I've seen for a large sample size. Most of my "good" rifles that I've got to testing so far are in the 10-13fps range for SD. I'm not doing any prep, either, though. Just load and shoot.
It was a Brux in Accurate Ordnance Sendero Contour. I just told them to chamber it for the 140 grain hybrids. My notes from that day say:
Flash Hole Deburred, Primer Pocket Uniformed, Trimmed to 1.910", Deburred, Chamfered, Neck Turned .014/.0145", .290" Neck Bushing, CBTO 2.190"
 
  • Like
Reactions: Ledzep

Sig Marine

Sergeant - USMC 1968-1970
Belligerents
Dec 29, 2013
546
298
69
So Cal
ES and SD are also relate to several factors; brass prep, weighing/sorting of brass and/or bullets and/or primers, neck turning as well as type of scale used when measuring powder. I know a couple of shooters who use (and can afford) a Prometheus and their numbers are always better than those of mine using the tools available to the working man.
 

Winny94

Major Hide Member
Belligerents
Nov 19, 2013
1,515
548
219
ES and SD are also relate to several factors; brass prep, weighing/sorting of brass and/or bullets and/or primers, neck turning as well as type of scale used when measuring powder. I know a couple of shooters who use (and can afford) a Prometheus and their numbers are always better than those of mine using the tools available to the working man.
High end scales can make it more efficient, but you can absolutely get sub 5-10 SDs w/ a chargemaster, beam scale, etc. All about consistency. (efficient cartridge design also helps as well)
 

kthomas

Sergeant
Belligerents
Jun 17, 2009
2,775
3,007
219
32
Tucson, AZ
If you look ledzeps graph, the difference between the shot 10 and 30 are minimal.

Aside from the looks of it the yellow is the only normal one and the others trending one way or the other leads me to suspect something else is influencing.
I don't know where the crossover point is where you get the most statistical relevance per sample size for this type of data.

If you are really curious, you can easily test this yourself with your loads and your chronograph.

I can say with confidence that it's definitely more then the 3-5 shots that a lot of people use to brag up their SD/ES numbers.

I've found a sample size of 10 to be a lot more statistically meaningful then 5 shots, so that's a good start.
 

Sig Marine

Sergeant - USMC 1968-1970
Belligerents
Dec 29, 2013
546
298
69
So Cal
High end scales can make it more efficient, but you can absolutely get sub 5-10 SDs w/ a chargemaster, beam scale, etc. All about consistency. (efficient cartridge design also helps as well)
I agree that what you throw powder with does not guarantee low numbers as I have gotten good numbers from the means you list. I think getting good ES/SDs comes from being consistent with your methods and that you reach a point where you have to decide if the energy spent for minor improvements in your numbers is worth it to you. Guys who shoot paper for score obviously feel it’s worth their time and money but I can’t justify more than I do just shooting steel. Now, this may change as I get closer to my shooting goals this year and the room for error gets smaller.
 
  • Like
Reactions: kthomas

_Raining

Private
Online Training Access
Belligerents
Feb 14, 2017
404
178
49
I don't know where the crossover point is where you get the most statistical relevance per sample size for this type of data.

If you are really curious, you can easily test this yourself with your loads and your chronograph.

I can say with confidence that it's definitely more then the 3-5 shots that a lot of people use to brag up their SD/ES numbers.

I've found a sample size of 10 to be a lot more statistically meaningful then 5 shots, so that's a good start.
There are calculators for confidence intervals, like what I posted earlier:
Avg Confidence Level of 95% = 2812.9747 ± 0.911

This means that you can be 95% confident that my average MV is between 2812.0637 and 2813.8857

The same calculation can be done for SD's. So if you end up with:
41 grains produces an SD of 10 over 30 shots gives us 95% CI of the SD to be 7.96 to 13.44.
41.2 grains produces an SD of 8 over 30 shots gives us 95% CI of the SD to be 6.37 to 10.75.
These values overlap, so you can't be 95% confident that the 41.2 grain charge produces a tighter SD then the 41 grain load.
They still overlap for a shot count of 100. If we bump that # up to 1000 shots then we get 7.66 to 8.37 and 9.58 to 10.46 for measured SD of 8 and 10, so in that case you could be 95% confident that the 41.2 grain load is in fact producing better SD then the 41 grain load, but then your barrel is shot out. So realistically we need to drop that % confidence level down bc you ain't gonna be 95% confident for close values with low sample sizes.

95% confidence (30 shots each):
4 -> 3.19 to 5.38
8 -> 6.37 to 10.75

95% confidence (10 shots each):
4 -> 2.75 to 7.30
8 -> 5.50 to 14.60

80% confidence (10 shots each):
4 -> [3.13 , 5.88]
8 -> [6.26 , 11.76]

40% confidence (10 shots each):
8 -> [7.35 , 9.49]
10 -> [9.19 , 11.86]

30% confidence (10 shots each):
8 -> [7.59 , 9.15]
10 -> [9.48 , 11.44]

Don't get me wrong, 30% is better then nothing but I would rather bump up those sample sizes.

For shits and giggles, here is a sample size of 3 and 5:

80% 3 shots:
4 -> [2.64 , 12.32]
8 -> [5.27 , 24.65]

80% 5 shots:
4 -> [2.87 , 7.76]
8 -> [5.74 , 15.51]

60% 5 shots:
4 -> [3.27 , 6.23]
8 -> [6.54 , 12.46]

65% 5 shots:
4 -> [3.18 , 6.51]
8 -> [6.35 , 13.02]
 

Sig Marine

Sergeant - USMC 1968-1970
Belligerents
Dec 29, 2013
546
298
69
So Cal
I don't know where the crossover point is where you get the most statistical relevance per sample size for this type of data.

If you are really curious, you can easily test this yourself with your loads and your chronograph.

I can say with confidence that it's definitely more then the 3-5 shots that a lot of people use to brag up their SD/ES numbers.

I've found a sample size of 10 to be a lot more statistically meaningful then 5 shots, so that's a good start.
I always wonder about the small or odd number shot groups guys seem to post showing their low numbers; culling for effect I guess.
 
  • Like
Reactions: kthomas

Ksracer

Private
Belligerents
Feb 14, 2017
183
65
34
My GF and I shoot a pair of similar rifles in PRS.
Dasher, Bartlein, Impact, LabRadar, etc.
The only real difference is I throw powder with an Auto Trickler setup and she throws with a chargemaster.
I Chrono them every chance I get, but usually only 5-10rnds each before I pick up and move to a different location and start a new string.
My SD's run about 4 and hers around 6 for an average string Those aren't cherry picked, those are normal. I'm on my 4th or fifth barrel, I can't recall ever seeing either rifle go over 8sd, including the cold bore.
I'm having a really, really hard time believing that a 30 or 50 shot consecutive string would push my SD 5 times higher.
 
  • Like
Reactions: 2aBaCa

2aBaCa

Major Hide Member
Belligerents
Minuteman
Jan 27, 2019
2,487
2,173
119
Behind enemy lines Northern Commiefornia
Haha your true high resolution SD could be 3 or it could be 5 you'll never know unless you shoot 50rd string.

My GF and I shoot a pair of similar rifles in PRS.
Dasher, Bartlein, Impact, LabRadar, etc.
The only real difference is I throw powder with an Auto Trickler setup and she throws with a chargemaster.
I Chrono them every chance I get, but usually only 5-10rnds each before I pick up and move to a different location and start a new string.
My SD's run about 4 and hers around 6 for an average string Those aren't cherry picked, those are normal. I'm on my 4th or fifth barrel, I can't recall ever seeing either rifle go over 8sd, including the cold bore.
I'm having a really, really hard time believing that a 30 or 50 shot consecutive string would push my SD 5 times higher.
 

captainmorgan460

Private
Belligerents
Minuteman
Dec 16, 2017
117
87
34
Sioux Falls, SD
This topic came up in another post in the range reports section. The moral of the story is that if you're using any sort of analysis on a sample size of less than 20-25, you're lying to yourself.

This is .223 data with 5 different charge weights every .2 grains from 23.5-24.3 of Varget. Sorry for not labeling well; it was originally only for personal use. On the left side is SD in fps, on the bottom is the number of samples. It represents a running SD as the number of shots/samples increases over 30-35 rounds.
View attachment 7258249

If you look at the data from 5 shots it's a complete lie. If you look at the data at 10 shots it's a lie. From 20-30 shots you're kinda getting an idea, and IMHO I should have taken this to 50 shots because I think it would tell me that changing charge weight really doesn't change SD appreciably at all.

Most notably, follow the dark blue and orange lines as sample size increases and think about how your perception of a load would be at 5, 10, 15, 20, etc.. shot sample sizes...

Now with that said, most of these ended up hovering around 20fps. Another rifle (6 creed) I shot a 30 shot sample and it was at 12.5fps SD after 30.

From what I have seen so far I would say a TRUE SD of between 10-15fps is excellent. Any 5-10 shot sampling is a waste of your time, other than potentially grabbing a quick average velocity to plug into your calculator. However, even average velocity bounces around and settles out after 20 samples, too.

That's a game changer. For sure haven't seen that before. Thanks for sharing.
 

Ledzep

Chancellor
Belligerents
Jun 9, 2009
2,570
1,705
219
NE
Aside from the looks of it the yellow is the only normal one and the others trending one way or the other leads me to suspect something else is influencing.
Yeah data looks much better when you dismiss 80% of it. ;)

The same experiment can be done with dice. Roll it enough and you'll have strings that roll 4, 5, and 6 more than 1, 2, or 3 and visa versa. It doesn't mean that the test is skewed.
 
Last edited:
  • Like
Reactions: phlegethon

MarinePMI

Ban Cat Handler
Staff member
Commercial Supporter
Online Training Access
Belligerents
Jun 3, 2010
6,108
6,195
219
San Diego, Ca
Just to add: when working in the commercial rimfire market, it was standard practice to test with 10 rd groups for a quick sampling, and 50 rd groups for "no shit, do we need to tweak the load, priming, or other factors that go into loading accurate rimfire ammo, before setting up the machines for a production run".

At first the larger numbers you see, leave you a little disappointed. But after awhile, those numbers seem more realistic. My point being, most don't understand what is really realistic: numbers wise and sampling size.

Sometimes I think people just use 5 shot sampled SD's and ES's as "liquid luck". It becomes false reassurance to many.
 
  • Like
Reactions: kthomas

Ksracer

Private
Belligerents
Feb 14, 2017
183
65
34
I like how you dismiss mathematical formulas about statistics and probability with the same confidence a flat earther dismisses the earth being a globe.
My Chrono data looks similar to what you posted above. Yet your SD wasn't 15 when you added it all together. I think what the take away here is that if you cherry pick strings of 5 or 10, yes, your true SD is probably significantly higher. But if you're consistently getting strings with 2's, 4's and 5's, it's probably not going to climb to 20 plus over 50rnds.
 

phlegethon

Gunny Sergeant
Hessian
Online Training Access
Belligerents
Minuteman
Nov 4, 2018
874
666
99
My Chrono data looks similar to what you posted above. Yet your SD wasn't 15 when you added it all together. I think what the take away here is that if you cherry pick strings of 5 or 10, yes, your true SD is probably significantly higher. But if you're consistently getting strings with 2's, 4's and 5's, it's probably not going to climb to 20 plus over 50rnds.
That’s right. But consistent strings of the same ammo isn’t a sample size of 5, it’s larger, so it has more power. This is more relevant if the sample size really is small, for example during load development.
 
  • Like
Reactions: _Raining