Groups are to small

[Bullgator]

So I watched the podcast by Hornady titled your groups are to small today. So I had about 80 rounds of new Gun Werks brass loaded that needed shooting so I shot a 80 shot string at 300 yards . My ES was 137.9 fps. and my SD was 36.7 average feet per second was 3309.7 .
With my Horizon 22creedmoor my question is what should I expect them to be and does it really matter as long as my group was good ? Which is another question what is considered a good group at 300 yards for a old one eyed man?

 

[JR1200W3]

Well, 3" would be 1 moa. 1.5" would be a half moa. I usually go for 1" groups at 300. I'm happy there but if I'm sub 1.5" I will begrudgingly take it.

If your groups are too small, try some Hornady bullets. That will probably fix the problem.

 

[supercorndogs]

E.S of 30-40 is pretty good

E.S of 100 pretty bad.

It depends on what you are doing with the rifle if it matters.

A point blank zero coyote rifle. Not much diffrence.

A rifle set up for shooting 700+ yards, you are gonna want to be in the pretty good category or batter. SD will give you better idea of likley dispersion vs E.S that can be effected by one large outlier.

 

[JR1200W3]

I'm taking a stats class right now. Distributions, Standard Errors, Confidence Intervals, inference, Hypothesis testing, etc. As I'm taking this class, all these discussions about SDs and ES keep invading my thoughts.

Something experienced shooters have learned intuitively is that the more rounds you fire as you're chrono'ing helps your SD's settle down over time. A lot of your variations in sampling will happen in the beginning. The first 5, 10, or even 20 rounds. But after 20 or 30, more rounds will actually help your SDs look better. This is due to the law of large sample sizes. And how Normal distributions solidify over larger samples. A larger sample creates a more accurate and narrower CI on the normal distribution. Now maybe not with your range, ES. But in terms of SD, you can get the bulk of your variation pretty soon.

 

[secondofangle2]

Well, because hardly anybody ever shoots 80 shot groups at 300 yards, it’s difficult to know. My wild guess is that 3MOA, 80 shot group even if you exclude five or 10 outliers would be pretty damn good.

Same thing with standard deviation: the habit of the entire gun community is to do repeated trials with tiny sample sizes and then hand pick the best ones to place in their mind or online as representative of their group size or standard deviation.

If people recorded mean radius of every single shot out of their gun as well as standard deviation, we would get a better (and BIGGER ) picture of reality.

Having said that a standard deviation of 36 seems pretty high

Also, did you realize you burned up 10% of your 22 Creedmoor barrel with this little test?

 

[Ledzep]

Long term repeatable SD:

Best: Single digit, 5-9fps (rare to see <8fps for a true SD over a large sample, but occasionally happens)
Good/Common: 10-15fps
Okayish: <20fps

Group Size
Best I've seen floats around 0.5-0.6 MOA
Common for PRS style rifles is 0.6-1.2 MOA
Thin profile and light weight rifles are more subject to heat and shooter fundamentals, 1.5-3 MOA can be common.

Regardless of what the weak point is (shooter, barrel, optic, ammo, heat, etc.), the point we're trying to get across is that when you combine it all and shoot a large sample test, you know what the entire system is capable of, and you can adjust max range engagement for acceptable hit probability accordingly.

 

[Harman117]

Take a picture and post it! Let's see that 80 round group

 

[straightshooter1]

So I watched the podcast by Hornady titled your groups are to small today. So I had about 80 rounds of new Gun Werks brass loaded that needed shooting so I shot a 80 shot string at 300 yards . My ES was 137.9 fps. and my SD was 36.7 average feet per second was 3309.7 .
With my Horizon 22creedmoor my question is what should I expect them to be and does it really matter as long as my group was good ? Which is another question what is considered a good group at 300 yards for a old one eyed man?

So, what did you group measure that you got at 300 yds.?

At 300 yds. I'd consider a "good group" to be ~1.5" or less.

An ES of 138 and an SD of 36.7 is pretty bad, in my book. I once had an 50 shot ES of 97 with an SD of 33 when testing a new primer. That was the worst I've ever gotten. When fire forming some new Alpha brass, 80 shots gave me an ES of 29 with an SD of 6.2 that produced .418 MOA group ( this being out of a .308 Ruger Precision Rifle, but with a Krieger barrel). Typically my high round counts yield SD's in the single digits. The factory ammo (mostly match grade) I've fired has yielded SD in the mid teens and ES's no worse than 50.

Hope this gives you some perspective. . .???

 

[JaoeyP]

So I watched the podcast by Hornady titled your groups are to small today. So I had about 80 rounds of new Gun Werks brass loaded that needed shooting so I shot a 80 shot string at 300 yards . My ES was 137.9 fps. and my SD was 36.7 average feet per second was 3309.7 .
With my Horizon 22creedmoor my question is what should I expect them to be and does it really matter as long as my group was good ? Which is another question what is considered a good group at 300 yards for a old one eyed man?

Was this a new barrel?

 

[RegionRat]

A suggestion for a different way of judging your shooting performance when you are shooting long strings at 300 yards (or any other standard known distance) is to just use a standard target and score it.

For example, highpower with a sling, or F-Class from rests, can use a standard 300 yard target and then you would be able to say if your score was High Master, Master, etc.. The F-Class 10 ring at 300 yards is 2.85" and the X ring is 1.42".

Not counting sighters, those games run 15 to 20 shot strings at 300 yards. A High Master runs a >97% score.
When those scores run into "clean" or 100%, then we go by X-count.

There are many folks who have "cleaned" the 300 yard line with all X during matches, but that is not seen at every match. Those are typically 3 strings of 20 shots.

So, the story changes based on if you are shooting with a sling, body support, versus from bag rests.

I don't know how heavy the bbl section runs on the Horizon 22 CM rigs, but if those bbls are not very heavy then the heat from extended strings will burn them up fast. I have 22 CM bbls cut three at a time and just don't worry about it, but I still keep the shots to say 5 at a time with time to cool off in between.

 

[nn8734]

I'm taking a stats class right now. Distributions, Standard Errors, Confidence Intervals, inference, Hypothesis testing, etc. As I'm taking this class, all these discussions about SDs and ES keep invading my thoughts.

Something experienced shooters have learned intuitively is that the more rounds you fire as you're chrono'ing helps your SD's settle down over time. A lot of your variations in sampling will happen in the beginning. The first 5, 10, or even 20 rounds. But after 20 or 30, more rounds will actually help your SDs look better. This is due to the law of large sample sizes. And how Normal distributions solidify over larger samples. A larger sample creates a more accurate and narrower CI on the normal distribution. Now maybe not with your range, ES. But in terms of SD, you can get the bulk of your variation pretty soon.

I had to take a few stats classes for my degree and you're learning the basics which def helps understand MV central tendency and dispersion data better. In my experience for any given centerfire rifle, SD starts out small (0-10 rounds where most people erroneously make their population attributions) then grows until it peaks around 20-25 rounds, give or take then begins to drop back towards a number that's more representative of the parent population, usually getting there by 30-40 rounds, as you say.

When I think Im ready to put a load into production, I'll do drop validation, firing 40-80 rounds of a given load receipe and record the entire string...My SDs in those cases are usually in the teens or low 20s for 5.56 and low-mid teens for .308, which is fine for my own requirements (hitting steel at 1000m and in)...I used to record my sessions only a two-three times a year as I hated carting out that LabRadar but now w/the little Garmin Xero, I record every single session (40-100 rnds).

 

[secondofangle2]

Good Lord, some of you guys shoot a heck of a lot more than I do. Three barrels at a time? Might as well go 4 at a time just like Perelli tires on a Porsche

 

[Baron23]

Good Lord, some of you guys shoot a heck of a lot more than I do. Three barrels at a time? Might as well go 4 at a time just like Perelli tires on a Porsche

What chu’ talking about, Jackson

 

[diggler1833]

I'm assuming the OP fired 80 rounds in a session with a built-for-hunting 22CM, and probably in some sort of hot environment with a moderate pace. The barrel's throat loves him.

Barrels are consumables, I'm not completely ignorant to that fact. However we learned that pretty quickly the OP would have seen some drastic ES and SDs... so this load is really only good for short-range use, even if accuracy was acceptable up close. So what was learned from say, 20 rounds to 80? I've only had SD get lower as the round count climbed over 10... not gone up (but it is possible with poor reloading habits I suppose).

*****

OP, sorry that your velocity was all over the place man. Since AFAIK, ADG makes Gunwerks' brass, I would consider it quality stuff. Even with virgin cases and clean necks, that velocity spread is problematic. If you don't mind sharing your equipment and procedures, some people in here might have some insight to help you out.

Regarding accuracy expectations, that depends on the rifle and purpose. I'm usually content if I can string 10 shots inside 1 MOA @ 300 yards consistently - meaning that I can repeat that group. Environmental factors are going to start impacting here as much as you and your load. Obviously if I do better, that makes me happier. If I shot F-Class, then I'd want to be like that with 20 shot strings.

One eye would make for slower time on target probably, but shouldn't affect your personal accuracy limitations.

 

[Ronws]

Take a picture and post it! Let's see that 80 round group

Reminds me of what I have learned here. Show us the paper or it didn't happen.

 

[Ronws]

@Bullgator

I have watched that and the follow-up and two associated videos, such as mean radius, at least 4 times, not kidding.

I learn something new each time.

Group the rifle for the job. Jayden even mentions this. For example, my activity is hunting, then 7 of the 3 round groups with cooling in between is going to be most germaine to what I do. Especially on the first shot. I may have to take a second shot. Hunters I know have never had to take a third shot. And I feel, as do others, if you are taking more than 3 shots at an animal, something is dreadfully wrong and you need to stop and fix whatever it is.

Also, I heard of a guy, Wade at Texas Predator Hunting (TPH Podcast) do a 5 shot group of one single cold bore shot each day for five days. The dispersion was the same as if he took 5 shots at once.

Also, when to group a rifle. Obviously after break-in. And clean when precision suffers. I would think 50 to 80 rounds should be sufficient.

As to the stats, those an academic interest, IMHO. What size is the group?

"The target doesn't lie." - Erik Cortina

But I have also known Cortina to say that if the rifle is not shooting 3/8" at 100 yards in the first 3 shots or so, he is not going to waste more time and ammo. If it is not doing it in the first few shots, 73 more is not going to make it better.

So, either by factory ammo that works or work up a handload that works and just stick with that.

I have watched Tom Rivers - Simple Living. He will work up a load for his hunting rifle. He is not even doing competitions. He just wants predictable precision for hunting. Granted, he is in the Southeast and the forests are thick.

 

[straightshooter1]

Good Lord, some of you guys shoot a heck of a lot more than I do. Three barrels at a time? Might as well go 4 at a time just like Perelli tires on a Porsche

How about 4 barrels at a time?

 

[JR1200W3]

I had to take a few stats classes for my degree and you're learning the basics which def helps understand MV central tendency and dispersion data better. In my experience for any given centerfire rifle, SD starts out small (0-10 rounds where most people erroneously make their population attributions) then grows until it peaks around 20-25 rounds, give or take then begins to drop back towards a number that's more representative of the parent population, usually getting there by 30-40 rounds, as you say.

When I think Im ready to put a load into production, I'll do drop validation, firing 40-80 rounds of a given load receipe and record the entire string...My SDs in those cases are usually in the teens or low 20s for 5.56 and low-mid teens for .308, which is fine for my own requirements (hitting steel at 1000m and in)...I used to record my sessions only a two-three times a year as I hated carting out that LabRadar but now w/the little Garmin Xero, I record every single session (40-100 rnds).

In clinical research it's common for studies to have small sample sizes and their results struggle to show a significant difference through a P value and Alpha. So Meta-analyses are used to aggregate independent comparisons and make an inference through a narrowed down confidence interval. Comparing the difference of means, Null and Alternative hypothesis, bootstrapping, T tests, and finally meta-analysis have consumed my last two weeks.

My wife has 20 years of experience in clinical research and a Doctorate in Microbiology so for the first time we can have a conversation about stats, lol. I was picking her brain about Meta-analysis this evening for an assignment and it occurred to me that one of the counter-arguments to the massive sample size argument is the value of reproducibility. Or the guys that say, "But every time I Chrono my five rounds groups I'm getting SDs under 8fps, so is a 30rd group really necessary?" Aggregating the SDs six individual 5 rd groups and demonstrating a consistent trend and tendency...that is reproducible has a ton of value. And it is a form of meta-analysis that is considered the gold standard in clinical research to deal with small sample studies.

I could see two sides to these arguments.

1. You need a larger sample size to have a better estimate of your average MV and SDs.

Or...

2. Everytime I shoot a five round group, my SDs are less than 10, over 6 shooting sessions.

With a 30rd group, all fired on one day, from one loading session, in A set of conditions.....even though you have a large sample size it doesn't mean that when you go out again and shoot another 30rd group that it's going to be reproducible. While if you can demonstrate reproducibility over multiple loading sessions, atmospheric conditions, etc....your meta-analysis of many small samples may have more value.

On the other hand, there's going to be a difference in means from multiple small samples, inevitably. Say you do shoot 6 x 5rd groups and your 6 different means are 2950, 2954, 2959, 2952, 2963, and 2945. You should be loading a summarized mean of means or the center of a 95% CI generated by those means into your Kestrel. Not just the last 5rd mean you shot. Whereas a 30rd group will greatly simplify the mean, as long as it's reporducible.

 

[nn8734]

1. You need a larger sample size to have a better estimate of your average MV and SDs.

I favor this argument. To me, it's more representative/indicative of how any random round or sample/subset of rounds in my entire prod lot is likely to behave on any given day.

Note that I am not saying it's the right answer - Im not qualified to tell you or anyone else what's right vs wrong when it comes to stats arguments. Just my opinion...

 

[JR1200W3]

I favor this argument. To me, it's more representative/indicative of how any random round or sample/subset of rounds in my entire prod lot is likely to behave on any given day.

Note that I am not saying it's the right answer - Im not qualified to tell you or anyone else what's right vs wrong when it comes to stats arguments. Just my opinion...

I think I'm going to load 30rds and shoot them at 100yds. Take photos of the group every 5rds. Then calculate the mean, SD, 95% CI for the total 30, and then do the same for the series of 5rd groups within the overall sample. And then randomize the 5rd mini-samples from the total 30rd population. See what the differences would be. I guess if my SDs and range is pretty tight, I won't see much difference. But it might be interesting to see if there is a difference between these two arguments.

 

[nn8734]

I think I'm going to load 30rds and shoot them at 100yds. Take photos of the group every 5rds. Then calculate the mean, SD, 95% CI for the total 30, and then do the same for the series of 5rd groups within the overall sample. And then randomize the 5rd mini-samples from the total 30rd population. See what the differences would be. I guess if my SDs and range is pretty tight, I won't see much difference. But it might be interesting to see if there is a difference between these two arguments.

Yea will be interesting if your mean of means in the series of five rounders is substantially different. It shouldnt be if both samples (30 rnd string and the series of five-round groups) are all shot in the same conditions, etc.

Will these rounds be factory ammo or your hand loads?

 

[JR1200W3]

Yea will be interesting if your mean of means in the series of five rounders is substantially different. It shouldnt be if both samples (30 rnd string and the series of five-round groups) are all shot in the same conditions, etc.

Will these rounds be factory ammo or your hand loads?

I don't really buy ammo. The most likely would be my 25GT. 37gr H4350, Alpha brass, 135 Hybrids. I get sub 8fps SDs everytime I shoot it. If I was thorough, I should probably do that gun and load, and then something that usually produces more variation. I took my SP-10 out today to see if I could get a beater load shooting halfway decent. Norma 130gr GTs, 42.5gr StaBall, Hornady brass, Ginex primers. I shot 3 x 5rd groups and a 10rd group. SDs were 19, 8, 5.7, and 19.

 

[nn8734]

I don't really buy ammo. The most likely would be my 25GT. 37gr H4350, Alpha brass, 135 Hybrids. I get sub 8fps SDs everytime I shoot it. If I was thorough, I should probably do that gun and load, and then something that usually produces more variation. I took my SP-10 out today to see if I could get a beater load shooting halfway decent. Norma 130gr GTs, 42.5gr StaBall, Hornady brass, Ginex primers. I shot 3 x 5rd groups and a 10rd group. SDs were 19, 8, 5.7, and 19.

I keep a small amt of factory ammo around for T/E purposes as well as to break in new barrels prior to load dev but beyond that, I load all my ammo.

My approach treats those sep groups as one and would avg the SDs and compute the range across all of those rounds...Testing both ways is fine - esp if you take a long break between samples and let the barrel cool each time. Maybe that would change MVs materially, maybe not - you will see when you do your testing...

 

[JR1200W3]

.Testing both ways is fine - esp if you take a long break between samples and let the barrel cool each time. Maybe that would change MVs materially, maybe not - you will see when you do your testing...

That's a good point. Any other tips to high round count testing? I don't normally do that. Could be helpful to know.

I watch Preston Moore's videos and always wonder how clean he's truly breaking those 30rds

 

[nn8734]

That's a good point. Any other tips to high round count testing? I don't normally do that. Could be helpful to know.

I watch Preston Moore's videos and always wonder how clean he's truly breaking those 30rds

Yes - dont focus on group size at 100m as it's only weakly correlated to MV consistency. In fact, I wouldnt shoot at 100m at all during these tests. Id do drop validation on steel at progressively further distances so you can kill two birds with one stone (gather mv test data to confirm/refute hypothesis and get some good dope on your ammo to see what it's doing at distance to record in your dope book). Will keep you focused on target vs group size which lots of folks get hung up on when also trying to get MV consistency data).

Do grouping drills separately where you're focused on one particular thing or another (ie zero shift at 100m when shooting prone vs off a tripod or barricade, etc).

 

[JR1200W3]

Yes - dont focus on group size at 100m as it's only weakly correlated to MV consistency. In fact, I wouldnt shoot at 100m at all during these tests. Id do drop validation on steel at progressively further distances so you can kill two birds with one stone (gather mv test data to confirm/refute hypothesis and get some good dope on your ammo to see what it's doing at distance to record in your dope book). Will keep you focused on target vs group size which lots of folks get hung up on when also trying to get MV consistency data).

Do grouping drills separately where you're focused on one particular thing or another (ie zero shift at 100m when shooting prone vs off a tripod or barricade, etc).

Hmmm. That just opened another can of worms. We've known for a while that your shooting position can yield different MVs. Prone vs standing barricade is going to provide a different resistance to the gun in recoil and it's been demonstrated that less resistance will produce slower velocities. I don't think I want to introduce that level of variable.

I agree that SDs and ES don't really correlate to accuracy at close range, only in the sense of vertical at distance and the (unofficial) law of good vs shitty ammo. Rimfire ammo will show us this fake law occasionally.

 

[nn8734]

Hmmm. That just opened another can of worms. We've known for a while that your shooting position can yield different MVs. Prone vs standing barricade is going to provide a different resistance to the gun in recoil and it's been demonstrated that less resistance will produce slower velocities.

No, not suggesting you do that in combo w/this other test...Do that after you have put the load into production...I always do so I can see what the shift looks like...

 

[Doom]

Which is another question what is considered a good group at 300 yards for a old one eyed man?

There is no true one number answer for 300yds. The answer is dependent on many factors. A better question is what do you need and what can you consistently shoot at 100yds. At 100yds external ballistic factors, such as environmental play very little into the group size, it is primarily internal ballistics, rifle harmonics, and SHOOTER. Your goal at 300 is 3x whatever YOU can do at 100yds. Likely very difficult to achieve but still the goal. Shooting a 80 round group and picking on two shots tells you next to nothing about anything other than the fact that as the shot count increases the extreme spread is going to increase. If you are a competitive shooter then the shot count that matters is whatever string count in a relay you shoot for record. Anything beyond that is noise.

 

[Doom]

Yes - dont focus on group size at 100m as it's only weakly correlated to MV consistency. In fact, I wouldnt shoot at 100m at all during these tests. Id do drop validation on steel at progressively further distances so you can kill two birds with one stone (gather mv test data to confirm/refute hypothesis and get some good dope on your ammo to see what it's doing at distance to record in your dope book). Will keep you focused on target vs group size which lots of folks get hung up on when also trying to get MV consistency data).

Do grouping drills separately where you're focused on one particular thing or another (ie zero shift at 100m when shooting prone vs off a tripod or barricade, etc).

My post kinda goes opposite your recommendation but the reason is the OP needs to understand that MV is purely a function of the ammunition consistency which definitely affects long distance group size and is not a factor per se (very small factor) in defining what the rifle in his hands is capable of. Determine that at 100yds so you know what to expect as the goal at longer distance.

 

[diggler1833]

I don't really buy ammo. The most likely would be my 25GT. 37gr H4350, Alpha brass, 135 Hybrids. I get sub 8fps SDs everytime I shoot it. If I was thorough, I should probably do that gun and load, and then something that usually produces more variation. I took my SP-10 out today to see if I could get a beater load shooting halfway decent. Norma 130gr GTs, 42.5gr StaBall, Hornady brass, Ginex primers. I shot 3 x 5rd groups and a 10rd group. SDs were 19, 8, 5.7, and 19.

I've been reading your posts man, and I like your approach. I was once forced to do Statistics I and II, then Quantitative Data and Analysis, and a couple other statistics-related classes for my final degree. I'll never argue against more data or inputs, but I do believe that sometimes one can collect data that is relatively unnecessary to their practices. Will an 80 round sample provide substantially more accurate data? Yes. Does it apply to the guy who is going to be shooting a few rounds in field conditions from unknown positions? Not nearly as much *I believe* because external factors are going to be the random variable and influence the previous data.

I'm interested in seeing just how much your mean radius and velocity readings changes from your 5-shot group aggregates, to the end product of your 30-round group. I've always been in the camp of doing (3-4) five shot groups, to include shooting a couple at distance to proof a load (vertical dispersion as well as accurate velocity), and then sticking with it if the results are fairly consistent. That is simple man's terms.

If you could also record the environmental factors throughout your group shooting (wind etc..), that would be awesome.

 

[Aftermath]

Which is another question what is considered a good group at 300 yards for a old one eyed man?

I am led to believe that your groups will improve if you are not only an old one eyed man but also a fat one.

 

[nn8734]

My post kinda goes opposite your recommendation but the reason is the OP needs to understand that MV is purely a function of the ammunition consistency which definitely affects long distance group size and is not a factor per se (very small factor) in defining what the rifle in his hands is capable of. Determine that at 100yds so you know what to expect as the goal at longer distance.

My recommendation is based on my own experience; I know a lot of folks, including me, get hung up on group size at 100m putting too much pressure on themselves to keep it tight. They end up throwing that last shot in the sample and may actually reject the load recipe when it was shooter error that opened the group...It's like getting ball bound when playing golf (if you play at all).

Agree w/your point about vertical dispersion at distance (horizontal isn't important as wind is a much greater influencer of hz dispersion at distance than the load itself, assuming shooter runs the fundamentals consistently-correctly. If the load is stringing up and down, that's an obvious problem. I start to see that on target when SDs hit the low-mid 20s in most short and long action cartridges. Anything in the teens and less should be GTG throughout the round's entire supersonic envelope.

 

[Baron23]

In clinical research it's common for studies to have small sample sizes and their results struggle to show a significant difference through a P value and Alpha. So Meta-analyses are used to aggregate independent comparisons and make an inference through a narrowed down confidence interval. Comparing the difference of means, Null and Alternative hypothesis, bootstrapping, T tests, and finally meta-analysis have consumed my last two weeks.

My wife has 20 years of experience in clinical research and a Doctorate in Microbiology so for the first time we can have a conversation about stats, lol. I was picking her brain about Meta-analysis this evening for an assignment and it occurred to me that one of the counter-arguments to the massive sample size argument is the value of reproducibility. Or the guys that say, "But every time I Chrono my five rounds groups I'm getting SDs under 8fps, so is a 30rd group really necessary?" Aggregating the SDs six individual 5 rd groups and demonstrating a consistent trend and tendency...that is reproducible has a ton of value. And it is a form of meta-analysis that is considered the gold standard in clinical research to deal with small sample studies.

I could see two sides to these arguments.

1. You need a larger sample size to have a better estimate of your average MV and SDs.

Or...

2. Everytime I shoot a five round group, my SDs are less than 10, over 6 shooting sessions.

With a 30rd group, all fired on one day, from one loading session, in A set of conditions.....even though you have a large sample size it doesn't mean that when you go out again and shoot another 30rd group that it's going to be reproducible. While if you can demonstrate reproducibility over multiple loading sessions, atmospheric conditions, etc....your meta-analysis of many small samples may have more value.

On the other hand, there's going to be a difference in means from multiple small samples, inevitably. Say you do shoot 6 x 5rd groups and your 6 different means are 2950, 2954, 2959, 2952, 2963, and 2945. You should be loading a summarized mean of means or the center of a 95% CI generated by those means into your Kestrel. Not just the last 5rd mean you shot. Whereas a 30rd group will greatly simplify the mean, as long as it's reporducible.

I don't believe that anybody said that the total data set needs to be shot sequentially....but there are considerations of environment, barrel condition/cleanliness, and condition of powder/primers/cases that can significantly impact the validity of aggregating independent data sets (groups, if you will).

Cheers....and I absolutely loathed statistics when I had to take it...oh, almost 50 years ago.

 

[JR1200W3]

I don't believe that anybody said that the total data set needs to be shot sequentially....but there are considerations of environment, barrel condition/cleanliness, and condition of powder/primers/cases that can significantly impact the validity of aggregating independent data sets (groups, if you will).

Cheers....and I absolutely loathed statistics when I had to take it...oh, almost 50 years ago.

Mmmm, the way I interpret the statements, I think the mindset that you need to shoot a 30rd group vs 6 x5rd inherently implies sequentially. It's the only distinguishing difference.

I do agree that bore condition, heat in the chamber, and rapidly changing atmospherics are variables in both methods. I've been thinking about this all morning, in between making box plots in Posit.

Specifically the idea that 30 demonstrates natural variation that 5 (6x5rds) won't. I have a hard time with that idea. I know that in pure statistics it's a truth. Natural variation in data sets is an absolutism. But in reloading and shooting the type of variations were talking about aren't natural. There's a cause. Variation at the level of 2956 vs 2949 sure....but not 2956 vs 2986. Something happened to that second loaded round that caused a 30fps jump and as reloaders we want to control the variables to eliminate that much variation. So....

It doesn't make sense that if you were to load 30rds and then shoot them in two different methods....either a complete 30rd string or a series of six 5rd groups over a longer stretch of time(or even at different locations) that you would get higher variations in the 30 and then those variations would never appear in the 6x5 because we're constantly stopping at 5. The type of significant variations were talking about here are caused by a reloading component or method, so if that round that's going to be 30fps faster exists in the 30rd group,... it's going to exist somewhere in the six 5rd groups. The only difference that may exist in the 30rd group is, as you mention, increased heat and bore condition changes that could occur from a 30rd string that you won't get in a slower 6x5. Which means the variation is induced by your method, not just the pure idea of a larger sample size. A statistical mind doesn't agree in theory. But in the context we're talking about here, it can't be true.

I think at the end of the day you and I are kind of saying the same thing, I'm hyper focusing on the pure fact of statistics doctrine that says larger data sets will show more variation than small data sets. Because that's the logic behind the "groups are too small" crowd. I completely understand where they're coming from. They're coming from undeniable truth in statistics, but the variations we think are significant in reloading come from something we did as reloaders to cause it. So it's not just random chance that you can cut off at 5rds and start over.

I think there is some room for a counter-argument in the above where over a 30rd string, we may see a trend that can be isolated within a 5rd string and the mean and SD will appear better. As in RDS 1-10 are pretty even and then we just get a gradual increase in rounds 11-15, and it normalizes 16-30. That trend will be compared against all 30 in a 30rd string although it will be muted in mean and SD by a larger data set, but still there. In a series of 6x5 that trend will get isolated into one 5rd group and may have a slightly higher SD than the other 5rd data sets, but won't ever get compared to some extremes in the remaining five sets. I understand this can be a function of the 30 vs 6x5, but I think that level of natural variation will hide pretty well in a 30rd string still.

 

[JR1200W3]

I've been reading your posts man, and I like your approach. I was once forced to do Statistics I and II, then Quantitative Data and Analysis, and a couple other statistics-related classes for my final degree. I'll never argue against more data or inputs, but I do believe that sometimes one can collect data that is relatively unnecessary to their practices. Will an 80 round sample provide substantially more accurate data? Yes. Does it apply to the guy who is going to be shooting a few rounds in field conditions from unknown positions? Not nearly as much *I believe* because external factors are going to be the random variable and influence the previous data.

I'm interested in seeing just how much your mean radius and velocity readings changes from your 5-shot group aggregates, to the end product of your 30-round group. I've always been in the camp of doing (3-4) five shot groups, to include shooting a couple at distance to proof a load (vertical dispersion as well as accurate velocity), and then sticking with it if the results are fairly consistent. That is simple man's terms.

If you could also record the environmental factors throughout your group shooting (wind etc..), that would be awesome.

Will do. I'm going to shoot at an indoor range, all prone, on the same target. The only thing that will change is 30 vs aggregating 6x5.

I think you hit the nail on the head. At least in my argument. Statistics state that larger data sets will show more variation. It is a doctrinal truth. If you're arguing against that, you're just ignorant of math. But it's a bit out of context in our environment. There's natural variation/ random chance and then there's outliers that are so noticable they are caused by something we did as reloaders. Natural variation and random chance: 2950, 2953, 2948, 2949, 2951. Reloading induced outliers caused by a mechanical or chemical catalyst: 2950, 2953, 2948, 2949, 2989. That isn't random chance. That is an outcome we seek to control and eliminate. That 40fps increase in velocity can represent a .2mrad to .5mrad vertical at 1000 depending on your BC. Since it is caused by an anomaly in reloading or component manufacturing, it will be present in a 5rd or 30 round group. It doesn't need a larger data set to present itself.

And these types of outliers are often culled or corrected for in studies because they aren't representative of the population. P value corrections, flagged outliers. In our case, it isn't random chance that a 30rd group will demonstrate that a 5rd group won't.

 

[CK1.0]

I tend to prefer larger sample sizes with my loads versus just shooting a few 3-5 shot groups and trying to judge things based on that. But IDK if it takes 80 rounds to know what you've got.

I like to shoot my 20-30 rounds in the form of successive smaller groups (meaning I'll shoot a bunch of successive groups back to back while keeping the chrono running), before making decisions about making load changes. I typically use 20 shots as my measuring stick, but sometimes I will stretch that shot count up to 30 just to prove things out to myself.

To me, all this stuff always comes back to "cone of fire"...

IMHO, the recipe for a good load is pretty simple: "Can I print acceptably small groups at 100 yards on command?" (for me, that equals .3" on good days, ~.5" even when the monkey pulling the trigger isn't that good), then, "What do my SDs look like?" I'm looking for an SD of ~7 or better over 20 shots.

That way, if I know what my gun does up close and know that my speeds are stable, I can be pretty confident in what my cone of fire looks like.

I've heard guys say that sometimes loads with the lowest avaerage SDs don't print the smallest groups at 100, but I don't give shit about that or necessarily believe it, because IME I've never had a load that yields low SDs shoot bad, and tend to think small groups have more to do with the shooter than necessarily the magic load.

FWIW, the attached PDF is what I use as my zeroing, all-purpose 100-yard target (in case it might help someone else). It lets you know what you've got pretty quickly, a typical practice string might look like this:

 

[JR1200W3]

I'm not sure what you expect us to take away from that photo.

First you say, "I tend to prefer larger sample sizes with my loads versus just shooting a few 3-5 shot groups and trying to judge things based on that."

And then you post this:

I'm looking at it thinking, (top to bottom, left to right) "Shit, Shit, Good, Good, Shit, Shit, Shit, Good, Shit" and left with such a mixed bag of good and shit that I would consider this a Null hypothesis proven out. Looks random to me.

I mean "a few" typically means 3 and you shot 9, but this kind of fits the definition of, "... a few 3-5 shot groups and trying to judge things based on that."

Even if you overlaid all those dots, it would be a kind of ugly 27rd group. Certainly not a .3-.5".

What does this demonstrate to you?

 

[CK1.0]

I'm not sure what you expect us to take away from that photo.

First you say, "I tend to prefer larger sample sizes with my loads versus just shooting a few 3-5 shot groups and trying to judge things based on that."

And then you post this:
View attachment 8743779

I'm looking at it thinking, (top to bottom, left to right) "Shit, Shit, Good, Good, Shit, Shit, Shit, Good, Shit" and left with such a mixed bag of good and shit that I would consider this a Null hypothesis proven out. Looks random to me.

I mean "a few" typically means 3 and you shot 9, but this kind of fits the definition of, "... a few 3-5 shot groups and trying to judge things based on that."

Even if you overlaid all those dots, it would be a kind of ugly 27rd group. Certainly not a .3-.5".

What does this demonstrate to you?

That’s just a practice target, shot positionally off a bag/prop, or modified prone off a prop, or tripod rear/cattle gate, or whatever else I was working on that day… don’t get your panties in a bunch.

That’s not a pic of my best first-prize groups, just a random target. I was trying to show what a decent cone of fire looks like to me, and you can pretty much tell from looking at it what was me and what was the gun, those dots are 1/4” and most of those groups clip a dot or better (which is what I’m looking for). IIRC that target was modified-prone off a bench or one of those giant square water containers (I say that because I usually have a couple/few groups uglier than those depending on what I was doing ).

I don’t have a pic of one of my load development targets as I don’t really take pictures of those anymore because I’m always looking for the same shit: I want all the rounds to touch, and be able to do it back to back to back, etc, period. I want my average group size off a bench or prone to be .3” (to ~.5” on bad days).

My point was shoot enough to know what consistently looks good and wasn’t just a fluke, try to a get a good idea what you’re looking at as far as what you’ve got for a cone of fire, you can’t do that shooting too little or trying small samples of everything as far as different loads, you’ve got to shoot enough to know what’s you and what is the gun/load (and if the chrono numbers looks good it’s usually our fault IME).

Same gun/load, off a bag and various props at 750 yards:

 

[JR1200W3]

Went out and shot 30rds today as promised. Created a new thread to show the results as courtesy to the OP.

Thread '5rd groups aren't statistically significant! Wanna bet?!!' https://www.snipershide.com/shootin...-statistically-significant-wanna-bet.7265325/

 

[Bullgator]

I am led to believe that your groups will improve if you are not only an old one eyed man but also a fat one.


View attachment 8743669

30 more pounds and I’ll be good to go.

 

[Bullgator]

Take a picture and post it! Let's see that 80 round group

These are two 300 yard targets.





GROUP


hehwood Casey


Birchwood Casey


Shoot-N © Jarget




ards


髦


$


G

 

[Bullgator]

Probably 300 rounds

 

[Bullgator]

Well, because hardly anybody ever shoots 80 shot groups at 300 yards, it’s difficult to know. My wild guess is that 3MOA, 80 shot group even if you exclude five or 10 outliers would be pretty damn good.

Same thing with standard deviation: the habit of the entire gun community is to do repeated trials with tiny sample sizes and then hand pick the best ones to place in their mind or online as representative of their group size or standard deviation.

If people recorded mean radius of every single shot out of their gun as well as standard deviation, we would get a better (and BIGGER ) picture of reality.

Having said that a standard deviation of 36 seems pretty high

Also, did you realize you burned up 10% of your 22 Creedmoor barrel with this little test?

Take a picture and post it! Let's see that 80 round group

II don’t have pic of the 80 .

 

[Bullgator]

So, what did you group measure that you got at 300 yds.?

At 300 yds. I'd consider a "good group" to be ~1.5" or less.

An ES of 138 and an SD of 36.7 is pretty bad, in my book. I once had an 50 shot ES of 97 with an SD of 33 when testing a new primer. That was the worst I've ever gotten. When fire forming some new Alpha brass, 80 shots gave me an ES of 29 with an SD of 6.2 that produced .418 MOA group ( this being out of a .308 Ruger Precision Rifle, but with a Krieger barrel). Typically my high round counts yield SD's in the single digits. The factory ammo (mostly match grade) I've fired has yielded SD in the mid teens and ES's no worse than 50.

Hope this gives you some perspective. . .???

Thanks that tells me how far I have to go. I’ve only been reloading for about 4 months now I’ve bought every gadget or tool they make. Powders ,brass and bullets watched every YouTube video on the web. I just started fooling with this 22 creedmoor . Iam on my second barrel on my Seekins Hit 6.5 PRC . My set is basically I can load a few rounds walk 30 feet and shoot out to 300 yards my point is I’ve shot a lot of rounds. I can find great groups and low es/sd but I struggle to the same load to produce the same results from day today .

 

[straightshooter1]

Thanks that tells me how far I have to go. I’ve only been reloading for about 4 months now I’ve bought every gadget or tool they make. Powders ,brass and bullets watched every YouTube video on the web. I just started fooling with this 22 creedmoor . Iam on my second barrel on my Seekins Hit 6.5 PRC . My set is basically I can load a few rounds walk 30 feet and shoot out to 300 yards my point is I’ve shot a lot of rounds. I can find great groups and low es/sd but I struggle to the same load to produce the same results from day today .

You're loads are not producing the same es/sd's? What kind of differences are you getting?

I don't get exactly the same ES's and SD's with the same load either, but they're very close, where I'm just looking for SD's in the mid single digits. Here's a couple of pic I took last year of my data for my 6.5 PRC and one for my .308 (and I'll point out that those results are not all the same load, as I'm a tinkerer ):

 

[sinister]

Here's some trivia:

March 31st, 2017

500-Round Group at 300 Meters — Now That’s a Test!​

For load development, some guys shoot 3-shot groups. Other guys shoot 5-shot groups, or even 10-shot strings. But for testing its projectiles, Sierra Bullets takes it to another level entirely. A while back Sierra was testing its .30-Caliber 175gr HPBT MatchKing in the Sierra underground tunnel. The results are show above — a FIVE HUNDRED Round group!

The tunnel testers sent five full boxes of bullets down-range. Here are 500 Shots shot in a 300 meter tunnel. The group size is 2.82 inches (that’s edge to edge of the farthest shots, less the bullet diameter). This was a pressure/velocity test for a commercial customer. The Cartridge was .308 Win, loaded at 2.800″.The powder was Reloder 15. A 26″ barrel was shot from a return to battery rest. The gun was cleaned every 125 rounds and two foulers shot.

What do you think — could you beat this group from a bench for 500 rounds?

One Facebook poster joked: “500-round group? Everyone knows anything less than 1000-round groups are a waste of time and statistically irrelevant.”

Sierra’s 300 Meter Testing Tunnel
Ever wonder how (and where) Sierra tests its bullets? The answer is underground, in a 300-meter test tunnel located under Sierra’s factory in Sedalia, Missouri. The photo above shows the construction of the tunnel back in May, 1990. Like most bullet manufacturers, Sierra does live-fire bullet testing of its projectiles. Sierra’s 300-meter test range is the longest, manufacturer-owned underground bullet test facility in the world. Sierra offers free tours of the test tunnel as part of Sierra’s Factory Tour Program.

 

[chungus]

What kind of groups are you getting with that horizon? I have one and at 100 yards I have not been able to get a 3 shot group less than 1/2” no matter how many loads, bullets, powders, I have tried. It almost always shoots under 1” regardless of the load so it isn’t horrible but it is my only high end rifle that I can’t get a 1/2” group out of.

 

[Bullgator]

What kind of groups are you getting with that horizon? I have one and at 100 yards I have not been able to get a 3 shot group less than 1/2” no matter how many loads, bullets, powders, I have tried. It almost always shoots under 1” regardless of the load so it isn’t horrible but it is my only high end rifle that I can’t get a 1/2” group out of.

Well to be honest I can get 1/2 groups almost with anything. I’ve shot some.300s and can often get .250 on two out of a 4 shot group. Mine likes 75 grain ELD match with 4350 at 39.1 grains up to 40.2 grains with a coal at 2.75 with the Hornady comparison and Gunwerks brass using a .224 expander. I will say that 2.075 seems to be the best coal across all the powders I’ve tried. The last powder was CFE223 at 44.2 grains it shot around 4,000 fps without any pressure signs but I dropped way back to .40 grains.