Define a 1MOA rifle

[jcampbellsmith]

I think you can gain some insights by looking at competitions, one opportunity is F-class. Their results are published - Results – Great Britain F Class Rifle Association

Their targets are of known dimensions. Look at the recent Europeans. The FTR winner shot 250 5's. 1/2 minute bull is the target. Their season has just finished. They clearly aren't content with 1 MOA rifles.

Similarly, .22 shooters and PRS shooters have their handles on what is acceptable accuracy. The recurring theme when I get involved in any of these sports is build quality. The winners are spending lots of money on the best kit and the best ammunition. Their measure is where they come in competitions.

Personally, I won't shoot a deer rifle that doesn't place 3 shots in less than 0.5 MOA at 200yds when fully rested.

Regards

JCS

 

[billgatese30]

Why would I?

You're an engineer. You deal daily with measuring tolerances and ensuring precision.

What, to you, would be a robust way to measure the accuracy of a given set up?

Measuring the results accurately is another discussion in itself. Gauge R&R is another topic we could spiral into but I suggest not today

 

[jcampbellsmith]

Measuring the results accurately is another discussion in itself. Gauge R&R is another topic we could spiral into but I suggest not today

Electronic targets are one answer. I'm off shortly to shoot indoors on electronic targets. My results above are using a Silver Mountain Target.
Regards
JCS

 

[tackb]

Thanks.

You say 10X10rd groups.

Would all 10 groups need to be under 1"?

The reason I ask is that when I run simulations, for this to be satisfied, the x and y error need to be very small (having an SD of 0.1). I don't think very many rifles would satisfy this. The CEP for a set up that satisfies this would need to be 0.1".

i think yes ?

for the system to be truly a 1moa rifle it would need to deliver that for an acceptable barrel life

i'd be confident that my comp rifle with match ammo would under the described conditions shoot ten 10rd groups of 1 moa , it mostly shoots way better , way better! but i am the weak link

very interesting discussion to be fair , makes me consider stuff i haven't thought about , i shot a less that 1/2'' group at 200m with that comp rifle the other day but that does not make it 1/4moa rifle (or shooter!)

 

[Mungo]

10 shots is still too small of a sample size in my opinion.

As I stated above, 25 is generally accepted as the minimum for a statistically appropriate sample size, but if you could go to 50 or 100 then even better, as this will allow you to predict the population standard deviation more accurately than you could from a smaller sample. However increasing the sample size is a law of diminishing returns in its ability to accurately predict the population SD and of course we would have other factors such as barrel cleanliness coming into it as well.

Eliminate what variables you can or reduce as far as practicable (wind/ammo variation/barrel temperature varianxes etc), then shoot a statistically significant sample size, assess for a normal distribution, then remove outliers of >3SD either side of the estimated population mean. This is generally accepted as good statistical practice unless there are strong practical reasons for deviating.

Also just to clarify, I've done this subject as a career for over 20 years for multi billion dollar manufactures of everything from specialist batteries to medical devices. I'm a process control subject matter expert and a six sigma black belt (for what its worth).

You'll appreciate this then...

This are groups of 25. Simulated with POI error varying from an SD of 0.1 to an SD of 2.

 

[nun_hunter]

Measuring the results accurately is another discussion in itself. Gauge R&R is another topic we could spiral into but I suggest not today

Exactly by Simon Winchester is a very good read for anyone interested in precision and accuracy.

 

[billgatese30]

You'll appreciate this then...

This are groups of 25. Simulated with POI error varying from an SD of 0.1 to an SD of 2.

View attachment 391795

All that tells is that SD must be below a certain value, which is a constant relationship.

https://encrypted-tbn0.gstatic.com/images?q=tbn:ANd9GcR7w4E_wqiUydT_fu6hYT1qBOQLdtnQ0QMaig&s

6SD is =99. 9997% of a normal distribution. 8SD is 99.99999% recurring. 1/6 = 0.1667, 1/8 is 0.125. Therefore we know we need SD to be in the 0.125~0.1667 range to be a 1MOA rifle 99.odd % of the time. What we don't know though is what the actual SD is. The only way to find that out is to shoot 25-50 shots, check its normally distributed (and nothings creeping in to throw our data) and calculate the estimated SD for the population. With a sample this size I wouldn't worry about the ES and focus on the SD. ES is more useful when sample sizes are small which can cause inaccurate SD calculations fue to the formulas for SD.

 

[Mungo]

All that tells is that SD must be below a certain value, which is a constant relationship.

https://encrypted-tbn0.gstatic.com/images?q=tbn:ANd9GcR7w4E_wqiUydT_fu6hYT1qBOQLdtnQ0QMaig&s

6SD is =99. 9997% of a normal distribution. 8SD is 99.99999% recurring. 1/6 = 0.1667, 1/8 is 0.125. Therefore we know we need SD to be in the 0.125~0.1667 range to be a 1MOA rifle 99.odd % of the time. What we don't know though is what the actual SD is. The only way to find that out is to shoot 25-50 shots, check its normally distributed (and nothings creeping in to throw our data) and calculate the estimated SD for the population. With a sample this size I wouldn't worry about the ES and focus on the SD. ES is more useful when sample sizes are small which can cause inaccurate SD calculations fue to the formulas for SD.

Yes.

In the field, most people are shooting 3 or 5 shots groups and using extreme spread as their measure of group size. As the fig below reveals, this means that it can be very easy to convince yourself you have a 1 MOA gun.

 

[billgatese30]

I think to revert back to the original question; how do we verify if a rifle is <1MOA, I think the easiest way to answer that in one sentence would be "Shoot a large enough sample size to determine an accurate estimate of population SD, and then have a x% confidence of all shots going within 1MOA". However you/your client would have to fill in the "X" they would be happy with. Me, I'd go 95% as that tends to be the standard accepted in stats unless there's a good reason to be 90, 98, 99! etc.

 

[billgatese30]

Also, on your charts above, what calculation are you using to the resultant ES as it looks like a similar number of data points, but they are titled 3 shot, 5 shot, 10 shot etc. I don't understand what is being used to predict a larger result given the same SD?

 

[Utectok]

I’m a purist a true one moa rifle must put every shot inside an inch at 100yds. So basically most 1/2 and 1/4 moa rifles are actually moa. And most moa rifles are about 1.5moa. I can group my rifle at 1/2moa but can put 30 shots in an inch so it’s really about an moa rifle. Which is actually quite unusual in my experience. I think some f class guys have 1/2moa rifles that shot 20 shots in 1/2 moa but 99 per cent of three shot 1/2” guns are shooting 1.25” really if you sample it properly

 

[Mungo]

I think to revert back to the original question; how do we verify if a rifle is <1MOA, I think the easiest way to answer that in one sentence would be "Shoot a large enough sample size to determine an accurate estimate of population SD, and then have a x% confidence of all shots going within 1MOA". However you/your client would have to fill in the "X" they would be happy with. Me, I'd go 95% as that tends to be the standard accepted in stats unless there's a good reason to be 90, 98, 99! etc.

I think this is the right approach, and is roughly where I started. I think I may have expressed myself poorly at the start, because what I was really looking to see was whether there is a standard value of X in your above example.

As you say 0.95 is the standard in statistics (at least in Biology - Physicists often use 0.99 or beyond).

So really I should just go with what I tell my students in basic stats lectures: n=30, alpha=0.05 and call it good.

 

[Mungo]

Also, on your charts above, what calculation are you using to the resultant ES as it looks like a similar number of data points, but they are titled 3 shot, 5 shot, 10 shot etc. I don't understand what is being used to predict a larger result given the same SD?

The ES is the absolute distance between the two furthest apart shots in the group. So each group only returns one value. Each point represents the ES of a group of n shots simulated with the SD given on the X axis.

 

[billgatese30]

The ES is the absolute distance between the two furthest apart shots in the group. So each group only returns one value. Each point represents the ES of a group of n shots simulated with the SD given on the X axis.

Ok, got it. Basically this is a good representation of how increased sample size gives a more predictable result and as I said earlier, the larger you get its a law of diminishing returns which is why the difference between 3 & 5 is greater than between 5 & 10.

 

[Stalker62]

The reason I ask is that when I run simulations, for this to be satisfied, the x and y error need to be very small (having an SD of 0.1). I don't think very many rifles would satisfy this. The CEP for a set up that satisfies this would need to be 0.1"

I have been approached to do some statistical analysis of load development data. He wants to know if he has a 'true' 1MOA rifle.

I asked him what criteria he'd like me to use to define a 1MOA gun. He couldn't say, and it turns out to be quite hard to find any standard definition online.

So: what would people say is a numerical definition of a 1MOA gun? The options I have at the moment:

What do people think?

Which is why a statistically appropriate sample size is required. Then you can determine flyers which can be excluded from the dataset.

And that just measuring at 100yds/m, does not make a 1”/1moa, rifle, because to me, it needs to repeat that accuracy across any distance the rifle will be shot, and we know that whilst we hope it is, there’s no guaranty the accuracy will be linear at 100, 200, 300 etc’.

The reason I ask is that when I run simulations, for this to be satisfied, the x and y error need to be very small (having an SD of 0.1). I don't think very many rifles would satisfy this. The CEP for a set up that satisfies this would need to be 0.1".

Measuring the results accurately is another discussion in itself. Gauge R&R is another topic we could spiral into but I suggest not today

So really I should just go with what I tell my students in basic stats lectures: n=30, alpha=0.05 and call it good.

May I please be excused from this thread?

 

[Lateral]

May I please be excused from this thread?




View attachment 391802

Yes

 

[billgatese30]

May I please be excused from this thread?




View attachment 391802

We're all good at something. I'm better at calculating 1 MOA than I am at shooting it

 

[Mungo]

We're all good at something. I'm better at calculating 1 MOA than I am at shooting it

What is they say? Those that can't, teach...

...and those who can't teach, go into management.

 

[billgatese30]

What is they say? Those that can't, teach...

...and those who can't teach, go into management.

And if you can't do that you go into sales or HR!

 

[tackb]

And if you can't do that you go into sales or HR!

and the pinnacle of uselessness - politics !