Garmin ES/SD calculations vs Excel

[RabbitSlayer]

I spent an entire season chasing ES and SD lol. Must have been like 1000 rds from several guns. I have decided the internet is using small samples and totally filled with BS. Doesn't matter how you calculate it, cuz the target doesn't lie....

 

[Straightshooter1]

I spent an entire season chasing ES and SD lol. Must have been like 1000 rds from several guns. I have decided the internet is using small samples and totally filled with BS. Doesn't matter how you calculate it, cuz the target doesn't lie....

I look at velocity SD's and ES's separate from what I get on paper, even though there is some relationship. Velocity SD's and ES's simply tells me how well I'm reloading (e.g. having to do with brass prep, powder charges and bullet seating). When I started precision reloading, I too was chasing SD's with a goal of finding what it takes to get consistent single digit SD's. And like so many shooters, typical sample size was 5 shots; more shots to confirm a load. It took a while to find what I needed to do to get there.

The big thing was getting my powder charges consistent to +/- .02 grs. and that took getting an consistent and accurate electronic scale (the FX-120i did it for me after trying other cheaper scales). Along with that, the consistent sizing of the cases where shoulder bump and neck tension is very consistent, which is best done with quality brass. Then, finally, getting consistent seating depths.

Now that I've got a Garmin chrono, I have larger samples do to its ease of use. While I used to do small sample sizes where I regularly get single digit SD's (like 5-8), I find my larger sample sizes going really well. . . like 50 shots of AR-Comp behind 169 SMK's gave me SD of 5.4 and an ES of 23 fps; 30 shots of AR-Comp behind 177 SMK's with an SD of 8.2 and an ES of 28 fps; another 30 shots of that 169 SMK load had a 8.8 SD and 25 pfs ES. With my 6.5 PRC two weeks ago - 30 shots with N565 pushing 142 SMKs gave me an SD of 4.9 and an ES of 18 fps with 5 round groups under .350" (it's not always THAT good ) .

While what's on the target if is the main determination for a good load, knowing SD's and ES's keeps me aware of whether I'm reloading consistently or not.

 

[RabbitSlayer]

I look at velocity SD's and ES's separate from what I get on paper, even though there is some relationship. Velocity SD's and ES's simply tells me how well I'm reloading (e.g. having to do with brass prep, powder charges and bullet seating). When I started precision reloading, I too was chasing SD's with a goal of finding what it takes to get consistent single digit SD's. And like so many shooters, typical sample size was 5 shots; more shots to confirm a load. It took a while to find what I needed to do to get there.

The big thing was getting my powder charges consistent to +/- .02 grs. and that took getting an consistent and accurate electronic scale (the FX-120i did it for me after trying other cheaper scales). Along with that, the consistent sizing of the cases where shoulder bump and neck tension is very consistent, which is best done with quality brass. Then, finally, getting consistent seating depths.

Now that I've got a Garmin chrono, I have larger samples do to its ease of use. While I used to do small sample sizes where I regularly get single digit SD's (like 5-8), I find my larger sample sizes going really well. . . like 50 shots of AR-Comp behind 169 SMK's gave me SD of 5.4 and an ES of 23 fps; 30 shots of AR-Comp behind 177 SMK's with an SD of 8.2 and an ES of 28 fps; another 30 shots of that 169 SMK load had a 8.8 SD and 25 pfs ES. With my 6.5 PRC two weeks ago - 30 shots with N565 pushing 142 SMKs gave me an SD of 4.9 and an ES of 18 fps with 5 round groups under .350" (it's not always THAT good ) .

While what's on the target if is the main determination for a good load, knowing SD's and ES's keeps me aware of whether I'm reloading consistently or not.

Pretty much same thing for me, but I think I am at at a point where the only thing that is going to improve my numbers is better brass. Lapua or alpha on the next go. Better brass cut everything in half going to starline in 223, but Lapua 223 is unobtanium. I have mountains of Hornady 6 creed brass from the match ammo. Im like 2 or 3 barrels out from needing brass lol. I consistently beat hornady match for 20 shots. My ES is around 30 to 40. Hornady match is 50 to 70. Now with the garmin the samples will be huge. I have sent thousands of dollars into the berm with magneto speed haha. I now sample everything and zero with 20 or 25 shots.

 

[Retired shooter]

Hello All

Would one or two of you "RSG" real smart guys, enlighten this older, not computer literate, dude, what program to use to download velocity data from the Magneto V3 . Windows 10 operating system.

Don't know how to take RabbitSlayers comment about the Magneto speed , good or bad, however that's what I am attempting to use.

Thank you.

 

[Linko]

I understand the large sample size importance. As Elon Musk says : "more data more better"

This brings up a question regarding the value of load development chrono data. "experts" (not me) perform load development based on group size and group format (vertical seperation as an example) "ladder test"

If the groups being shot to develop a load are around 5 (don't burn out the barrel), how much value is the chrono data? I am guessing what the groups look like is first and add the chrono data in the analysis.

 

[AccelR8]

My 2 cents (I am by no means a statistician); I have a Magnetospeed (not a Garmin). I noticed the number it spits out for SD was different than the Excel formula I initially chose. So I took the raw shot data and calculated the SD the old fasioned method by hand with pencil and paper. The result was a number that matched the Magnetospeed output for SD. Then trying the various Excel formulas I found it to be the STDEV.P that matched both the MS and by-hand methods.

 

[AccelR8]

Thanks to rijndael for the posted reply.

I learned this stuff many, many years ago before spreadsheets and personal computers. And haven’t used it much since then. When I worked through my uncertainty (in my previous post) I was unable to understand how the .S formula worked. I mean I could read the formula but I didn’t understand how it applied to the data in the cells...anyway the link the page rijdael provided spells it out nicely, even I can understand!

 

[Doom]

I understand the large sample size importance. As Elon Musk says : "more data more better"

This brings up a question regarding the value of load development chrono data. "experts" (not me) perform load development based on group size and group format (vertical seperation as an example) "ladder test"

If the groups being shot to develop a load are around 5 (don't burn out the barrel), how much value is the chrono data? I am guessing what the groups look like is first and add the chrono data in the analysis.

The follow shows what the 95% Confidence Intervals looks like based on 5 or more shots looks like for mean and standard deviation assuming the measured mean is 2900 for and the standard deviation is 10.

The issue with low round count is due to sampling. It is less of a problem with the mean velocity because the sampling error tends to be more normally distributed around the true mean. This is not the case with standard deviation. The probability is that for small samples the SD is biased on the low side of the true standard deviation and this is would also be true of the extreme spread.

 

[Doom]

The Confidence Interval can be interpreted as if you took an additional infinite number of x samples then 95% of the means and standard deviations of those samples would fall within the confidence interval.

 

[Dimner]

There are multiple stddev funcitons in Excel.

My guess is STDEV is actually STDEV.S rather than STDDEV.P, based on this:

STDDEV.S = 1.5
STDDEV.P = 1.3

The differences are explained here: https://exceljet.net/formulas/standard-deviation-calculation

View attachment 1552921

This is what I found as well when I get lazy with my chronograph and just do one big string even though I'm testing multiple powder increments. Find the excel formula that matches what your chrono uses. Or.... Email Garmin and ask for their formula. All that matters is you use the same formula for everything.

 

[Dimner]

Also... Don't forget that the target is far better data than sample size, standard deviation, extreme spread or any of that.

Don't get bogged down in the minutiae.

 

[Webster]

Also... Don't forget that the target is far better data than sample size, standard deviation, extreme spread or any of that.

Don't get bogged down in the minutiae.

Group size is the ultimate number regardless of the distance.

 

[Straightshooter1]

Group size is the ultimate number regardless of the distance.

I used to feel that way, but now I find Mean Radius tells more about a group.

Now that I'm chronoing large series of shots, I've found it interesting actually seeing velocities increasing as ambient temperature rises. Like today, velocity increased ~12 fps as temperature went from 79F° to 88° over 3 hrs. and I could see my SD climbing right along. I was only shooting at 100 yds, so the velocity changes didn't really have any effect on paper.

 

[Linko]

This thread turned into a very interesting discussion.

 

[jxb]

I used to feel that way, but now I find Mean Radius tells more about a group.

Now that I'm chronoing large series of shots, I've found it interesting actually seeing velocities increasing as ambient temperature rises. Like today, velocity increased ~12 fps as temperature went from 79F° to 88° over 3 hrs. and I could see my SD climbing right along. I was only shooting at 100 yds, so the velocity changes didn't really have any effect on paper.

I'm thinking this would make sense because when I run a load in GRT the pressure increases as the temp goes up. I've started doing my calculations based on 85deg to assure I'm not over pressure during the summer here.

 

[Fast14riot]

Any group statistic is near meaningless without a SD attached to it. Mean radius is good, but whats the SD of it? Absolute groups size (aka agg.) is good too, but whats the SD? There are a few LR/BR shooters who record data of every round fired for the life of a barrel. Some interesting trends can be found in that data.

 

[Straightshooter1]

Any group statistic is near meaningless without a SD attached to it. Mean radius is good, but whats the SD of it? Absolute groups size (aka agg.) is good too, but whats the SD? There are a few LR/BR shooters who record data of every round fired for the life of a barrel. Some interesting trends can be found in that data.

I take it you're not talking about velocity SD's???

 

[Fast14riot]

I take it you're not talking about velocity SD's???

Correct, looking at the SD of group size (in whatever form its measured) is whats important.

 

[swd]

Until there is a new class called chronograph racing the best results the chronograph sees and the best tune the rifle has on paper are rarely ever the same..

 

[OMB]

Great explanation. Is there a formula to predict accuracy based on sample size? Is there a formula in Excel to predict accuracy, something like an R squared number. 5 or 10 shots probably gives a poor accuracy for ES or STDEV. I wouldn't get hung up on small number differences comparing loads. The botton line is group size and small ES if your shooting long distance. Short range bench comp they don't worry about getting low ES. You can shoot small groups at 200 yards with an ES of 20.

ES is a sample-based random variable and as such should not be used to evaluate anything. For small samples, ES will correlate highly with the sample SD but as the sample gets larger, SD will get better as an estimate of the the true underlying SD but ES will increase.