TargetTalk

has anyone found any ammo that is suitable for 100 yard prone this coming season,I do not plan on buying $1000.-1200 ammo that will not group any better than $300 to 600
dege41

After extensive testing, my Anschutz groups the same (no statistically significant difference) with CCI Standard Vel, RWS, and Eley Match Rifle. Huge price differences; no performance difference. (note: the average distance from mpi differences for 30 shot groups were in the 2-3 tenths of an inch range at 100 yards)

You have to test each individual rifle with many different brands and do the math.

Oh By The Way- the performance differences between lots of the same exact ammo can be larger than the performance differences between different brands . . .

Steve's reference about the lot numbers I agree with. May need to test several lots of a brand and then if you find one that shoots buy all you can afford. I shoot both 3p and rimfire benchrest. I have always had the best luck with Lapua Midas. But from now on I think I will stick with Lapua Master or Eley Match (black box) semi auto bullet style. Paying over $1000 a case is getting stiff. In benchrest many are finding the Eley semi auto bullet style drifts less in the wind than the EPS with the nipple. Also deal with someone that will sell you a couple of boxes each of different lots and then will have some of those lots available it your testing is good. Lapua sells out quickly when new lots arrive in the states so yoou need to test quickly as soon as you get it.

Steve Swartz wrote:After extensive testing, my Anschutz groups the same (no statistically significant difference) with CCI Standard Vel, RWS, and Eley Match Rifle.

Steve,
How do you quantify your shot group, and what statistical test do you use to establish significance?

Jeroen Hogema

My expectations for 22 match ammo, has to be able to group 10 shots at 100 yards at less than .800 of an inch.
For years we were able to better this by trying a lots of different numbers.
How ever in the last few years it seems like there is no ammo that will consistantly produce small groups, I know it is not the rifle becaus I have saved back some of the good ammo lots just as a comparison between old and new ammo.
The last good ammo I had bought was Lapua Multi-match, I have tried Red box Eley, Midas Lapua, RWS R 50, Wolfe of all Grades and have found that the Wolfe MT shoots as well as the high priced ammo, but not good enough for 100 yard shooting in smallbore prone matches
I am not new at this game I shoot about 10,000 round of 22 match ammo a year
dege41

Shoot 30 shots, calculate centroid (average x, average y point) then measure 30 distances from centroid. For each 30 shot group (sample), we then compare "average distance from centroid" using the "standard deviation of average distance from center."

From an experimental design perspective, I shoot 6 x 5 shot shot groups so I'm not shooting one ammo all 30 shots at the beginning and another batch of 30 at the end when I'm tired etc. . . .

Steve Swartz

The riflemen in the Pheonix area have been reporting excellent results with Wolf Match Extra and SK Match. They have an extensive thread about it on Shooters Journal.

The best name brand ammo has become so expensive that some have taken to buying the new model Anschutz and three or four custom barrels, the barrel cost of which is equal to a case of Eley Red. It only takes a few minutes to swap out a barrel so the next step is to purchase ammo and see which barrel gives the best results. In the long run they believe that this will be cost effective in selecting "knot lots" because it gives a wider selection of ammunition and extends barrel life.

I never thought I live to see the day when it was quicker and cheaper to change a barrel than select and buy a lot of ammunition.

Hap Rocketto

Steve Swartz wrote:Shoot 30 shots, calculate centroid (average x, average y point) then measure 30 distances from centroid. For each 30 shot group (sample), we then compare "average distance from centroid" using the "standard deviation of average distance from center."

So for one group, you get one average and one standard deviation. Do you get into formal statisticsal tests with that= E.g. a t-test to compare 2 groups?

Steve Swartz wrote:From an experimental design perspective, I shoot 6 x 5 shot shot groups so I'm not shooting one ammo all 30 shots at the beginning and another batch of 30 at the end when I'm tired etc. . . .

Do you analyse the 6x5 shots as 6 separate groups, or do you treat them as one 30 shot group?

Jeroen Hogema

Every shot is a data point- providing distance from the centroid. Shoot 30 shots on one target.

Yes, statistical tests (o.k., I cheat and use z scores on 30 shots instead of t scores; less rigorous but still quite useful when looking for subtle differences).

I don't give a rip about "how big a cloud of X shots make." I care about "how far will shots fall from each other."

Standard "NRA Style" group testing is somewhat misleading in this respect. And why on earth do we want to thorw out 4/5 datapoints in the first place? I want *every* shot to be meaningful.

(consider two identically sized [by ES] groups; one with a four shot knothole and one flyer vs. a circle with all shots around the edge. NRA "group testing" would say "these groups are identical." WTF, O?)

Steve Swartz

I know a guy who sent his Anschutz to Eley to test their ammo on it, the Tenex performed slightly better that the Club stuff at 1/ the price, moral of the story; its not always the expensive stuff that does the trick!!


Steve

WA

Steve Swartz wrote:Every shot is a data point- providing distance from the centroid. Shoot 30 shots on one target.

Yes, statistical tests (o.k., I cheat and use z scores on 30 shots instead of t scores; less rigorous but still quite useful when looking for subtle differences).

Steve,

Indeed, when you have the possibility to use the data from each individual shot, you'd better do so. See e.g. Leslie (1994). When you are trying to pick the best from 2 sets of ammo, you can use his results to make statements like
- when using 30 shots per group,
- and when using the mean radius (MR) to select the best one,
- and when the 'real' difference between the 2 sets is 40%,
- then you have about 97% probability of picking the right one.

He also demonstrated that mean radius is a better metric than shot group radius. (Still, for my air rifle pellet testing this won't work - even pellets that are no good for my rifle will mostly be creating 1 big hole. Not possible to measure x,y coordinates from paper)

He used various metrics to select the best group, no matter how small the difference. You are using a statistical test to make a statement about the significance of a difference. So where Leslie would say "this one has a lower MR, so It's probably the better one", you might say "the difference is so small that I can't be sure which one is better".

Thus introducing statistical power: what probability do we have of finding a significant effect (given an effect size that we want to be able to detect).
I ran some simulations to get some feeling for this. I tried the t-test which is more appropriate than the z statistic because we don't know the actual standard deviation. (I know I am violating at least 2 t-test assumptions, so I am cheating too, but hey - you've got to start somewhere...)

- when using 30 shots per group,
- and when using the mean radius (MR) as the metric,
- and when the 'real' difference between the 2 sets is 40%,
- and when using a p-level of 0.05 in a t-test,
- then you have about 65% probability of obtaining a significant effect.

This 65% is not very impressive, I'd say... And that is for a 40% effect size, which is not my idea of subtle differences.
So if you want to be able to a statement about statistical significance of differences between 2 groups,
- you either need even more shots than 30,
- or you will only be able to obtain significant effects when the differences between the 2 lots are huge.

Perhaps the approach of picking the lot with the smallest MR and forget about statistical significance is more practical...

Jeroen Hogema


Leslie, J.E.III. (1994). Is "Group Size" the Best Measure of Accuracy? (used to be at http://www.floridahighpower.com/Features/WHICHONE.pdf).
Published as "Is 'Group Size' the Best Measure of Accuracy?", The Canadian Marksman, 129, no. 1 (Autumn 1994): p46-8.

I found that the radial standard deviation (square root of the sum of horizontal and vertical variance) was slightly better than the mean radius, but it was much more intense to calculate. I also found that the information content of additional shots diminished, and that by the time you have 25 data points (shots) each shot adds very little to the probability of determining the smaller lot. Of course, we all agree that the typical gun mag test of extreme spread on a five shot group is useless!

The link to my two articles is:

http://www.shootersjournal.com/Features.htm

The link worked this morning.

Jack

Well, O.K., fine- but the method predates the 1994 article you refer to by at least 50 years . . . I'm using the same methods that have been in general use for serious firearm testing (military, primarily) for decades.

CEP as a measure of "accuracy" (precision, actually) goes back to Army ordnance evaluation circa 1930s.

Haven't read the Leslie article but I'm sure it's fine for the audience.

Steve

Steve:

No argument from me: I got the formulas from a 1965 book by a gentlemen named Frank Grubbs, a mathematician at Aberdeen Proving Grounds. I even found a reference at using RSD in testing the accuracy of the M1917 rifle.

I make no claim to discovering these stats. I just used them and tried to ween people off of the extreme spread stat that is so common.

Jack

I know this is very old topic, but searching on "selecting ammunition" I have found it and tried to read above (3-4 posts) mentioned articles. Unfortunately, links do not work anymore. Do you know if it is possible to find them elsewhere? If you know any other links relative to the subject would be great to know it too.

BTW
For selecting ammo in detail, I use similar mathods as described by Steve Swartz. Unfortunately, in spite of the fact that ESTs become more and more present, still many of my colleagues need to do that using paper targets shooting 10shot groups. Then I advise to base the analysis on surface measurements that is quite simple now (using for example Photoshop Elements soft). It works quite well...

Great no BS thread!

Rover wrote:Great no BS thread!

Yeah, so as I understand these articles are not worth reading... ? Usually I prefer to verify it myself, but ok...

First hit on google: http://fuzzylimey.net/coachtalk/groupanal.html

I have been told in the past, that you should clean the barrel before testing a new lot, true or false.

EJ wrote:First hit on google: http://fuzzylimey.net/coachtalk/groupanal.html

That I have found, of course, via googling already; I have meant rather these two articles by Jack Leslie (shootersjournal) as no titles were mentioned... Anyway, thank you very much.