Quick reference guide for SD required to be confident you have a single digit SD load.

[Feniks Technologies]

Ok, let’s break some hearts. Here’s a list of what SD you need to have based on your sample size to be fairly confident you have a single digit SD load.

If your sample SD for that string size is higher than this, you *may* still have a single digit SD. But this is the number you need to be as confident as possible (generally speaking).

Flame suit ready.

- 3 shots: 1.5 fps
- 5 shots: 3.2 fps
- 10 shots: 5.1 fps
- 15 shots: 5.9 fps
- 20 shots: 6.4 fps
- 25 shots: 6.7 fps
- 30 shots: 7.0 fps
- 50 shots: 7.5 fps
- 100 shots: 8.1 fps


Basically, if you shoot a sample size over your chrono and the SD is larger than that, you have a chance (% of chance depends how much higher the SD is) that your load won’t be single digit in the long run.

This doesn’t mean that SD’s at or below these numbers won’t be higher. It also doesn’t mean SD’s above this won’t be single digit.

This is just the maximum SD you can have for a given shot string and be as confident as possible that you have a long term single digit SD.

 

[OREGUN]

I shot 1 and my SD was zero. I’m good, right?

 

[OREGUN]

This is actually pretty interesting. I usually get bored after 10-15 shots (a mag or so) over the chrony, max. It’s not uncommon to see SDs in the range of 5-8. It’s good to have a reference to decide if that’s a reasonable value or…suspect. I hardly reload at all anymore but nonetheless, I appreciate the realism applied to statistical analysis of chrony data.

 

[Jefe's Dope]

What does SD of 9 over 35 round sample size and a SD of 7.6 over a 20 round sample size work out to?

 

[Feniks Technologies]

What does SD of 9 over 35 round sample size and a SD of 7.6 over a 20 round sample size work out to?

35/9 = 7.3 - 11.8

20/7= 5.3 - 10.2

It’s just a 95% confidence interval. So, based on your specific sample data, you can be 95% confident your SD is between those two numbers.

 

[Feniks Technologies]

This is actually pretty interesting. I usually get bored after 10-15 shots (a mag or so) over the chrony, max. It’s not uncommon to see SDs in the range of 5-8. It’s good to have a reference to decide if that’s a reasonable value or…suspect. I hardly reload at all anymore but nonetheless, I appreciate the realism applied to statistical analysis of chrony data.

Yea. IMO, it’s fairly unrealistic to attempt to achieve a very specific sd number like say 5fps. Unless you’re willing to shoot a lot of rounds and do a ton of work at the loading bench.

It’s much more realistic to look at a number you don’t want to go over as well as how much time you want to spend on brass prep and such.


This sort of statistical analysis is what revolutionized Poker game theory. In the past players would attempt to put an opposing player on a specific hand and make their decisions.

Then people came along and used a strategy where you put your opponent on a range of hands he is likely to have in this situation and your average odds against that range. Then make a decision based on pot and implied odds to make the correct play.

IMO, this will eventually be the way that ammo evolves to give shooters the best possible chance to make good ammo and predictions.

As well as alleviate a lot of the “my load fell apart” puzzles. When it’s just performing as it should in the long run.

 

[OREGUN]

35/9 = 7.3 - 11.8

20/7= 5.3 - 10.2

It’s just a 95% confidence interval. So, based on your specific same data, you can be 95% confident your SD is between those two numbers.

How are you getting that range. Teach a man to fish, so to speak. Please.

 

[Feniks Technologies]

How are you getting that range. Teach a man to fish, so to speak. Please.

Google “sd confidence interval calculator.”

Several will come up. Then just plug in your info. 95% confidence is typically what most will use. It my not be exactly perfect depending. But it will get you close enough to help make more informed decisions with the data you have collected.

You can do it long hand, but it’s a bit of work.

 

[CoryT]

Ahh, the tyranny of statistics.

 

[Jefe's Dope]

How to calculate SD from ES? Example- ES=35 for a 50 round sample.

 

[Feniks Technologies]

How to calculate SD from ES? Example- ES=35 for a 50 round sample.

Not really going to be a thing.

You can figure out your likely ES from your SD. But not too much the other way around.

ES is only using two shots of your data and unless you have a ton of it, likely won’t be valid.

Many don’t want to face it, but ES is barely better than useless for us. You can use it in certain situations to eliminate things. But not to validate.

And, if you have enough data for have a confident ES, you have enough data to make a confident SD which will give you more info anyway.

The whole “I’ve seen high ES with a low sd but never a low ES and a high sd” (I used to use this logic myself until I was more educated) signifies a misunderstanding of how the data extrapolates.

 

[Feniks Technologies]

But for a sample like 50 rounds, you could divide by 4 and have a semi decent idea.

But, unless you somehow lost the individual data for everything except the fastest and slowest shot, you have 50 rounds of data.

ES uses 4% (two shots)
SD uses 100%

Fairly easy to see why one is better metric.

 

[LV Precision]

This is data but it basically means nothing. Will this change how you shoot? No, unless you're shooting benchrest.

 

[OREGUN]

This is data but it basically means nothing. Will this change how you shoot? No, unless you're shooting benchrest.

Well, single digit SD seems like a worthy goal for PRS shooting if you want to be confident in the consistency of your ammo at 1000 yards. Would you be happy shooting at .5mil squares at 1000 yards with ammo that had an SD of 12 and an ES of 48? You might stay on the plate. Maybe.

 

[RegionRat]

How to calculate SD from ES? Example- ES=35 for a 50 round sample.

I'll agree with the above and add a suggestion.

If the distribution of that 50 round sample forms a very smooth Bell Curve (Normal Guassian Distribution) with the average being centered, you could divide 35/5 ~ SD 7
However the long term answer could be anywhere from divide by 4 for 35/4 ~ SD 8.75 to as low as 35/6 ~ SD 5.83

If the data is well behaved, you get a pretty good 5 to 6 Sigma answer after 30 shots, with very little improvement towards 50 shots.
On the other hand, if the data is still changing and drifting around past 30 shots, the risk of estimating the ES or SD is that it could still be wrong.

 

[RegionRat]

This is data but it basically means nothing. Will this change how you shoot? No, unless you're shooting benchrest.

Well, it turns out that short range BR shooters are probably the last folks to look at velocity stats.

In fact, they tend to throw their charges and not give a care in the world about the speed since they will add or subtract clicks as the day goes on in order to keep themselves in tune.

It is the MR and LR folks who should be paying attention.

Once we get out past 600 yards or so, it won't change how you shoot, but it will start to affect how you score.... YMMV

 

[Meety Peety]

But this is the number you need to be as confident as possible

No one tells me what I need to be confident!

 

[Feniks Technologies]

This is data but it basically means nothing. Will this change how you shoot? No, unless you're shooting benchrest.

You don’t know what you don’t know.

 

[Feniks Technologies]

This is data but it basically means nothing. Will this change how you shoot? No, unless you're shooting benchrest.

Let’s take a specific example.

A 5 shot group with a 5sd.

That’s means you could have up to a 14sd. Which means you can have up to a 56es 95% of the time.

A 600yd kyl rack is a miss on a 1/2 moa target if you end up on the wrong end of that. In todays very close matches, that could very well be a loss.

But, you’ll have guys scratching their heads wonder why their 5sd load was going over or under targets.

Same idea when the targets are larger. You notice for some reason you’re impacting low or high, make and adjustment, and then you miss low or high.



Like most things, you need to have good fundamentals before a lot of this stuff matters.

But, if you don’t understand how this stuff works, you’ll spend time reworking a load or confused, or even trying to diagnose a rifle or shooter problem which was actually just ammo performing as it should. You just misinterpreted the sample data.


I’d suggest spending more time learning how this stuff works, as well as testing it via shooting more. It’ll be much more productive than replying to threads here without the required knowledge to make such blanket statements.

 

[LV Precision]

You don’t know what you don’t know.

Ok Bryan Litz.

 

[Feniks Technologies]

And as always, this is just more information to help anyone wanting to become a more well rounded marksman. Since many make their own ammo.

Feel free to ignore it and keep doing whatever it is you’re doing.

 

[Feniks Technologies]

Ok Bryan Litz.

Responses like this is why guys like Litz and others no longer post on here and other sites. And you have to pay a membership to get their info.

Jacob and Lindy from RO used to post a lot. Cortina used to post a lot on other forums. Many, many top shooters in the country read here but don’t post for the same reason.

None continue to do so because of these responses. Just something to keep in mind.

 

[Wildfire1550]

Good info

 

[LV Precision]

Responses like this is why guys like Litz and others no longer post on here and other sites. And you have to pay a membership to get their info.

Jacob and Lindy from RO used to post a lot. Cortina used to post a lot on other forums. Many, many top shooters in the country read here but don’t post for the same reason.

None continue to do so because of these responses. Just something to keep in mind.

You’re saying you can shoot the difference of 1% improvement in hit probability? Not likely.

 

[Yondering]

View attachment 7881354
You’re saying you can shoot the difference of 1% improvement in hit probability? Not likely.

That's a good chart with some useful info. And it goes hand in hand with what Feniks has said here though, and doesn't contradict any of it. If you think it does, you aren't listening/reading. I think you and most everyone else here also know dam well that he didn't say what you're claiming he said; that's a strawman argument and just your attempt to pick a fight.

Feniks is right about the reason a lot of knowledgeable shooters quit posting here, because of foolishness like this. You're just looking for an argument without even knowing what you're arguing about.

 

[Feniks Technologies]

View attachment 7881354
You’re saying you can shoot the difference of 1% improvement in hit probability? Not likely.

Tell me you don’t understand any of this without telling me you don’t understand any of this.

 

[Feniks Technologies]

For anyone else reading along and not sure……

If the target is big enough, of course it doesn’t matter.

But that’s with anything. You can say SD doesn’t matter at 2500yds if you use a 25ft tall target.


Also, where he lacks understanding in his copy/paste scenario is that it assumes an actual known and completely valid SD.

We can almost never have a known sd without hundreds of rounds fired. And even then it’s not perfect.

Let’s take the normal 5 shot string and compare a 5 and 10sd.

5sd = 3-14sd

10sd = 6-29sd

So, now we aren’t just talking the difference between 5 and 10sd. We are talking the possible difference in 3 and 29sd and anywhere in between.

Or for each one we are talking 3-14 and 6-29.

Even on the better, as you can see by the copy/paste job, it’s no longer 1%. It’s now in the 4% range.

So, if you want to keep that 1% at 1% and you’re shouting 5 shot strings, you’ll need a maximum of 3.2sd to be 95% confident your margin of error is 1% or so.

And, if you shoot a 5 shot string and get a 10sd and think (like it seems out special guest above thinks) that it’s a 1% difference only…..and therefore no need to try to get better…..

You’ll be scratching your head the time when it’s a 20+ sd and now your 1% just turned into 6-10%. Oh and that’s also assuming zero wobble, a perfectly center zero or offset input, no mirage, no light refraction, no ammo temp increase, aerodynamic jump properly accounted for…..etc. otherwise you just added some tenths of mils to you error.

Special thanks to @LV Precision for helping with the visual.

 

[6.5SH]

The need to gather more samples coupled with the desire and real need to limit component use and loss of barrel life kinda makes the case for something like the firearm mounted radar chronograph in another thread.

 

[Jim Boatright]

When your rifle and loads start producing very uniform bullet speeds just out of the muzzle, it makes sense to use Extreme Spread (ES) for a group of shots as your performance metric instead of Standard Deviation (SD). We found in our development of monolithic copper ELR bullets that, by base-drilling our bullets during manufacturing, we could routinely get indicated ES values of 2 to 4 fps for 5-shot test firing groups as measured via Oehler chronographs with centers 10 or 15 feet ahead of the muzzle. We started using my old portable Oehler 35P chronograph with a 4-foot screen spacing and lately use an Oehler System 89 with 10-foot screen spacing centered 15 feet from the muzzle for 338 Lapua Magnum. [I expect the muzzle speed measurement error itself is about 1.5 to 2.0 fps.] We drill a 0.45-caliber hole in the base of the bullets deep enough to pass through the boat-tail and most of the way under the Rear Driving Band (RDB) of our copper ELR bullets. At a cost of 10-grains in bullet weight, this base-pressure porting greatly improves the gas sealing ability of these half-hard (H02) copper bullets by allowing the RDB to expand elastically at peak base-pressure something like an old lead Minie ball. We use very fast twist Schneider P5 test barrels for their inherent gas sealing design. The complete lack of copper fouling is a nice side benefit of this approach.

 

[Feniks Technologies]

When your rifle and loads start producing very uniform bullet speeds just out of the muzzle, it makes sense to use Extreme Spread (ES) for a group of shots as your performance metric instead of Standard Deviation (SD). We found in our development of monolithic copper ELR bullets that, by base-drilling our bullets during manufacturing, we could routinely get indicated ES values of 2 to 4 fps for 5-shot test firing groups as measured via Oehler chronographs with centers 10 or 15 feet ahead of the muzzle. We started using my old portable Oehler 35P chronograph with a 4-foot screen spacing and lately use an Oehler System 89 with 10-foot screen spacing centered 15 feet from the muzzle for 338 Lapua Magnum. [I expect the muzzle speed measurement error itself is about 1.5 to 2.0 fps.] We drill a 0.45-caliber hole in the base of the bullets deep enough to pass through the boat-tail and most of the way under the Rear Driving Band (RDB) of our copper ELR bullets. At a cost of 10-grains in bullet weight, this base-pressure porting greatly improves the gas sealing ability of these half-hard (H02) copper bullets by allowing the RDB to expand elastically at peak base-pressure something like an old lead Minie ball. We use very fast twist Schneider P5 test barrels for their inherent gas sealing design. The complete lack of copper fouling is a nice side benefit of this approach.

Yea, it’s like most things. If you’re *actually* capable of world class ammo, these posts and such aren’t for you as the knowledge base required for such things is already beyond simple statistics.

But, most shooters aren’t actually able to produce real sub 5sd ammo consistently. If nothing else, just the time it takes to properly prep and keep things that uniform/consistent.