Standard deviation versus accuracy?
- Thread starter grobin
- Start date
68% of your shots will fall in a band of two SDs (one on either side) around your mean (average) velocity. 95% will fall in a band of four SDs, and 99.7% will be within 6 SDs. This assumes you have enough MV measurements to calculate a statistically significant SD.
The wider the spread of MVs, the larger the likely vertical spread of your groups. That spread may or may not be significant to you depending on the distance you are shooting and your precision goals. You can calculate the predicted vertical differences by plugging MVs (and your target's distance) into a ballistics calculator.
Since accuracy nodes are at times dependent on "barrel time" (which likely varies as MV does), a high SD may result in some rounds falling outside the node. That may result in additional precision loss.
(Note that SD can also be affected by inconsistency in your actual powder charge weights.)
OTOH, sometimes the combos producing the best SDs do not produce as small a group size as other charge weights. Only your targets can tell you that.
The wider the spread of MVs, the larger the likely vertical spread of your groups. That spread may or may not be significant to you depending on the distance you are shooting and your precision goals. You can calculate the predicted vertical differences by plugging MVs (and your target's distance) into a ballistics calculator.
Since accuracy nodes are at times dependent on "barrel time" (which likely varies as MV does), a high SD may result in some rounds falling outside the node. That may result in additional precision loss.
(Note that SD can also be affected by inconsistency in your actual powder charge weights.)
OTOH, sometimes the combos producing the best SDs do not produce as small a group size as other charge weights. Only your targets can tell you that.
Rick Courtright
Hawkeye
grobin said:
Hi,
In my own experience, not enough of one to trust as a reliable predictor. However, working toward loads with small SDs often helps give a little psychological edge to one's confidence in their loads. Yet even then, as Twoboxer said, sometimes bad is good and vice versa! So ya gotta wear out some paper to find out for sure.
Rick C
Ka6otm
Blackhawk
- Joined
- Dec 21, 2002
- Messages
- 753
I took 3 quarters of Statistical Analysis in college, got As in all 3, and started and ran 2 different Statistical Process Control programs in different companies.
In general, the OP is correct and Standard Deviation is an indicator of quality (or in this case, accuracy).
However, it doesn't seem to work all that well with guns. I have been subscribing to Handloader magazine for around 30 years and every issue has loads they tried and they include data for each one, including Standard Deviation.
What I've learned from reading these hundreds of tests is that sometimes the lowest Standard Deviation is the best grouping load and sometimes it's not. It's essentially random and not a good indicator at all.
In general, the OP is correct and Standard Deviation is an indicator of quality (or in this case, accuracy).
However, it doesn't seem to work all that well with guns. I have been subscribing to Handloader magazine for around 30 years and every issue has loads they tried and they include data for each one, including Standard Deviation.
What I've learned from reading these hundreds of tests is that sometimes the lowest Standard Deviation is the best grouping load and sometimes it's not. It's essentially random and not a good indicator at all.
Jimbo357mag
Hawkeye
Sometimes I think SD is an indicator of how accurate your reloading is and the accuracy is an indicator of how windy it was or how steady you were that day. It's all good.
I've got a type of .22 ammo that gives four distinctly different reports when shooting it--from very subsonic to normal supersonic to a near-.22 mag crack. Velocity is obviously all over the place. And yet the stuff, from several rifles, prints very small groups at 50 yards--much smaller than some other ammo that all sounds the same.
There is NO reliable predictor of accuracy. You have to shoot the gun/ammo combo to see what it can/will do.
There is NO reliable predictor of accuracy. You have to shoot the gun/ammo combo to see what it can/will do.
I just use Extreme Spread (ES) anymore. You will find over many tests, you can expect an standard deviation (SD) to be about 1/3 of the Extreme Spread. The tighter your ES, the more chance your load is going to be accurate -- but of course not always (guns can be 'finicky')! Remember that a bullet traveling at a different velocity is going to impact in a different place on the target just due to ballistics. That's why you would 'like' as close to 0 ES as possible! BTW you always shoot at least 10 rounds for a meaningful SD statistically. I usually shoot 15 per test.
Looking at a snapshot of some loads :
SD and ES
8 31
10 30
12 44
11 31
10 35
27 88
etc.
Looking at a snapshot of some loads :
SD and ES
8 31
10 30
12 44
11 31
10 35
27 88
etc.
Hey, no short cuts, you need trigger time. I agree with what has been said. But, add smaller SD is a possible indicator of better/smaller groups.
This is how you will find out. Shoot for accuracy when getting the information of your ballistics.
This is how you will find out. Shoot for accuracy when getting the information of your ballistics.
Similar threads
