Crunching the 380 — A ballistics analysis (Part III)

Not Part II.2, III. OK, on to what I thought would be the subject of visualizations, namely .380 firearms’ effect on rounds’ effectiveness. Still, having stumbled on this non/light/high expander breakout, I continue to find uses for it in this round.

This is a bit of a 3 bears post, so let’s get on to the first wrong bear: FMJ and other non-expanding rounds.

Again ‘non’ was chosen because the median of each type was non-expanding. You can see some of the Hydra Shok (from three testers) and two of the Ultimate Defense (from one test with a Diamondback) were doing pretty well. Tl;dr: If those interest you, try them in your firearm for function and performance. And a side note: debuting this spring, Federal is introducing their “Hydra Shok Deep” in .380. Looking forward to seeing people test those.

Other than that, I’d discount this whole class except to say, again, they certainly get the penetration part done, and then some, and so if that’s your angle, just buy plain old FMJ. That’s not my recommendation, but there it is.

OK, now on to the other bad bear, the ‘high expander’ rounds.

You can clearly see the variations on the theme here, but really as a whole you can expect a good sized hole, and without putting too fine a point on it, you’re kinda hoping it doesn’t stop short. It’s kind of unnerving that the Saber and TAC-XPD are so inconsistent between testers. The PDX1 and Train and Defend have the virtue of being very consistent (the Guardian Gold only has one test of 5 rounds from a G42). It’s not worth putting more graphs up just to tell you that when mapping these by barrel length they don’t really vary. Not really a ‘bad’ bear after all! The only caveat is that we chose the FBI spec as our benchmark, and they’re not really there.

OK, so did I write all this just to agree with the oft-cited Shooting the Bull 410 videos? Let’s see.

Again, the .380 is not a duty round. None of these drop in at perfect 12″-18″ penetration with full expansion, but they’re not designed to (or they are, in their 9mm configurations, and .380 appeared a bit of an afterthought?) FBI spec emphasizes penetration first and expansion second, and these have that same priority. There are a lot of them though, so this is where the firearm plays into things. I’m going to pick three to make some points.

Let’s start with the almost incumbent winner, the XTP (at least you’ll get that answer and a citation if you ask on almost any forum about .380 ammo).

In the interest of space, I posted these smaller, but in short: XTP seems to work the same across barrel lengths, and penetrates more (and expands less) through cloth. Some inconsistency (see the left side) with the pocket pistols, so check with yours…

Now for the favorite of nearly all 9mm respondents, the HST (in 99gr ‘micro’ for .380, though)

This graphic uses the subset of our expansion/penetration matrix but only for the HST rounds. Why? Well, there’s only 14 rounds in my sample set, and two of those only have FPS recorded. It’s not possible to make an accurate assessment of how HST performs with short and long barrels with this little data.
Worse yet, five of the 12 dots on this chart are all non-expanders. They also happen to be from one source (Lucky Gunner, using a Glock 42). I have no explanation about why they had such odd results. The remainder, though … 3 in the perfect zone, 3 just shy of perfect penetration, and one with plenty of penetration and light expansion. So promising! So few samples! Someone, please, stop shooting oranges and ribs and put some of this HST into gel, with a caliper and a ruler.

Now for the round that provoked so many sad faces across YouTube videos: Hornady’s Critical Defense. This one has a polymer plug in the nose to both inhibit clogging and provoke ‘controlled expansion’ (and yes, that’s squarely in what I’d call ‘light’ and they mapped that way). Did it work?

This doesn’t look bad! Only a few in the goldilocks zone, but at least half beyond the 12″ mark. Clearly, penetration correlates opposite to penetration, though. And yet … on YouTube, all we see are (FBI-spec) failures. What’s going on? Two more pictures explain.

Lots of lines everywhere. The first one shows that HCD improves in penetration as barrel length increases; there’s a clear delineation between the .380 pocket pistol lengths (your LCPs, Picos, TCPs) and the ones which are like subcompact 9s (G42, Walther, Bersa Thunder). I didn’t grab a FPS trend because it correlates exactly the same. The HCD is a light shooter and isn’t making its ideal velocity out of sub-3″ barrels.
The second demystifies the expansion. The gel-only shots vary widely on expansion, which increases a lot by barrel length; with denim barrier, it’s very consistent and mild. When the polymer plug gets to do its job, it works.

Conclusion on the Hornady Critical Defense: I don’t recommend it for pocket pistols; should be great in subcompact size guns. I should temper that a bit. It’s actually OK; but it comes at a premium price, and for pocket pistols the XTP is better for less.

Frankly, while one could tinker with each ammo choice, and maybe you the reader would like another dozen graphs. Not today. I have one more stat, and that’s enough to make a solid recommendation. Based on one site that claims to have figures for the FBI scoring at 70% penetration (in the ideal range, with almost nothing for < 12″ and sharp drops for >18″), 20% expansion, and 10% consistency, I composed a faux FBI score for each round. There’s no ‘consistency’ score because I was scoring each round so that’s a sample size of 1 and no standard deviation… so really it’s 7/9ths and 2/9ths.

Anyway, I was prepared to throw all this into a decision tree model and see what predicts closest adherence to the FBI spec. Remember, penetration weighted overwhelmingly here. Fortunately, my visualizer now has a rudimentary modeler built in, and it quickly came up with an answer. By a good margin, the XTP rounds win (6.15 average/9, vs all rounds being an average 3.15). Generic JHPs, Golden Saber, SIG V-Crown, HST, and Critical Defense made the 2nd tier. Generic FMJ and the Civil Defense round were worst, for opposite reasons (FMJ goes straight through, CD makes hardly any impression).

So there’s the recommendation:

  • If you have a pocket .380 (anything with a barrel < 3″), begin and end your search with the XTP bullet. They’re relatively inexpensive, are loaded by several manufacturers, penetrate in the right range, and expand mildly. (Then practice!)
  • For the sub-compact .380s, there’s no clear winner, but XTP, HST, and HCD rock. I am personally a fan of the HST here, but that’s in my firearm knowing it’s 100% reliable and pushes that 99gr bullet well. (Then practice!)
  • If you live in Florida or Hawaii maybe consider some of those rounds I tagged as high expanders. Grudgingly, I have to accept there’s something to that 12″ standard, and a big wide hole that stops the threat a couple minutes after he kills you is not too helpful. (But practice!)

And lastly; yes, .380 ACP works. If you’re an officer and need to stop threats behind steel car doors, wood, and auto glass as part of your job, no, but for the general public, yep.

I’d planned a part IV with deeper analytics but, unless I find a reason, that’s off for now. Might be fun to develop a predictor for the missing data on some rounds, but the existing data drove conclusions that are compelling to me, and the time is better spent developing better shot placement…

Crunching the 380 – A ballistics analysis (Part II)

Thanks, readers, for commentary on the intro (or, as one put it, lots of words to say nothing). Hopefully not nothing:

– This is about .380 not 9 or 40 or 45 because 380 is the sort-of-marginal caliber with respect to the FBI spec
– .380 is not a ‘duty’ caliber but we’re going to use the FBI spec as a useful benchmark, but only the bare/denim tests, not the barrier tests.

OK, so a bit about the data. Lots of sources from ammunition sellers to YouTube bloggers have done .380 tests, and of those I have only selected those that provided FPS and, ideally, penetration and expansion into organic or non-organic ballistic gel. That’s still over 330 data points of:
– Brand/Model/Weight(gr) of ammunition
– Tester, URL, and test date
– Firearm and barrel length
– Test condition (bare gel or 4 layers+gel)
– FPS, penetration, and expansion
Here I have to add a note. Many of the testers listed each round’s FPS/penetration/expansion, which is ideal for drawing analyses. Many sorted their 5-round groups by FPS, penetration, and expansion independently, so that while it’s good data we don’t know how they correlate. In this segment we’re going to see visualizations that don’t rely on making that connection; in the next one some linear analysis will attempt to fill in the correlation for us.
– From the grains and FPS I’ve calculated a power factor ((grains*FPS)/1000) partly to see if it’s possible to bring a .380 to “BUG” IDPA competitions.

A quick look at all of the rounds together for penetration and expansion clearly shows there are very different designs going on here:

I’ve put ideal lines in at 12″ and 18″ penetration and .531″ expanded size (50%, although I could’ve put it at .525 as the bullets start at .35″) to show what might be termed the ‘goldilocks zone’ for rounds; enough but not too much penetration, and full expansion. Not a lot in that zone between green and red and right of the black line!

What this led me to do immediately is classify the types or models of ammo into:
– Non-expanders (FMJ and things that act like that);
– Light expanders (ones that mostly don’t get bigger than .531″); and
– High expanders (the ones that do)
This required that I go back and slice up the data by model.

Just for those keeping score, the .35 and .36 ones are going under “non”; from there to the Copper-Only Solid are “light”; and beyond that “high.” An anomaly here; the Federal Hydra-Shok went under “non” as their median (out of 21 samples) was .36″ although the average was .42″; so what you’re seeing here are lots of rounds that didn’t expand, and a few that mushroomed out to .5 or better.
Also this chart just listed the model, but there are a few makers of rounds that are just called “JHP” so I labeled them appropriately without redoing the above chart. Same thing with several makers using the XTP bullet; but all of them came out in the “light” category.
Lastly, doing this reminded me of how many tests I’d collected that didn’t list expansion data. Those will have to wait until I test out a model to fill in the missing data from the FPS and penetration data; they’re blank for now.

OK, without even doing any math, or caring which firearm one is using, I can already draw some hypotheses:
– The .380 is still not a duty round! (But I just found a page that claims to have the rating scale for the FBI test, and it’s 70% penetration, 20% expansion, and 10% consistency; if I can verify that I’ll have some calculations with it)
– Like any other caliber, if you want to be sure of making a hole through something, use FMJ. There are a few ‘defense’ rounds which I will list at the end and act just like FMJ, so just buy FMJ. Just note you will probably make a hole through whatever’s behind your intended target too.
– The brands that expand massively almost never penetrate massively. Some get kinda close so when we get into firearms and barrels, some may pan out; we’ll see.
– The brands that are designed to expand lightly are most of the dots in the 12″ to 18″ range with some expansion, and nearly all the goldilocks rounds are those too. Again, much more clarity coming with firearms and barrels.

This went a different direction than I’d intended when starting to write. I’d planned on diving straight into barrel length and ended up adding features to the study and writing all about that instead. Part II.2 will proceed on course…

Crunching the 380 – a ballistics analysis (Part I)

It’s said that racing began with the production of the second car. Long before that, though, and in a much more serious way, people have debated the best firearm cartridge from the moment there were more than one.

Modern semi-automatic handguns come in fairly standardized calibers these days, ranging in size from the .22 LR through .25, .32, .380 (aka 9x17mm), 9mm (aka 9x19mm), .40, 10mm, and .45, . That’s a lot of choices, which means a lot of arguments! That’s excluding some less common variants, and all the revolver calibers.

Of these, the ones commonly cited as “duty” arms are typically cited as 9mm, .40, and .45. By “duty” one means, or ought to mean, firearms intended for use by law enforcement officers and other public defenders whose job may entail stopping and apprehending persons who are committing or have committed crimes. To this end the FBI has developed a protocol to, as best as presently possible, objectively evaluate a combination of a firearm and cartridge to stop an offender.

The FBI protocol

The FBI uses a uniform substance emulating flesh — 10% ballistic gel. While recognizing that people have bones, sinew, muscle, fat, nerves, and organs, this substance provides uniformity in testing. So: misses are useless, and hits that fail to be directed at a vital area may fail to stop an offender, but that’s the role of training, and less so the role of the arm itself. The FBI test is intended to provide an officer with an effective weapon with which to employ their training. This test involves:

  • Bare gelatin
  • Heavy clothing on gelatin
  • Light clothing behind a steel barrier
  • Light clothing behind a wallboard barrier
  • Light clothing behind a plywood barrier
  • Light clothing behind an automobile glass barrier

Lastly, this protocol is looking for 12″-18″ of penetration, and expansion of 1.5x the original size, in order for the round to have the best chance of effecting that stop (gel is not flesh, it’s a test).


This is about where we come in. The general public does not have a “duty” need. The officer has a duty to protect the public, and having engaged the offender, needs to stop them and, ideally, apprehend them, even when they’re taking cover or concealment behind steel, drywall, or other barriers. You or I on the street have a different need. Our need is to make the threat stop. If the threat runs away, great. If it takes cover, we can leave; again, great. In fact, some of the barriers above are things we would very much prefer not to shoot through, especially if we’re in a home, business, or office where family, co-workers, or bystanders might be.

Our ideal round is not necessarily the same as the officer’s — but there are many similarities. If we’re forced to shoot in self-defense (presuming our threat doesn’t quit, run away, or hide) we very much need our action to stop the threat — maybe more so, as we may not have an armed partner and government behind us. Accordingly, on ammunition sites and YouTube vloggers alike the barrier tests are usually not replicated, and you’ll find tests largely involving bare gel or denim/cloth/fleece “heavy clothing” (and the occasional bag of oranges). This set of articles uses the 12″-18″ FBI standard, not because the .380 ACP is a duty round, but because some objective standard is better than none. When making your own choice, keep in mind this is about personal defense, not duty.

That said, in the general case there are many calibers, and since the public is more likely to carry concealed, handguns other than full-sized duty arms come into play. This introduction is long enough, so I’ll get right to the meat of it by noting that a) hardly anyone makes truly “pocket” pistols in 9mm, .40, or .45; and between that and the rounds’ power, b) if those calibers are your choice, you need only avoid the extremes to find a viable round. This article isn’t about traditional “duty” calibers.

This series of articles is about the .380 ACP, which depending on your existing choices, friends, or experience might be thought of as the goldilocks round for self-defense, a good backup, or a complete waste of time.


When I first compiled my information and made a side remark to my initial findings, one Internet forum poster asked what conditions I used for my test. A very insightful remark! Depending on where you look, at this YouTuber or that site, you might conclude that Hornady XYZ or Speer ABC is complete junk or awesome. Not surprisingly, perhaps, you may notice that people test with different arms, in different weather, with the equipment they have; their “heavy clothing” may differ, and environment may play a role. Ideally, it’d be nice to assume that Winchester MNO is what it is and if a site gets certain results, you will too. A quick look at two or three sites will quickly cause you to lose that opinion! Clearly, shooting one, two, or five rounds into gel does not entitle us to make a definitive conclusion!

So: a brief word about data science. It isn’t magic, it isn’t “AI,” and it can’t predict everything. All it is, really, is the same averages and other math you might have used in a basic statistics class (or in Excel) but lots and lots of it. Adding lots of data turns “anecdote” into “information” as long as we keep the limitations in mind: we can’t “prove” anything this way but only infer and rule out the “null hypothesis” or contrary position. If I show that the moon is not made of cheese, you can’t say I’ve proved it’s made of wine. (Silly example but you get the point, I hope).

Accordingly, this study doesn’t “trust” or “distrust” Shooting the Bull 410, Lucky Gunner, or any other author. It pulls together 330+ rounds from 12 different authors, 12 different firearms, and 29 different types of ammunition, all in .380 ACP.

In the next installation I’ll describe all this data and share some pretty pictures (or as we say in the data science biz, visualizations); from there we’ll see if we can draw some estimates when authors publish partial data (say, they have the round, a given firearm, and a chronometer, but not ballistics gel). Finally, we’ll see if all this adds up to a way to decide, whether armed with a tiny Beretta Pico or a long-barreled S&W EZ380, which ammo choices are likely to result in the FBI-spec 12″-18″ penetration and 1.5x expansion — maybe just “a” standard, but currently the best benchmark we’ve got.

Refreshed Data – an iRacing Safety Analysis

Welcome (back) everyone! This was a bit of data crunching from 2018 related to e-sports, specifically the online simulation iRacing.

Understandably, when you bring racers in real-time in hyper-realistic simulations of racing from around the world, add cultures and distance (and internet latency) and add plenty of adrenaline, incidents happen. Worse yet, iRacers spend a lot of time preparing for racing (or at least many do) and a ruined race from mishap is highly upsetting, if at least not as painful or expensive as thrashing a $100k+ car.

Nonetheless, theories abound about how to prevent these, and what predicts a wreck. This work brings together 43k individual driver experiences over several seasons of racing and analyzes when crashes happen, on what courses, to which sort of drivers in which positions on the grid. Some conclusions and suggestions are posited. Enjoy!