The statistics take murders into account, you know.

My pleasure.

Various types of crimes are tracked by the FBI which then publishes an awful lot of statistical data. In question for us is expanded homicide table 8- same data is available on a different page that doesn't deep-link well up to 2021, but the result for all is the same- about 10k-12k firearm homicides per year.
Side note- rifles (including 'assault' rifles and other rifles like hunting rifles) are used in about 300-400 homicides/year, never more than the number of people who are punched and kicked to death. Suggests that maybe trying for 'assault weapon bans' is a waste of time that won't have much effect.

But back on track. 10k-12k firearm homicides per year, the vast majority committed with handguns or 'unknown type' guns. A gun might be 'unknown type' if it's not recovered- for example if there's a drive-by shooting and the perpetrators are not caught, you can't say for sure what kind of gun it was because even pistol rounds can be fired from certain rifles.

Measuring defensive gun uses (DGU) is much harder. In the vast majority of incidents (90-95%) the criminal sees the gun and runs away so there's not much to report. That means a great many go unreported, and of those that do get reported, there's no central tracking system the way there is for homicides. That means the only way to get any sort of number is with surveys and statistical analysis, which are of course open to the interpretation and opinions of the statistician crunching the numbers.

Wikipedia has a good page on that subject which I would encourage you to read. But to briefly summarize- anti-gun researcher Hemenway puts it at 55,000-80,000/year, pro-gun researchers Kleck and Gertz put it at 2.1 million/year, pro-gun researchers Cook and Ludwig put it at 4.7 million/year. More direct analysis of the government NCVS survey data put it between 100,000 and 370,000 DGU/year which is the area I think is probably most accurate. However the one thing just about every researcher involved seems to agree on is that the question hasn't been answered reliably and considerable uncertainty exists.

Thus, for the sake of argument, I take the lowest number from that- 55,000 DGU, and compare it with the highest number of say 12k firearm homicides, and I say there are AT LEAST 4.58x more DGUs as there are firearm homicides.

With that in mind, the argument that 'a successful DGU where your own gun isn't used against you is one in a million' becomes statistically impossible.

A lot of the whole 'owning a gun makes you more likely to get shot' bit comes from bad stat analysis and selection bias. Put simply, if you live in an unsafe area, you're more likely to get shot, but you're also more likely to want a gun for self-defense. That makes the connection between gun ownership and getting shot a correlation, not a causation; but many people confuse the two.

Another big misused stat is suicide. You've probably heard a stat like '35,000 people die of gun violence every year'. How does 12k become 35k? Simple answer is that the rest are suicides. But I think it's disingenuous to count suicides as 'gun violence' because the term 'gun violence' sounds like something that will happen to you, not something you do to yourself. There is a small correlation between gun ownership and suicide rate- I believe that's partially due to socioeconomic factors (the guy who lives in a bad neighborhood more likely has no money and thus is more likely to suicide) but it's also causative (happens because of the gun)- a gun will kill you instantly; whereas many other methods take time during which you may change your mind or fail in your suicide attempt. I still don't believe that self-harm is a valid reason to restrict gun ownership though, but I respect that many disagree with that.

Hope that helps! Does it give you what you were looking for?