The issue of gun accuracy (let's forget for a moment that when most people say accuracy, they mean precision) comes up a lot when discussing the game. Obviously, since engagement distances are very short, accuracy is reduced compared to real life, or no one would ever miss. In a previous article, I pointed out how the M-30 gets shafted about half as much as the D-25. Let's use some extremely scientific methods to determine the shaftedness ratio (SR) for some guns in the game: the ratio of in-game dispersion to real life dispersion. A high SR corresponds to a gun that is highly nerfed compared to real life.
The Soviets used a metric called average deviation to determine how precise a gun is. Average deviation is given for three axes. Since the game doesn't tell you the length-wise dispersion, let's focus on the horizontal and vertical deviations. A bunch of them are given here.
Comparing average deviation to maximum deviation is a bit hard, but not impossible. Due to the magic of a normal distribution, you can convert between the two. The current system is allegedly off at 3 sigmas (it will be 2.5 in 0.8.6). In a standard normal distribution, the 50% boundary lies at about 0.67 sigmas. Half of your shots should land within the closest 22% of the circle. This may not be true if the distribution isn't standard (we know it's normal, so please no whining about impulse in the comments), but for the purpose of comparing ratios, it doesn't matter that much.
Now, let's take the figures for the guns in the linked article, and figure out how badly they got shafted! I will use the 1000 meter figure to calculate precision, since they are all 0.1 meters at 100 meters, and that's boring. Since real life deviations aren't circular, but in-game ones are, let's use the larger of the values.
NOTE: the values, in most cases, are given to the nearest decimeter. This means that a small change that puts a gun from, say, 0.5 meters to 0.6 meters, has a sizeable effect on the SR.
Starting at the lowest caliber, we have the DShK 12.7 mm machinegun. Just like in the game, the dispersion isn't that stellar, 76 cm. At 100 meters, that's 7.6 cm. The 100% dispersion is 57 cm, and the 50% dispersion, using the 22% rule above, is 12.54 cm. That's a shaftedness ratio of 1.65.
Cranking up the caliber to 20 mm on the TNSh autocannon, we get 7 cm of dispersion at 100 meters, 100% dispersion in-game of 53 cm, 50% dispersion at 11.66 cm, and a shaftedness ratio of 1.66. Pretty consistent so far.
Getting out of the autocannon region, we advance to the American 37 mm M5 gun. Compared to the TNSh, it's a sniper, at 4 cm at 100 meters. In game, the 100% dispersion at 100 meters is 46 cm, 50% dispersion is 10.12 cm, and SR is 2.5. That's pretty nerfed!
Let's see how the German 3.7 cm gun does. 5 cm at 100 meters, with the same 46 cm of 100% dispersion as the American gun. That's the same 10.12 cm of 50% dispersion, and a SR of just over 2.
Moving up in caliber, the British 2-pounder gun (40 mm). The maximum deviation is 4 cm (although the horizontal deviation is an impressive 2 cm) at 100 meters. In game, it's 36 cm for a 100% radius, or 7.92 cm for a 50% radius, for a SR of 1.98.
Next up is the Soviet 45 mm model 1937 gun. Artillery tables give us a radius of 6 cm, either 46 cm in game, which converts to 50% radius of 10.12, and gives an SR of 1.68. Interestingly enough, this is the only gun where APCR scatters less than AP. Firing only gold, the SR goes up to 2.
Sadly, I have no data on 50 mm guns, so let's skip all the way up to the M2 75 mm American gun. The impressive deviation of 3 cm at 100 meters is increased to a not so impressive value of 47 cm at 100 meters, or 10.34 cm for a 50% dispersion, matching other similar tier guns. Since the real life precision was much higher, the M3 Lee scores an SR of 3.44. Yikes.
The T-34's 76 mm F-34 gun doesn't do much better. At the same 3 cm, both the Soviet and Chinese versions carry a 46 cm dispersion. At 10.12 cm for 50%, that's a SR of 3.37. Not as bad as the Lee, I guess, but just barely.
Skipping up to 88 mm, the Tiger II's gun gets some impressive results: 2.6 cm at 100 meters. The in-game stat is also impressive, 34 cm. The 50% dispersion is 7.48 cm, for an SR of 2.87.
Moving up a caliber, we get into artillery territory. The 105 mm German 1918 howitzer gets a dispersion of 4 cm (same as the 2-pounder, with an impressive 2 cm on the smaller axis), or 49 cm in game. That's a 10.78 cm 50% dispersion, and an SR of 2.7.
The 122 mm M-30 howitzer has a dispersion of 6 cm, and 55 cm on its TD version in-game. The 50% radius is 12.1 cm, giving an SR of 2.
The 122 mm D-25 has long been the subject of debate among those who care about
The biggest gun I have data on is the 152 mm ML-20S. Its real life deviation is an impressive 3.2 cm, but in-game, SerB's cruel nerf bat reduced it to 50 cm, or 11 cm for a 50% radius. The SR is 3.43. Hey, not as bad as the IS!
So, what do we see here? The low tier, low accuracy guns aren't penalized that much, or else you wouldn't be able to hit anything. The penalty at tier 1-4 is about 2. Tier 5s have a harsher fate, with the SRs going over 3.
Here is a sorted list, in summary.
Soviet 12.7 mm DShK: 1.65
Soviet 20 mm TNSh: 1.66
Soviet 45 mm: 1.68
British 40 mm 2-pounder: 1.98
Soviet 122 mm M-30: 2
Soviet 45 mm 20K (APCR): 2
German 3.7 cm: 2
American 37 mm: 2.5
German 105 mm 1918: 2.7
German 88L/71: 2.87
Soviet 76 mm F-34: 3.37
Soviet 152 mm ML-20: 3.43
American 75 mm M2: 3.44
Soviet 122 mm D-25T: 4.2
Yes, the bottom 3 guns are Soviet, but so is the top one. That's explained by most of my data being Soviet. If anyone has tables for Western WWII guns, I'd be grateful. Despite the cries of anti-German bias, Germans are clustered solidly in the middle.