Torpedo Bomber(TacBomber Class = 9) attacking Capital Ship or Destroyer
Until I get the philosophy sorted out, no reasonable model parameters can be set.
...So, it was considered shameful for a hussar to live to be 30 years old...
At Midway, a total of 41 Devastators, the majority of the type still operational, were launched from Hornet, Enterprise and Yorktown to attack the Japanese fleet. The sorties were not well coordinated, in part because Rear Admiral Raymond A. Spruance ordered a strike on the enemy carriers immediately after they were discovered, rather than spending time assembling a well-coordinated attack involving the different types of aircraft – fighters, bombers, torpedo planes – reasoning that attacking the Japanese would prevent a counterstrike against the US carriers. The TBDs from Hornet and Enterprise lost contact with their escort and started their attacks without fighter protection.
The Devastator proved to be a death trap for its crews: slow and hardly maneuverable, with poor armor for the era; its speed on a glide-bombing approach was a mere 200 mph (320 km/h), making it easy prey for fighters and defensive guns alike. The aerial torpedo could not even be released at speeds above 115 mph (185 km/h). Torpedo delivery requires a long, straight-line attack run, making the aircraft vulnerable, and the slow speed of the aircraft made them easy targets for the Mitsubishi A6M Zeros. Only four TBDs made it back to Enterprise, none to Hornet and two to Yorktown, without scoring a torpedo hit.
Source: Wikipedia
https://en.wikipedia.org/wiki/Douglas_TBD_Devastator
Some were killed by fighter planes, some by anti-aircraft fire.
There have been attacks in which out of six torpedo bombers sometimes one survived ...
Swordfish, Fairey Albacore, Nakajima B5N, SM.79, IL-4... There was little chance of the pilots seeing their grandchildren. The generals came up with a combat model in which a slow-moving airplane is long in range of the ship's anti-aircraft artillery fire.
Someone created a system of rewards and incentives for pilots and their relatives after the quite expected death of a pilot - essentially a kamikaze pilot. Way out there with very little chance of getting back in.
The death in the name of bright ideals, in the name of one of four warring empires, in the name of emperor or king, in the name of one of three odious leaders, in the name of one of four political and ideological systems... but it was still almost guaranteed doom.

For the homeland and industry, 20 airplanes were much cheaper, and - alas! - more profitable than the senseless loss of one Hood, on which 1,418 men died.
This is a very heavy and even psychologically difficult topic to express in percentages and model...
What does all this mean in terms of modeling? Trading a cheap chess bishop for a chess queen? Yeah, I guess so.
But, modeling will still have to be done! There are three ways to do this in the model.
1) A strong ship (primarily Battleship) has a much larger INI than a Torpedo Bomber. And is guaranteed to inflict some sort of casualties. The Torpedo Bomber then strikes the ship. The Bomber's NA must be very high for the strike to be felt.
The pros are a controlled result. Torpedo Bomber will never leave the battle without losses.
Cons - predictability of the result.
2) Battleship has equal INI.
Pros - intrigue of the battle. The player cannot be sure that exactly now this Torpedo Bomber will inflict losses in this particular attack.
Cons - other types of ships, weaker ones, get hit by Torpedo Bomber with completely predictable results.
3) Some kind of compromise between the first and second options.
None of the three proposed options are perfect. The problem is that in this specific case of Torpedo Bomber vs Ship, the distribution of random requires an inversed rather than a regular Gaussian. There is a higher probability of attacks on the edges of the Gaussians, and a much lower probability in the middle, where the regular Gaussians are bent "up".
Game developers, it's so easy to do in program code...
Torpedo bomber's parameters for third "compromise" option:
Naval Attack = 16, Ground Defense = 0, Initiative = "-4" (minus four). EXP = 300, STR=13.
That doesn't mean that these are the most correct parameters. Maybe I will go to option 1 or 2 after further experiments. At least I have marked a starting point for experiments on a wide field of modders' activity.