The most important factor for CAP numbers is the CAP amount setting set via the Routine air ops window. Remember that this setting is set for each force separately (main/scout/carrier)!

The amount of fighters dedicated to CAP as percent of total available fighters (this includes repairable aircraft as discussed later):

• Light: 33%
  
• Medium: 50%
  
• Heavy: 66%
  
• Maximum: 100%

These are very high confidence numbers as they were strongly repeatable in empirical testing (N = 32). Based on testing each carrier division will handle its CAP independently. E.g. with medium setting each division will seek to have half of its fighters on CAP. By contrast within the division each carrier may not contribute equally. In testing (with two ship divisions) one carrier could handle anywhere from 15-85% of the whole division’s CAP, with the other carrier filling the rest. I was unable to determine any pattern as to how this inter-division workload was determined.

The AI seems to allocate CAP during night time the same way as it would during day. The catch being that it assumes every fighter at the division’s disposal would be night capable, even when they are not. As an example lets imagine a CV division with two carries, each with 40 fighters. But only 10 fighters on each CV is night capable. With medium CAP setting the whole division seeks to maintain 40 fighters on CAP at all times (the division’s total available fighter pool being 80). But since there are only 20 night capable planes total these poor pilots will have stay airborne the entire night without rotation. The AI interprets that it needs 40 fighters on CAP during night and just accepts that there is a shortage compared to what it thinks should be airborne. The practical problem for the player is that the AI does not “spare” any night capable pilots from CAP operations during day time. So if your entire fighter complement is not night-capable expect the pilots that are to suffer fatigue faster, as they shoulder disproportional amount of flight ops during the night.

When a CV division seeks to fulfill the requirements as set out by the CAP settings for the force it counts damaged aircraft into the pool of total available.

Also while not the topic of this guide I did pay attention to reliability numbers during the testing and noticed the following: The amount of aircraft listed as damaged at the battle’s start does not seem to be affected by aircraft reliability. Even with good reliability rating I could have anywhere from 5-25% of my planes damaged at battle start. However good reliability did seem to affect the rate of repairs, i.e. planes would be made flyable within one in-game hour. Accidents seemed to be more affected by technology than aircraft reliability.


When a CV division is set to support another division in the order of battle it will allocate 1/3 of its CAP fighters to cover the parent division with the remaining 2/3 staying over the CV division. This is a high confidence assessment based on testing. The total amount of CAP fighters is determined by the CAP setting as discussed above. See Data Set 2 below for example distribution when supporting another division.


This setting is selected from the division control window. However despite testing I was unable to determine any pattern as to what the effects are supposed to be. In fact I’m unwilling to even give any averages as that would imply some sort of consistency. The only strongly observable trend was that Provide CAP to -setting is very ineffective. In many cases no fighters at all were ever deployed over the target division. And when they were there was no determinable factor as to how many fighters would be allocated to this task.

The intended effects of this order remains a total mystery to me and I would welcome any input on this matter.

Below are several data sets to illustrate in concrete numbers the findings discussed above. The values presented are not averages, but actual figures from random tests.

All data sets share the following factors:

• One CV division of two ships.
  
• Each CV has 44 fighters, so total fighter amount for the division is 88.
  
• Fighters have good reliability and values are taken once initially damaged airplanes have been repaired.
  
• No fighters are used for automatic recon missions.
  
• All percentiles are rounded to whole figures.

In this data set the carrier division is independent and not providing CAP to any other division. The data shows the effect of different CAP settings on fighter amounts.

In this data set the carrier division is part of the main force and is supporting a BB division. The numbers illustrate how CAP setting doesn't affect the proportional amount of planes, only absolute numbers. Notice the variation from the average discussed above. Such variation can always be expected during actual combat operations.

In this data set the carrier division is in the carrier force, while the BB division is in the main force. CV division's role is set to independent and it is ordered to 'provide CAP to' the BB division. As the numbers show the reliability of this system is very low. Similar results occurred in other tests and I'm unable draw any conclusions.

In this data set the carrier division is in the main force with the BB division. CV division's role is set to independent and it is ordered to 'provide CAP to' the BB division. What is clear that the 'provide CAP to' has some effect, unlike in previous example when the CV division was in its separate force.

In this case study the carrier division is in the main force with the BB division. CV division's role is set to support the BB division and it is ordered to 'provide CAP to' the BB division. In general I would say based on the testing the overall amount of CAP is not effected, but the distribution between the divisions can swing in unpredictable ways (I'm unwilling to give any averages as I would have low confidence in them). Personally I don't see any reason to use this order combination over just having the CV division in support role, as using just the support role is very reliable and predictable numbers wise.

These final example numbers are from tests where the CV division is in support role to a BB division, while it is ordered to 'provide CAP' to a different CA division. The aim was to see how the AI splits the CAP over three different divisions. In general the CAP was concentrated over the CV division. Quite often no CAP was provided over the CA division (as these four examples show). In general I would not recommend trying to use this arrangement in combat (if you value your CA division). As mentioned several times at this point I would only ever trust CV division in support role to prove reliable and consistent CAP numbers over another division.