Question on navy search

[Kull]

Personally I use the Search Arcs, but I can understand why others might not. Is 360 degrees "with penalties" better than a focused search? Maybe, particularly from ships or islands. Probably not when searching sea zones from a continent. I suspect that some of the "noise" when trying to run a test involves the reality that "arc" and "no arc" searches both get the 360 bonus from the first 4 hexes, factored by the decreasing likelihood of spotting things the further out you go.

[Arkham]

OK, so I've been running some tests on naval seach. The TL;DR is that not selecting arcs is better than selecting arcs.

Methodology:
I am running .26B with the updated AI, (tho thats irrelevant for this test). I ran the 1000 mile campaign scenario because I didn't want coastwatchers skewing the results. I was playing head to head mode, and as the IJN i created a SCTF of two CLs and two DDs and parked it 9 hexes away from Dutch Harber to the NW (164,46).

As allies, I stood down all my aircraft until the TF was in place, then set Cat Squadron VP-43 to 100 percent search. First I set it to search arcs 260-320, ensuring that each 10 degree radius would be allocated two planes. Search altitude was set to 6000 and the max distance was 12. This was to replicate searching for the enemy on a likely avenue of advance.

I ran ten turns at this setting and recorded the results in excel and included the number of operational losses and pilot fatigue at the end of 10 days. I defined a hit as a notification in the IJN ops report. The allies ops report was useless, with one drunk Cat pilot reporting seeing 12 enemy ships right outside Dutch Harbor. He was quickly sent stateside.

I then reloaded the game and reran the test. Only parameter I changed was changing the search arc to 000 - 000, or no arc. Same number of planes going out, same altitude and same search distance.

I then ran the test a third time, this time changing the percentage of search from 100 to 80 percent, or 9 planes doing a search. (12 * .8 = 9.6, but I'm assuming it rounds down to 9). {edit} I confirmed it does round down to 9 planes by checking the mission count after 1 day). I did this as this is going to be a more likely setting to help reduce fatigue and operational losses.

These are the results.


Based on that, it seems that setting search patterns is a waste of time and effort and shouldn't be done. While there is a big malus to search based on not setting a pattern, the fact that you have more planes rolling against each hex seems to more than make up for this. I'm not sure this is what the devs intended.

[Kull]

Good test methodology. It seems that whatever the "penalties" are coded to be, they aren't enough.

[Arkham]

Good test methodology. It seems that whatever the "penalties" are coded to be, they aren't enough.

Yeah, I agree. Maybe I should run this test on the beta patch .27 and let them know in that thread if the results are the same? I would think a hit result of an unfocused search of around .3-.5 would be much more realistic and fair. You want the player to have to think about avenues of approach and set search arcs appropriately. Right now it seems that the game is rewarding you for not thinking about that and being lazy.

[RangerJoe]

Good test methodology. It seems that whatever the "penalties" are coded to be, they aren't enough.

Yeah, I agree. Maybe I should run this test on the beta patch .27 and let them know in that thread if the results are the same? I would think a hit result of an unfocused search of around .3-.5 would be much more realistic and fair. You want the player to have to think about avenues of approach and set search arcs appropriately. Right now it seems that the game is rewarding you for not thinking about that and being lazy.

The update was supposed to be very limited but people pointed out more things that could be fixed. Then the beta was probably upgraded at least one more time and tested after each upgrade. I don't think that naval search had any problems or potential problems to fix. If you want to test the beta that you have, just change your current .exe to a .bak and put in the new .exe and see if it works.

[Arkham]

I just tested .272 (2022 release) and I got the same results. Focused search arcs gave worse results than not setting any arcs. I'll post in that thread too letting them know in case they want to adjust the malus value.

[Kull]

I just tested .272 (2022 release) and I got the same results. Focused search arcs gave worse results than not setting any arcs. I'll post in that thread too letting them know in case they want to adjust the malus value.

I agree. Visible Search arcs were one of the "cool new things" added for AE. Eye candy aside, a focused search area should always provide better results then a 360 search with the same assets. The penalty assessment is clearly broken.

[Arkham]

I just tested .272 (2022 release) and I got the same results. Focused search arcs gave worse results than not setting any arcs. I'll post in that thread too letting them know in case they want to adjust the malus value.

I agree. Visible Search arcs were one of the "cool new things" added for AE. Eye candy aside, a focused search area should always provide better results then a 360 search with the same assets. The penalty assessment is clearly broken.

It gets worse. I continued to test using .26b in 4 day increments after I realized I could set all my message delays to 0 and days per turn to 4. I did about a month's worth of testing in four day increments, and found that f I set the search value to 20 and the rest value to 80, I was achieving slightly below the focused searches at 100 percent in terms of hits. The sweet spot is probably about 40 percent search to get the same results as a focused search, but you can do it indefinitely and not ever have to worry about search ever again. Set and forget.

Feel free to test this out yourself and let me know if you see any problems with my testing methodology. Mind you, this is with a squad of 12 with NavS in the low 60s, so earlier war pilots are not going to be as good, but if you set your 1941/1942 patrol squadrons to about 60 percent you'll be training them up nicely.

[Kull]

I've always run Naval and ASW Search at 50%, and don't use arcs for any of the TFs and many of the islands. So my results have been pretty good, with very few attrition losses. You still want to train those pilots in the necessary skills, however. For example, the Japanese AI is pretty good at spotting my submarines but bomb hits are very rare. By late '42 most of my pilots have been trained up to 70 ASW and 70 Naval, so they not only spot, they're getting hits.

[stuman]

Well that will certainly save some time.

[Chris21wen]

I've always run Naval and ASW Search at 50%, and don't use arcs for any of the TFs and many of the islands. So my results have been pretty good, with very few attrition losses. You still want to train those pilots in the necessary skills, however. For example, the Japanese AI is pretty good at spotting my submarines but bomb hits are very rare. By late '42 most of my pilots have been trained up to 70 ASW and 70 Naval, so they not only spot, they're getting hits.

Basically the same. Only use arcs for land masses but if what you say is correct it seems pointless. Definitely a fix required if that is the case.

As far as automatic detction goes I thought it happened at one hex range.

Dectction between 1-4/5 hexes inproves considerably as more potential aircraft search them, up to 6 in the immediate adjacent hexes, gradually reducing as the range increases.

[Kull]

As far as automatic detction goes I thought it happened at one hex range.

Dectction between 1-4/5 hexes inproves considerably as more potential aircraft search them, up to 6 in the immediate adjacent hexes, gradually reducing as the range increases.

My understanding of the 4-hex rule was that "search" happens in every quadrant out to 4-hexes, but not necessarily detection. No idea what the formula looks like under the hood, but your comment about reduced chance "as the range increases" seems logical.

For AI players (i.e. your opponent isn't as dangerous as a human), I think there is still some value in Search Arcs since the Z key gives you a good visual reminder of where your land-based search assets have been deployed.

[Arkham]

For AI players (i.e. your opponent isn't as dangerous as a human), I think there is still some value in Search Arcs since the Z key gives you a good visual reminder of where your land-based search assets have been deployed.

But it doesn't actually.

I think what happens is that in a search arc 1 die is rolled for each plane that traverses a hex. Hexes that is close to base are going to have a lot of planes traversing so there are a lot of rolls. As the search arc goes further out more and more hexes are getting hit with only 1 plane, thus one die roll.

I would also imagine that they are calculating the percentage based on factors such as experience, plane type, fatigue, leadership and then applying modifiers based on the total amount of tonnage in a given hex and the altitude of the search plane. A small sub or PT boat is going to be a very small chance to detect from a plane at 15K, but a large fleet of CVs and BBs is much easier to detect at 6K. So all those factors are combined to get a search value, and the die roll for the plane traversing the hex has to roll better than that value to get a successful 'hit'

But what I've seen in my testing the other day is that searches that are not using arcs are getting better hits. So what I think is happening there is that they are applying a malus amount to the search value since its not focused. What might have been a .500 search chance to roll against is being reduced down to a .2500 search chance. However, since ever hex is getting a roll to account for the unfocused search, the rules of probability say that a lot of lesser chances are better than 1 or two stronger chances.

If my WAG is correct, the fix for this would be to change the non focus modifier down quite a bit. Should be simple to code really, IF what I'm guessing is true.

But then again, I might be talking out my posterior on how this is coded so...

[RangerJoe]

For AI players (i.e. your opponent isn't as dangerous as a human), I think there is still some value in Search Arcs since the Z key gives you a good visual reminder of where your land-based search assets have been deployed.

But it doesn't actually.

I think what happens is that in a search arc 1 die is rolled for each plane that traverses a hex. Hexes that is close to base are going to have a lot of planes traversing so there are a lot of rolls. As the search arc goes further out more and more hexes are getting hit with only 1 plane, thus one die roll.

I would also imagine that they are calculating the percentage based on factors such as experience, plane type, fatigue, leadership and then applying modifiers based on the total amount of tonnage in a given hex and the altitude of the search plane. A small sub or PT boat is going to be a very small chance to detect from a plane at 15K, but a large fleet of CVs and BBs is much easier to detect at 6K. So all those factors are combined to get a search value, and the die roll for the plane traversing the hex has to roll better than that value to get a successful 'hit'

But what I've seen in my testing the other day is that searches that are not using arcs are getting better hits. So what I think is happening there is that they are applying a malus amount to the search value since its not focused. What might have been a .500 search chance to roll against is being reduced down to a .2500 search chance. However, since ever hex is getting a roll to account for the unfocused search, the rules of probability say that a lot of lesser chances are better than 1 or two stronger chances.

If my WAG is correct, the fix for this would be to change the non focus modifier down quite a bit. Should be simple to code really, IF what I'm guessing is true.

But then again, I might be talking out my posterior on how this is coded so...

What are the criteria for the computer to send out aircraft for specific search arcs? Until you can definitely define and state the parameters of those criteria, you don't know how the search works. You can presume (or assume) all that you want but until you know those things, you are guessing.

[Nezlezar]

I've always run Naval and ASW Search at 50%, and don't use arcs for any of the TFs and many of the islands. So my results have been pretty good, with very few attrition losses. You still want to train those pilots in the necessary skills, however. For example, the Japanese AI is pretty good at spotting my submarines but bomb hits are very rare. By late '42 most of my pilots have been trained up to 70 ASW and 70 Naval, so they not only spot, they're getting hits.

Pardon the necro'ing of this post.

70 NS / 70 ASW pilot

Is the above pilot getting more hits using the Naval Search mission or the ASW mission? Have you noticed a difference in hit rate based on the airplane class (ignoring range)? Like, Catalina vs MB but they're both set at range 12 for example. Same pilot in both...same odds to hit or not?

[RangerJoe]

I've always run Naval and ASW Search at 50%, and don't use arcs for any of the TFs and many of the islands. So my results have been pretty good, with very few attrition losses. You still want to train those pilots in the necessary skills, however. For example, the Japanese AI is pretty good at spotting my submarines but bomb hits are very rare. By late '42 most of my pilots have been trained up to 70 ASW and 70 Naval, so they not only spot, they're getting hits.

Pardon the necro'ing of this post.

70 NS / 70 ASW pilot

Is the above pilot getting more hits using the Naval Search mission or the ASW mission? Have you noticed a difference in hit rate based on the airplane class (ignoring range)? Like, Catalina vs MB but they're both set at range 12 for example. Same pilot in both...same odds to hit or not?

The last post is not that old but what are you trying to hit?

[Chris21wen]

I'll stick my neck out here and I have no proof what-so-ever with no intentions of testing it as it's would be a real time consuming exercise. I'm going on what is in the manual and numerous tests and observations others have carried out. Only the programmers know for sure as the exact algorithm has never been published. Here's my take on it.

It is harder to find subs than ships. In the game this is simulated by the ASW range being half of the naval search range. The idea being the ac spends more time in a hex looking for subs than ships. This in turn is directly affected by the ac endurance. The further any ac goes from it's base the less time it can spend over any specific hex. Then there's the arc to considers, the further out you go the larger the area within the arc is.

This is why it has been stated many times that any search over 10 hex is not effective and hexes close to the base (1-4 hex) are. All that assumes the ac has a range over 10 hex but what about ac with less range, say a Jake with ext range 10 or Pete 4?

The effective range now should be less due to endurance. I don't know what that is but it appears to be 2/3 normal range for search, say 6 for Jake and therefore only 3 for ASW, but that is offset by the fewer hexes within the arc to search. Note searches can go out to their extended range.

Then you have height, weather numbers with all of that entails!

Ignoring range/endurance/height/numbers it does not matter what type of aircraft is being used for the search, it's the crew (pilot in the game) that do the looking. The ac type only becomes important when tech is involved and once the target has been found, the more weapons it carries the more the likelihood of a successful hit.

This leads be to the ASW/Search skills. These skills are used both to find and attack, the higher the better for hits (>60). However I lower skill levels are effective at finding. Having said that I not entirely convinced that navB/T and or lowN skills do not matter?

All this is open to debate as there is no published algorithm.

[Nezlezar]

Certainly going to have to base much of this on observation without evidence. I am not trying to bomb anything in particular...it's more of a question that everything else being equal, would a search mission have the same result as an ASW mission? That is, both theoretical missions have a 70/70 ASW/NS pilot and they both have a submarine say, 2 hexes away that they've spotted. One pilot was put on a Nav Search mission, the other, ASW. Does the designation of that mission have any impact on the hit chance? Or is the difference between the mission largely that the range (on ASW) is reduced and thus increases the spotting chance? Going further with this hypothetical situation...now they are both on the same mission but the only variable is the aircraft. Sure, the ordnance that hits will vary but did the hit chance change if the only different variable (same altitude, distance from target, etc) was the aircraft used? Like, a float plane versus MB.

[RangerJoe]

The ASW mission flown at 1k would have a greater chance of surprising and attacking the submarine while the submarine is on the surface.

The higher the altitude, the harder it is to detect ships but a task force is easier to spot. Hence the 6k altitude being the best for searching although DBs on partial search with the rest on naval attack are best at 10k for searching plus dive bombing.