Matrix Games Forums

ORIGINAL: geofflambert

ORIGINAL: obvert

ORIGINAL: geofflambert

I don't know about y'all, but if I'm sweeping I'm doing it at an altitude below what I think the maximum is for the CAP. However, if I come in too low and the CAP's too high, there may be no engagement. Likewise if I come in too high and the CAP's too low there may be no engagement. If you're using Sweep as a tactic, and no engagement occurs, that is what some people call a failure. Attrition may go against you on any given day, but if it is the right tactic to use (you desire to achieve air superiority and you believe you have the means to do it) attrition is what happens. Don't try to be too cute. The only reason to sweep at max alt is because you don't have air superiority or any hope of getting it but just want to stir things up so maybe some of your bombers get through.

Why would you sweep at an altitude lower than the highest CAP? Don't you then have the CAP gain the dive?

I've not experienced the CAP not being engaged by even a 42k sweep so far, and the next test has the highest CAP at 9k.

Well, I have experienced having CAP too high and not intercepting bombers, much less sweeps. My position stands.

edit: I've also experienced having my sweeps too high and not encountering CAP,

Definitely, I've had this happen with bombers for sure. Sweeps seem a different thing though.

ORIGINAL: Lowpe

ORIGINAL: geofflambert

ORIGINAL: obvert




Why would you sweep at an altitude lower than the highest CAP? Don't you then have the CAP gain the dive?

I've not experienced the CAP not being engaged by even a 42k sweep so far, and the next test has the highest CAP at 9k.

Well, I have experienced having CAP too high and not intercepting bombers, much less sweeps. My position stands.

edit: I've also experienced having my sweeps too high and not encountering CAP,

What you might be seeing is sweepers engaging a training squadron and there is no actual combat but the replay pops up for second or two.

I have never seen no combat between dedicated sweepers and enemy fighters on CAP. I have seen drastically short combats though because of the altitude difference and it seems more pronounced for Japan (as the sweepers) than the Allies.

right, and another possibility is that the CAP was exhausted from previous sweeps and attacks. I usually throw everything at the same target. I never tried to pin it down like Obvert is.

ORIGINAL: mind_messing

For future reference, can you record detection range? As in, in the CR where it says "Time to target X minutes"?

It's different for each group of course, and the interesting thing for me is that it does seem there is a correlation there between low CAP being successful and time to intercept.

So I've been taking screenshots of the CRs since they are overwritten each day, but I think I'll start also pushing a day ahead with each iteration so I can keep the CRs themselves. I'm as I say just getting a feel for what needs to be tested, and I haven't even started watching the combats yet. I would like to watch specific ones to see what happened over time. Sweeping with two groups does show just enough difference between them though that I think this will show a lot.

ORIGINAL: Lowpe

Bingo! I could tell you why, but lets see if you can figure it out.[;)]

I'm getting some ideas now, but i'll hold off until I can look through the CRs or match up CRs with the other data. It does feel like detection, time to target and time to intercept are big factors. I'll also have to test without radar (I've added a lot here to make sure some of it is working) and at different distances. I realize now 4 hexes may be too close for some radars. The US CPS-1 detects to 198 miles! [X(]

Just added the Time to Target notes from the CRs I had. Thought I had them all but not quite.

From that data though it doesn't look like the radar is working? Those are some very short detection times. Maybe I've got to tweak my test scenario. Type 3 is the best IJN radar for bases, right?

When I did my testing and added radar the detection times were all awful (about 10 or so minutes), but the results were drastically improved for the CAP.

Also, I don't think radar has any effect outside the hex it is in, but the longer ranges still help you. Confusing, I know, but it is all part of the abstraction. Short version, I don't think moving further than 4 hexes will have any effect, regardless of the radar range.

ORIGINAL: obvert

Just added the Time to Target notes from the CRs I had. Thought I had them all but not quite.

From that data though it doesn't look like the radar is working? Those are some very short detection times. Maybe I've got to tweak my test scenario. Type 3 is the best IJN radar for bases, right?

Maybe I missed it in your notes when I scrolled back up.

You should use the -archive switch to keep every CR. The one in the simple SAVE folder will be overwritten daily, but you will have dated ones in your archive folder within the SAVE folder.


Sweeps function differently than bombing raids. I have seen a few sweeps where the attacking fighters do their "Ki-84a Frank sweeping at 34000 feet" and the visual is shown over the target base before the CAP is completely depleted, so there seems to be a random function in here for when the battle is over... however, for the most part, CAP and sweepers will always fight because the sweeps are looking for the CAP, and vice versa. For bombing raids, if the CAP doesn't have time to get to the raid, then it just begins its attack runs on the target.

I have had fewer strikes get through CAP when CAP was too high than when CAP was too low. I would rather err on the high side, and again I don't know why anyone would ever sweep at anything other than their plane's maximum altitude, so would rather err on the high side there also.

I would love to see the second test redone with the P47s sweeping at 15,000 feet

Erik, regarding sweeps and radar. I tested this a while back.

Have a squadron sweep a base. Note the TTT. Then switch them to AF attack and note the TTT. Results used to be very...interesting a while back. Don´t know if its fixed now.

IIRC Radars work just for the target hex. However longer detection range means more "rolls" for each radar device - so a radar could have 1 roll with range of 20 (it is assumed it is in centre of the hex), 2 rolls with range 60, 3 rolls with range 100, etc.. or something in that way...

Succesfull radar detection is critical in getting your own planes into the air - and possibly above the attackers regardless of CAP setting altitude ("Group altitude set 20000, climbing to 17000 to 31000" message).

So here is another that I did late last night and accidently posted in John III's AAR. Whoops! Anyway, it's here now.

This is one to show the effects of an all high CAP at 31k. Quite a big difference! Should also point out that this is not a layered CAP, which based on earlier results seems more effective. I'll test a high layered CAP next for comparison.

ORIGINAL: BillBrown

I would love to see the second test redone with the P47s sweeping at 15,000 feet

We'll get there! I'l try to exhaust the variations. This is by no means skewed to help one side or another, and the fact that I've picked a very good Japanese plane is really to at least at first make sure that the defender does have a chance to do some damage against the P-47. After this series I'll mix up airframes too.

ORIGINAL: JocMeister

Erik, regarding sweeps and radar. I tested this a while back.

Have a squadron sweep a base. Note the TTT. Then switch them to AF attack and note the TTT. Results used to be very...interesting a while back. Don´t know if its fixed now.

Thanks Jocke. I'll test that next just to see radar use. I suspect that radar is working. Interestingly I'm starting to wonder though if the short TTT is actually a slight advantage for the defending CAP if placed low, since the groups intercept en masse rather than being hit piecemiel. We'll see as we go.

ORIGINAL: Barb

IIRC Radars work just for the target hex. However longer detection range means more "rolls" for each radar device - so a radar could have 1 roll with range of 20 (it is assumed it is in centre of the hex), 2 rolls with range 60, 3 rolls with range 100, etc.. or something in that way...

Succesfull radar detection is critical in getting your own planes into the air - and possibly above the attackers regardless of CAP setting altitude ("Group altitude set 20000, climbing to 17000 to 31000" message).

Thanks Barb.

Yep, the time to intercept message are very important I'm feeling but with low CAP maybe not in the way we're used to thinking about. As above, I wonder after seeing the first sweep getting hit hard by the layered low CAP if having them all low, not yet able to climb, meant they won the number game by so much they overwhelmed even the faster diving P-47s. I'll try to isolate this to work on my suspicians.

Ok. One more to help round out the high CAP option.

This one has layered CAP at 31k, 28k and 26k. The sweeps again are coming at 42k.

I noticed more fluctuation of results here, with some very good showings by the Franks and other runs where they're crushed at over 3:1. Odd. Trying to make correlations even in the miniscule differences i TTT, but so far nothing seems to fit. Let me know if you see anything.

(Forgot to make Time to target on center column here)

Even though you're doing 10 runs, that's still a pretty small n.

I have a hypothesis for variable results in layered CAP. There is probably some randomness in which unit gets dived on by the strato sweepers. I would guess that in the instances where the Franks achieved a positive ratio, more P-47s dove on the lowest band of CAP and were in turn dived on by higher Franks. When the Franks lost outright, it may have been because the P-47s dove on the highest first and then continued to dive before the Franks could climb high enough to dive.

It is more intense work in the editor to sort this out, but what I would do to sort out which unit was getting dived on is create some identical aircraft profiles that just happen to have different names. Like Ki-84r LOW, Ki-84r MIDDLEONE, Ki-84r MIDDLETWO, Ki-84r HIGH. You could then see messages like "P-47D25 diving on Ki-84r LOW" and know which unit is getting attacked.

ORIGINAL: Lokasenna

Even though you're doing 10 runs, that's still a pretty small n.

I have a hypothesis for variable results in layered CAP. There is probably some randomness in which unit gets dived on by the strato sweepers. I would guess that in the instances where the Franks achieved a positive ratio, more P-47s dove on the lowest band of CAP and were in turn dived on by higher Franks. When the Franks lost outright, it may have been because the P-47s dove on the highest first and then continued to dive before the Franks could climb high enough to dive.

It is more intense work in the editor to sort this out, but what I would do to sort out which unit was getting dived on is create some identical aircraft profiles that just happen to have different names. Like Ki-84r LOW, Ki-84r MIDDLEONE, Ki-84r MIDDLETWO, Ki-84r HIGH. You could then see messages like "P-47D25 diving on Ki-84r LOW" and know which unit is getting attacked.

I agree it's a small sample to start. I'm really feeling out what needs more testing and hoping to get just this kind of feedback to make the tests do more, or work to certain goals.

It still surprises me that a low CAP setup would be more successful than the other ranges tried so far. I know this is a bare bones test too, and I do want some color in it from differing airframes, pilot experience, and other stuff that wold fit game play experiences better.

Afte going through a full cycle with these I'll try one for another 10 runs and see if the results are close to the same percentage lost. If they ar then maybe 10 is enough to get a good idea of each situation. If not then maybe I'll have to make the set larger for later tests.

The diving on the lowest group would certainly make sense to give much better results potentially, assuming it's limited to diving on a single group. That's a long dive too. The highest defending group could be sitting there moving straight ahead level and watch the planes just dive by and blast em.

Need a much bigger sample though, but this is fascinating indeed.

Somewhere, The Elf (the air developer) is smiling at this attempt to divine his algorithms. [;)]

I personally would like to see the number of radars as a variable, say zero, one, three and six. There were some counterintuitive results in an AAR that made it seem like more radars may have made things worse. Would be good to put that to rest.

Cheers,
CC

ORIGINAL: obvert

So here is another that I did late last night and accidently posted in John III's AAR. Whoops! Anyway, it's here now.

This is one to show the effects of an all high CAP at 31k. Quite a big difference! Should also point out that this is not a layered CAP, which based on earlier results seems more effective. I'll test a high layered CAP next for comparison.

But these results might be a function of the altitude. No Japanese fighters performed well at that sort of height. Note how much the Frank degrades in maneuverability between 10,000k and 31,000k, whereas the P47 hardly degrades at all. So at 10,000 feet you have a much slower fighter with better maneuverability taking on a faster fighter with poor rating. However at 31,000 feet you have a slower fighter with poor maneuverability taking on a faster fighter that actually maneuvers fairly well at that height. All things considered speed is the strongest asset a fighter has but when you are slower and also lose maneuverability to boot, the results make sense.

So low CAP might be better for a Japanese aircraft, but then not ideal for any late war American. Worth testing.

And when you set up your test map you can turn off advanced weather effects and every day will be partially cloudy for consistency.

ORIGINAL: crsutton

ORIGINAL: obvert

So here is another that I did late last night and accidently posted in John III's AAR. Whoops! Anyway, it's here now.

This is one to show the effects of an all high CAP at 31k. Quite a big difference! Should also point out that this is not a layered CAP, which based on earlier results seems more effective. I'll test a high layered CAP next for comparison.

But these results might be a function of the altitude. No Japanese fighters performed well at that sort of height. Note how much the Frank degrades in maneuverability between 10,000k and 31,000k, whereas the P47 hardly degrades at all. So at 10,000 feet you have a much slower fighter with better maneuverability taking on a faster fighter with poor rating. However at 31,000 feet you have a slower fighter with poor maneuverability taking on a faster fighter that actually maneuvers fairly well at that height. All things considered speed is the strongest asset a fighter has but when you are slower and also lose maneuverability to boot, the results make sense.

So low CAP might be better for a Japanese aircraft, but then not ideal for any late war American. Worth testing.

And when you set up your test map you can turn off advanced weather effects and every day will be partially cloudy for consistency.

I did switch off advanced weather halfway through the first set, and started to get fewer thunderstorms. I'm also walking the test ahead each turn to get a different weather, hopefully.

The interest for me in the high results is the difference between high layered and high flat. The airframes may not be as good high, but they perform much better layered from 31k than flat, and by the numbers they are still more maneuverable than the P-47D2 up there. The results are almost as good though as when set to 10k, 10k, and 7k.

Moral of story (so to speak):

Counter "stratosphere" sweeps with layered CAP at altitudes where your planes have good maneuverability, it works. [8D]