Some tank duel tests (CMBN)

[MOS:96B2P]

BUMP.  This is an interesting topic by @Drifter Man with some good information backed up by testing.  Thought I would bump it. 

[Sgt.Squarehead]

Interesting.....Strongly supports the position that 'Target Arcs' don't actually improve spotting, yet they always seem to.

Perception is an interesting thing. 

[Drifter Man]

Cheers, this is actually a good time to get bumped.

I have been working on the spotting problem, but I got a bit too ambitious and it is going to take some time before I have a complete piece of work to show.

I have set up a test scenario with one vehicle in the middle and 49 enemy Panzer IVs placed around it at 12 to 6 o'clock positions, 200 to 1400 meters (in 200m intervals). The game runs for up to 7 seconds for me to see what contacts does the vehicle at the center see on the first spotting event, as a solid contact. Repeat 10,000 times and calculate the probability of seeing each of the enemy Panzer IVs.

I can replace the vehicle in the middle and measure the spotting ability of different vehicles. I keep all the data but only take those at 12 and 1 o'clock positions and convert them to a single number, which is easier to interpret. I call this number "spotting rating" and make it relative to Panzer IVH.

So, a Regular Panzer IVH with hatches open has a spotting rating of 100. And a tank that has spotting rating 105 can see, roughly, 5% better than a Panzer IVH. Note that it depends on distance quite a bit, but that would make things too complicated for a quick comparison.

Results so far (opened up / buttoned up):

• Pz IVH (late): 100 / 27
• Pz VD (late): 104 / 29
• Pz VA (mid): 109 / 39
• Sherman, no cupola: 109 / 35
• Sherman, with cupola: 110 / 39
• Sherman VC Firefly: 105 / 28
• M10: 220 / 217
• T-70M: 89 / 17
• T-34-76 (M1942 early): 97 / 21
• T-34-76 (M1942 late): 95 / 25
• T-34-85: 106 / 36
• IS-1, IS-2: 101 / 26
• Valentine: 100 / 24
• SU-76M: 212 / 211
• SU-85, SU-122: 95 / 16
• SU-85M, SU-100: 95 / 18
• ISU-122, SU-152, ISU-152: 104 / 29
• SU-57: 252 / 248

You can see there are two categories: fully enclosed vehicles and open-topped vehicles. The open-topped ones (M10, SU-57, SU-76M) spot more than 2x better than fully enclosed ones and do not suffer their ~3x spotting penalty when buttoned up.

A quick takeaway - if you are trying to get an spotting advantage by forcing the enemy M10 commander to duck inside, don't. It is not much help. The same will probably apply to Marders, Archers and the like.

Even at 10,000 trials there still are statistics involved, for instance I don't think the T-34 M1942 late should have a lower spotting ability than M1942 early when opened up.

Everything is automated, of course, with minimum time and effort required on my part. I just collect the results and set up a new test. Computer time is my bottleneck.

[dbsapp]

6 minutes ago, Drifter Man said:

Cheers, this is actually a good time to get bumped.

I have been working on the spotting problem, but I got a bit too ambitious and it is going to take some time before I have a complete piece of work to show.

I have set up a test scenario with one vehicle in the middle and 49 enemy Panzer IVs placed around it at 12 to 6 o'clock positions, 200 to 1400 meters (in 200m intervals). The game runs for up to 7 seconds for me to see what contacts does the vehicle at the center see on the first spotting event, as a solid contact. Repeat 10,000 times and calculate the probability of seeing each of the enemy Panzer IVs.

I can replace the vehicle in the middle and measure the spotting ability of different vehicles. I keep all the data but only take those at 12 and 1 o'clock positions and convert them to a single number, which is easier to interpret. I call this number "spotting rating" and make it relative to Panzer IVH.

So, a Regular Panzer IVH with hatches open has a spotting rating of 100. And a tank that has spotting rating 105 can see, roughly, 5% better than a Panzer IVH. Note that it depends on distance quite a bit, but that would make things too complicated for a quick comparison.

Results so far (opened up / buttoned up):

• Pz IVH (late): 100 / 27
• Pz VD (late): 104 / 29
• Pz VA (mid): 109 / 39
• Sherman, no cupola: 109 / 35
• Sherman, with cupola: 110 / 39
• Sherman VC Firefly: 105 / 28
• M10: 220 / 217
• T-70M: 89 / 17
• T-34-76 (M1942 early): 97 / 21
• T-34-76 (M1942 late): 95 / 25
• T-34-85: 106 / 36
• IS-1, IS-2: 101 / 26
• Valentine: 100 / 24
• SU-76M: 212 / 211
• SU-85, SU-122: 95 / 16
• SU-85M, SU-100: 95 / 18
• ISU-122, SU-152, ISU-152: 104 / 29
• SU-57: 252 / 248

You can see there are two categories: fully enclosed vehicles and open-topped vehicles. The open-topped ones (M10, SU-57, SU-76M) spot more than 2x better than fully enclosed ones and do not suffer their ~3x spotting penalty when buttoned up.

A quick takeaway - if you are trying to get an spotting advantage by forcing the enemy M10 commander to duck inside, don't. It is not much help. The same will probably apply to Marders, Archers and the like.

Even at 10,000 trials there still are statistics involved, for instance I don't think the T-34 M1942 late should have a lower spotting ability than M1942 early when opened up.

Everything is automated, of course, with minimum time and effort required on my part. I just collect the results and set up a new test. Computer time is my bottleneck.

Sounds interesting.

It would be nice to see the data on modern tanks in CMBS and CMCW. 

[Redwolf]

6 hours ago, Drifter Man said:

Cheers, this is actually a good time to get bumped.

I have been working on the spotting problem, but I got a bit too ambitious and it is going to take some time before I have a complete piece of work to show.

I have set up a test scenario with one vehicle in the middle and 49 enemy Panzer IVs placed around it at 12 to 6 o'clock positions, 200 to 1400 meters (in 200m intervals). The game runs for up to 7 seconds for me to see what contacts does the vehicle at the center see on the first spotting event, as a solid contact. Repeat 10,000 times and calculate the probability of seeing each of the enemy Panzer IVs.

I can replace the vehicle in the middle and measure the spotting ability of different vehicles. I keep all the data but only take those at 12 and 1 o'clock positions and convert them to a single number, which is easier to interpret. I call this number "spotting rating" and make it relative to Panzer IVH.

So, a Regular Panzer IVH with hatches open has a spotting rating of 100. And a tank that has spotting rating 105 can see, roughly, 5% better than a Panzer IVH. Note that it depends on distance quite a bit, but that would make things too complicated for a quick comparison.

Results so far (opened up / buttoned up):

• Pz IVH (late): 100 / 27
• Pz VD (late): 104 / 29
• Pz VA (mid): 109 / 39
• Sherman, no cupola: 109 / 35
• Sherman, with cupola: 110 / 39
• Sherman VC Firefly: 105 / 28
• M10: 220 / 217
• T-70M: 89 / 17
• T-34-76 (M1942 early): 97 / 21
• T-34-76 (M1942 late): 95 / 25
• T-34-85: 106 / 36
• IS-1, IS-2: 101 / 26
• Valentine: 100 / 24
• SU-76M: 212 / 211
• SU-85, SU-122: 95 / 16
• SU-85M, SU-100: 95 / 18
• ISU-122, SU-152, ISU-152: 104 / 29
• SU-57: 252 / 248

You can see there are two categories: fully enclosed vehicles and open-topped vehicles. The open-topped ones (M10, SU-57, SU-76M) spot more than 2x better than fully enclosed ones and do not suffer their ~3x spotting penalty when buttoned up.

A quick takeaway - if you are trying to get an spotting advantage by forcing the enemy M10 commander to duck inside, don't. It is not much help. The same will probably apply to Marders, Archers and the like.

Even at 10,000 trials there still are statistics involved, for instance I don't think the T-34 M1942 late should have a lower spotting ability than M1942 early when opened up.

Everything is automated, of course, with minimum time and effort required on my part. I just collect the results and set up a new test. Computer time is my bottleneck.

Those are awesome measurements.

Do you have an idea how this compares to infantry with/without binocolars?

The advantage of the open top vehicles is unexpectedly large. This could be highly influential on choices in Quickbattles.

[Drifter Man]

21 hours ago, dbsapp said:

Sounds interesting.

It would be nice to see the data on modern tanks in CMBS and CMCW. 

I am happy to share the method I use with anyone who wants to try the tests on different CM titles. It should be easily portable. Basic knowledge of python is required.

14 hours ago, Redwolf said:

Those are awesome measurements.

Do you have an idea how this compares to infantry with/without binocolars?

The advantage of the open top vehicles is unexpectedly large. This could be highly influential on choices in Quickbattles.

I have an idea - I think a tank with the hatch open is basically one guy with binoculars + a few more guys watching through narrow vision blocks. So, one infantry with binoculars should see a bit less than a tank with hatches open.

I also think that the open top vehicles have this big advantage because there are more guys looking out without restriction at the same time. When there are three pairs of eyes looking out, the chances of seeing something greatly increase.

[Bulletpoint]

21 hours ago, Drifter Man said:

A quick takeaway - if you are trying to get an spotting advantage by forcing the enemy M10 commander to duck inside, don't. It is not much help.

This is really interesting. Always thought the Tank Destroyers would spot much worse when the crew keep their heads down, but seems like the game simulates them peeking over the edge anyway, even though we don't see it graphically.

Unbuttoning might have a bigger effect at long range though, if opening up means the vehicle commander is using his binoculars. If I understand your test correctly, I believe you are measuring all distances out to 1400m but combining them into one single spotting rating. Maybe the signal of the improved spotting at long range when unbuttoned is being swamped by all the spotting power at close range where the whole crew can help out. Just an idea, maybe you already took that into account.

Thanks for sharing in any case. Looking forward to the data on the Marder.

[chuckdyke]

I will watch out how an SU 76 spots vs a T34/76. 

[Drifter Man]

Here are two examples of the complete data. The numbers give the measured probability (in %) of spotting a Pz IVH as a full contact in a single spotting event. Flat ground, grass, clear weather, midday. The first one is for a typical tank, the other one is for an open top tank destroyer. I would say that in both cases the effect of buttoning up (or lack thereof) is consistent at all ranges. For the SU-76M, there is a noticeable penalty to the sides and rear when buttoned up, but not in the ahead direction.

Left half = opened up, right half = buttoned up

T-34-85:

 

SU-76M:

T-34-76 (M1942 Late) - special for @chuckdyke

 

Edited October 25, 2021 by Drifter Man

[chuckdyke]

  So the SU 76 spots much better than the T34 which is fair. I am fond of the SU76 spots very well in Hammers Flank especially with infantry nearby. Thanks for your effort. 

[Drifter Man]

This is what the test setup looks like:

[Bufo]

58 minutes ago, Drifter Man said:

This is what the test setup looks like:

The spotting depends on the vision slits, which can differ for each side of the vehicle (so asymmetrical). For example the BMP-2 with the driver and gunner alone (infantry dismounted) is basically blind directly to the right side of the vehicle because neither the driver, nor the gunner has a vision block looking on that side.

So for some vehicles spotting is not the same for left and right sides of it. Hence the test should use a full circle, not just a half.

Edited October 25, 2021 by Bufo

[Drifter Man]

12 minutes ago, Bufo said:

The spotting depends on the vision slits, which can differ for each side of the vehicle (so asymmetrical). For example the BMP-2 with the driver and gunner alone (infantry dismounted) is basically blind directly to the right side of the vehicle because neither the driver, nor the gunner has a vision block looking on that side.

So for some vehicles spotting is not the same for left and right sides of it. Hence the test should use a full circle, not just a half.

Ideally yes - that was my first plan, too - but I had to consider two things:

• Spotting in the ahead direction is the most important thing anyway, and left-right asymmetries should not matter much when looking forward
• A full circle does not fit on the screen for me, so I would have to test two half-circles... twice as much time.

[Sgt.Squarehead]

Scratch that, I missed the point. 

Edited October 25, 2021 by Sgt.Squarehead

[Redwolf]

4 hours ago, Drifter Man said:

I have an idea - I think a tank with the hatch open is basically one guy with binoculars + a few more guys watching through narrow vision blocks. So, one infantry with binoculars should see a bit less than a tank with hatches open.

I also think that the open top vehicles have this big advantage because there are more guys looking out without restriction at the same time. When there are three pairs of eyes looking out, the chances of seeing something greatly increase.

Full agreement, but I don't think the latter should apply to straight to the front spotting. Looking straight ahead on open hatch tank should spot equally to an open top AFV, because the guys without binoculars looking to the sides don't contribute to that. And both still have the gun sight looking to the front in addition to the commander.

[Bulletpoint]

2 hours ago, Drifter Man said:

A full circle does not fit on the screen for me, so I would have to test two half-circles... twice as much time.

Maybe it would be better to evaluate the half circle of the whole 180 degree frontal arc then, instead of left and right sides?

[Drifter Man]

19 hours ago, Redwolf said:

Full agreement, but I don't think the latter should apply to straight to the front spotting. Looking straight ahead on open hatch tank should spot equally to an open top AFV, because the guys without binoculars looking to the sides don't contribute to that. And both still have the gun sight looking to the front in addition to the commander.

Could be - it is also possible that the non-binocs see quite well. I'll test a guy with binocs when I have time for him

19 hours ago, Bulletpoint said:

Maybe it would be better to evaluate the half circle of the whole 180 degree frontal arc then, instead of left and right sides?

That is an option, but I chose the other one. For most cases, having a 180 deg frontal arc would give me redundant data - I doubt that many vehicles have significantly asymmetric spotting - and I would lose the potentially interesting information about spotting to the rear, which could be interesting. Tradeoffs everywhere...

Although maybe I should change the circular plots so that they do not give false impression that the data comes from full circle.

[Redwolf]

So here is a test scenario maybe showing whether giving tanks an infantry sidekick improves spotting. Both sides have 10 T-34/85. Allied side also has one sharpshooter team per tank (see screenshot). Axis tanks spot for themselves only.

I don't have this automated, my sample size is rather small. So I want you suckers to also run the scenario (either side or hotseat) and report back - does the infantry sidekick appear to make a difference for you? For me, the side with the extra spotters wins every time with something between 10:3 to 10:7 losses.

 

9spott3485.btt

[Redwolf]

Attaching test scenario as zip file in case there are download problems.

 

9spott3485.zip

[Vergeltungswaffe]

I ran it 3 times, hotseat on iron and the spotter side won 10-3, 10-7, and 10-4.

Certainly appears to be an advantage to me.

[Drifter Man]

I reported something similar earlier in this thread. The infantry can provide a tentative contact to the tank, which is then more likely to turn that into a firm contact it can shoot at. When the tank has intel before the duel starts, it wins 71% of the time, all else being equal.

I also confirmed that a tentative contact from an infantry units is the same as from scenario early intel.

Table 7. Supplementary tests. Effect of a tree standing directly in front of the Defender. Effect of a contact marker provided to the Defender by giving him 100% intel strength. Both AFVs are stationary on Grass and the crew hatches are open.

Attacker		Defender
Grass	44%	54%	Grass + 1x Type B Tree as protection
No intel	29%	71%	100% intel

[domfluff]

Yes, this is horizontal sharing in action. Partial contacts upgrade to full contacts, so if you have 10 sets of eyeballs spotting without ports to look through, the infantry will spot first, then share this contact with the armour.

[domfluff]

Notably, this is also the correct usage for BMPs post-dismount - if you do dismount (and you definitely shouldn't in all cases), then keeping the BMP within 4 action spots of the dismounts is the correct procedure in most cases. Equally, the BMP should be area-firing at partial contacts, but that's secondary.

[MOS:96B2P]

9 hours ago, Drifter Man said:

infantry can provide a tentative contact to the tank, which is then more likely to turn that into a firm contact it can shoot at. When the tank has intel before the duel starts, it wins 71% of the time, all else being equal.

+1.  Interesting and useful.  Thank you for all the work that goes into your testing. 

[KGBoy]

On 10/27/2021 at 2:45 PM, MOS:96B2P said:

+1.  Interesting and useful.  Thank you for all the work that goes into your testing. 

I concur!