GA statistical musings

[GetAssista]

ORIGINAL: Lowpe
Thanks, how do you mod it for less than 10 factories?

F4 cell describes it.
More clear, mentioned COUNTIF function is in the same column, you should modify its range (which reflects # of facs) for the whole column

[Lowpe]

ORIGINAL: GetAssista

ORIGINAL: Lowpe
Thanks, how do you mod it for less than 10 factories?

F4 cell describes it.
More clear, mentioned COUNTIF function is in the same column, you should modify its range (which reflects # of facs) for the whole column

I read that, but I don't think it was in English.[:D] My bad,

So, if I were to allocate three size 30 factories on Dec 7, 1941 to research the Frank B (one of my favorite planes that I never use) it would come in March 10, 1944 with the engine bonus.

That seems way too early, an almost 15 month acceleration (6/45 to 3/44). I must be doing something wrong.

I ran a second, with a 12/45 plane with 1 factory and engine bonus and got 9 Oct 1944. I certainly am doing it wrong.[:(]

[GetAssista]

ORIGINAL: Lowpe
So, if I were to allocate three size 30 factories on Dec 7, 1941 to research the Frank B (one of my favorite planes that I never use) it would come in March 10, 1944 with the engine bonus.

That seems way too early, an almost 15 month acceleration (6/45 to 3/44). I must be doing something wrong.

I did the same, and am getting July-August 44 in arrival. Check that you changed COUNTIF range in all the formulas in the column F.
Other than that yeah, engine bonus is pretty powerful. Once all 3 facs are repaired you would crank up advances every 16-17 days

[pws1225]

Good post, it was wondering how factory repair worked. That's GA.

[GetAssista]

And the saga continues.
Today I want to bring your attention to some of the pecularities of HQs and preparation.

TL/DR: Don't leave home without your army/corp HQs - presence is more important than preparation.

The idea was to gain more understanding in what defines adjusted AV. Many things affect it, so I tried to isolate two of the more important and less understood ones - HQ and preparation percentage. Tests have a single Japanese ID attacking a single US regiment in a clear terrain base with forts at level 2. Commanders were edited to have all stats at 60, experience for infantry was set at 50, morale at 99. TOE of ID had 450 infantry squads and 450 support, TOE of US regiment is standard. Attack commensed at the start of scenario, so presumably clear weather all the way. I toggled prep between 0 and 100 both for attackers and defenders, and sometimes added HQs for the attacking side, both army/corp and command ones.

Many of the combinations are yet to be tested, so I want to point out two most important findings (for me)

1. The effect of army/corp HQs is much more important than ~10% to AV which I've seen circulated. HQc have a chance to double the adjusted AV, and all but eliminates leaders(-) penalties, that are usually disasterous (x2 to x4 drops in AV). Command HQ can further double adjusted AV on its own, and the effect is cumulative. With ID+HQc+HQ all at prep 100 I got 1994 adjusted AV once from 450 base. In ~50% attacks AV doubled, while in ~20% quadrupled.

2. Mere presense of the HQc is a big boon (more than x2) to average adjusted AV. Even with prep at zero. See on the picture how adjusted AV of an 100-prep division behaves if 0-prep HQc is added to the hex. With HQc in hex, attacking LCU stops getting leaders(-) almost entirely, while w/o HQ 70% of attacks got this negative modifier which generally halved adjusted AV.


3. Prep itself matters for LCUs, but not as much as HQ presense. Defending regiment with base AV 127 gets average adjusted AV ~50 with prep 0, and ~70 with prep 100. Adjusted AV is still consistently lower than base, but preparation alleviates some of the negative effect

[BillBrown]

Have or will you test if the HQ(s) have to be in the combat hex or is it effective if they/it are just withing their command range?

[Lokasenna]

ORIGINAL: BillBrown

Have or will you test if the HQ(s) have to be in the combat hex or is it effective if they/it are just withing their command range?

I would be very surprised if they didn't function just within range, not necessarily within the same hex.

@GA - did the units have enough Support devices without the presence of the HQs in the hex? That is the one indirect thing that an HQ in the hex could do outside of what you appear to have just tested. If they didn't have enough Support without the HQ units, I'm hesitant to chalk this difference up to the presence of HQs alone.

[BillBrown]

I would agree that they should, but there are some nooks and crannies in the code.

[Lokasenna]

ORIGINAL: BillBrown

I would agree that they should, but there are some nooks and crannies in the code.

This one would qualify as a bug.

[GetAssista]

ORIGINAL: Lokasenna
@GA - did the units have enough Support devices without the presence of the HQs in the hex? That is the one indirect thing that an HQ in the hex could do outside of what you appear to have just tested. If they didn't have enough Support without the HQ units, I'm hesitant to chalk this difference up to the presence of HQs alone.

I've done some tests earlier on that urban legend that support helps in adjusted AV. Had a tank regiment attack completely stripped of support. Adjusted AV was roughly equal to base AV even without any support squads in the hex.

Also:

ORIGINAL: GetAssista
TOE of ID had 450 infantry squads and 450 support

[szmike]

- how much HQ preparation affects combat?
- does command HQ affects combat the same way as HQc? say when LCU is in command range of unprepped command HQ, but no HQc around

[Lokasenna]

Ah, I missed the TOE line.

I'd also be interested to know if having a command HQ within range, but no corps or army HQ, results in a boost to AV. I suspect yes.

[GetAssista]

ORIGINAL: szmike
- how much HQ preparation affects combat?
- does command HQ affects combat the same way as HQc? say when LCU is in command range of unprepped command HQ, but no HQc around

ORIGINAL: Lokasenna
Ah, I missed the TOE line.

I'd also be interested to know if having a command HQ within range, but no corps or army HQ, results in a boost to AV. I suspect yes.

Had time to do some testing today with command HQs. Have some interesting results to report

TL;DR: LCU prep % helps in getting HQ bonus. HQ level of prep does not matter much, as long as HQ preps for the attacked base

As usual I ran a bunch of scenario starts with land battles in indentical bases happening, with IJ infantry division (TOE 450 squads+450 support) attacking US regiment (TOE 130 squads/mgs + 170 support). I wanted to get prefighting artillery duel and randoms associated with it out of the test. Command HQ with varying levels of prep was present either in hex, or in range, or absent altogether. ID also had varying prep for the base. All other editable variables (exp, morale, fatigue/disruption, commander stats) were set identical for all LCUs.

Results in the form of average observable adjusted AV (in % from base AV) for different combinations of factors is summarized in the table


Edit: updated a picture to a) take a peek at how middle prep works for LCUs. Seems like average HQ bonus linearly increases with it.
And b) made sure that non-prepping HQ bonus is not because of just its support squads

[Lokasenna]

What is the n on these?

It doesn't look like "in range, other" and "in hex, other" has a significant difference from each other. What about in range with prep?

[GetAssista]

ORIGINAL: Lokasenna
What is the n on these?

What do you mean n? Each % number is an average of 100 attack results as noted

ORIGINAL: Lokasenna
It doesn't look like "in range, other" and "in hex, other" has a significant difference from each other. What about in range with prep?

Yes, just the presense of HQ matters even if no relevant prep is there.
I did not test prep 100/50/0 in range instead of in hex, assumed HQ works like it should and provides the bonus while in range. Too much work for too little potential gain to test if this is indeed true (I assume it is).

edit: update the picture some

[Lokasenna]

My eyes glazed and I missed the 100.

[Lowpe]

GA, command HQ with no subordinate HQs? I wonder if range plays a part...i.e. the greater the range the less the bonus?

Good information.[&o]

A 2014 post by Alfred on HQs...


1. LCU looks to see if a Corps HQ is in range. If a Corps HQ is found, proceed to points 2-6 below. If no Corps HQ is found, proceed to points 7-8 below



2. If Corps HQ is within range, the level and target of the Corps HQ is checked and a combat bonus within the range 0-10% may be given.

3. If more than one Corps HQ is within range, the levels and targets of the additional Corps HQs are not checked.

4. If a Corps HQ within range was found, the LCU then sees if a Command HQ is within 2x range.

5. If Command HQ is within 2x range, the level and target of the Command HQ is checked and an additional combat bonus within the range 0-90% may be given.

6. If more than one Command HQ is within 2x range, the levels and targets of the additional Command HQs are not checked.



7. If no Corps HQ is found by the LCU, it then looks to see if a Command HQ is in 1x range.

8. If a Command HQ is within 1x range, the level and target of the Command HQ is checked and a combat bonus within the range 0-10% may be given.



9. The combat bonus applies to combat conducted inside the named base.

10. An Army HQ = a Corps HQ. The only difference is that a Corps HQ usually has only a one hex range whereas an Army HQ might have a range of up to five hexes.


A LCU may therefore receive assistance from:

(a) one Corps HQ only, or
(b) one Corps HQ and one Command HQ, or
(c) one Command HQ only

Alfred

[Blackhorse]

GetAssista,

I just returned to the forum after a multi-year hiatus. I'm scanning the forums to glean whatever knowledge I can that I missed while I was away. This is great analysis.

Did you do any separate analysis on the effects of prep and HQs in non-base hexes, and/or on defending units?

TL;DR: LCU prep % helps in getting HQ bonus. HQ level of prep does not matter much, as long as HQ preps for the attacked base

As usual I ran a bunch of scenario starts with land battles in indentical bases happening, with IJ infantry division (TOE 450 squads+450 support) attacking US regiment (TOE 130 squads/mgs + 170 support). I wanted to get prefighting artillery duel and randoms associated with it out of the test. Command HQ with varying levels of prep was present either in hex, or in range, or absent altogether. ID also had varying prep for the base. All other editable variables (exp, morale, fatigue/disruption, commander stats) were set identical for all LCUs.

Results in the form of average observable adjusted AV (in % from base AV) for different combinations of factors is summarized in the table


Edit: updated a picture to a) take a peek at how middle prep works for LCUs. Seems like average HQ bonus linearly increases with it.
And b) made sure that non-prepping HQ bonus is not because of just its support squads




Attachment (1)

< Message edited by GetAssista -- 6/24/2017 1:39:51 PM >

[GetAssista]

ORIGINAL: Blackhorse
GetAssista,

I just returned to the forum after a multi-year hiatus. I'm scanning the forums to glean whatever knowledge I can that I missed while I was away. This is great analysis.

Did you do any separate analysis on the effects of prep and HQs in non-base hexes, and/or on defending units?

Thanks!
I did not make a separate testing for non-bases. I assume HQ presense bonus will still be there (50% -> 70% adjusted AV), but not any of the prep bonus.
Also I assume defenders have it exactly the same as attackers with respect to HQ bonuses at least. Other bonuses to AV are very different for defenders compared to attackers so it might be worth testing some time in future.

[Blackhorse]

ORIGINAL: GetAssista


Thanks!
I did not make a separate testing for non-bases. I assume HQ presense bonus will still be there (50% -> 70% adjusted AV), but not any of the prep bonus.
Also I assume defenders have it exactly the same as attackers with respect to HQ bonuses at least. Other bonuses to AV are very different for defenders compared to attackers so it might be worth testing some time in future.

Bolded above for emphasis. I suspect that HQ benefits are different at non-base hexes, and that Defender AV bonuses differ from Attacker AV bonuses. Should you find the time to test those hypotheses, many of us would be keenly interested in what you find.

Regardless of whether you do, or not, here's a salute for the work and analysis you've already put in.