Question Re. Expected Combat Loss Indicators

[MJY]

Based on approximately 50-60 SCI/SCII campaigns, it appears to me that the computer-generated on screen expected combat loss indicators in both versions of the SC system under-estimate expected attacker casualties about twice as often as they under-estimate expected defender casualties. One would think that random probability would have the expected combat loss indicator right, say, 50% of the time with attacker and defender losses under-estimated about 25% of the time each.

My questions are as follows. First, have other players noticed this (e.g., any hard data). Second, are these skewed probabilites deiberately programmed into the software as a means of encouraging agressive play and offensive strategies from players that would otherwise be risk-adverse? And, third, is there a patch for this/in the works for this?

I look forward to hearing from the Forum.

[ChrisND]

As far as hard data goes, I am taking a statistics class right now and I have to do a project/study. I have picked the damage model for my study. Expect a full report in a few weeks.

My hypothesis is that this is selective memory, but we shall see.

[MJY]

Normal Dude, thanks for the quick reply. I would be interested in seeing the results when available. (A good opportunity to put the stats I learned in grad school to good use.)

For myself, I did track results for random periods of several SCI games. My results were insufficiently large-n to be statistically meaningful, but I felt I recorded enough results to discount selective memory/observer bias.

[targul]

Is it statisically possible that stat classes are a waste of time?

I remember mine in undergrad and grad school and I am pretty sure I had a great time and learned alot of stat. That made me believe I could convince anyone anything is wrong given the parameters I set fot the model.

Anyway if you examine the various items not included in the odds you probably find your problem. It tells you want is not included so the odds are only a basic not final odds are given.

[Liam]

I have always pondered those little #s from the first time I noticed then in SC1, it was very helpful then. From what I understand and I'm not sure this is factual those #s are 100% accurate but they do not factor in the Dice and with tech in SC2 I believe the Dice are exponential...

i.e. IW2 would have a factor of 2 strength points possible destroyed, likewise for defense

I'm not the Math Pro here though

[pzgndr]

Are you playing with FOW on or off? It makes a difference in how the estimated losses are calculated. With FOW on, some things like HQ leader bonuses and experience are hidden and not included in the calculations. So yes, these estimates are skewed but that's a deliberate feature.

Is there evidence that these loss estimates are incorrect when FOW is turned off, and accounting for the +/-1 variance? That would indicate a bug.

[Retributar]

...In other word's ..."Houston!!!...We Have a Problem!!!...A MAJOR MALFUNCTION!!!"

[MJY]

Pzgndr,

Thanks! FOW may well be the explanation. It certainly makes sense. (I have never played SCI or SCII without FOW.)

Is your conclusion based on looking at the code, game play experience, or something else? I will wait and see if other players support this conclusion and if they do I will consider this topic definitively settled and close the topic.

Thanks!

[pzgndr]

MJY, it's based on Hubert telling me that's what he did and me writing it up that way in the User Manual.

From page 40:

As an aid, expected combat losses are displayed at the top of the screen prior to resolving combats. This is a fairly accurate estimate for beginners when the fog of war game option is turned off. However, if fog of war is being used then some hidden enemy factors such as HQ command rating and combat morale bonus are ignored in the calculated estimates, and so the experienced player should expect slightly different results.

[MJY]

Thanks again, Pzgndr, for the quick response. But to what user manual are you referring? Page 40 of the manual that came with my disc refers to plunder and convoy raiders and the FOW section of my manual makes no reference at all to the expected loss indicators.

I assume this means there is a 2nd edition of the user manual out there somewhere? On-line I presume?

[jon_j_rambo]

You gotta take a college course to learn this game!

[Bromley]

p. 36 of the manual.

[jon_j_rambo]

I never went to class or bought books.

[Guest Mike]

I have no class and I'm often bought to book.

[pzgndr]

I assume this means there is a 2nd edition of the user manual out there somewhere? On-line I presume?

There was a pdf update to the manual provided with the v1.05 patch. Check your SC2 directory.

[MJY]

Thanks to all who responded. I'm now going to close this topic/thread. When Normal Dude completes his statistical study w/o FOW I assume or, even more interesting, to demonstrate the statistical impact of FOW -- now there's a study -- I look forward to seeing a new thread.

Thanks again!

[MJY]

P.S. Ooops. I guess I can't close a thread I started. I'm such a Luddite!

[jon_j_rambo]

I used the free text books at the library, or borrowed off others. Just couldn't afford or justify $50+ for a one term book. College was easy enough...finished with GPA 3.4.