Huh, must have remembered wrong. MW would be cool though.

_________________
http://www.troublemakergames.co.uk/
Epic: Hive Development Thread

You're overvaluing the average. If 5 hits wins me the combat, which it often can, I'm a lot more worried about wiffing and rolling below average. The odds of actually hitting the average are pretty poor anyway (22.5% for 2x4+ and 40% for 2+).

Of course, I'm wondering if Dave's math isn't a bit off in other ways. Given that http://onlinestatbook.com/java/binomialProb.html is actually correct I get an 80% chance of at least 5 hits, which is greater than the 74% for 5+ (giving a decrease of 10%, not 17%).

It's something that's been bugging me for a while - I constantly get surprised in assaults when I calculate what should, on average, happen, because it never does. A 2-point swing is very significant in epic assaults, and 2-point swings happens all the time with this math. I often see people use the expected outcome to compare stats between different number of dice, and I think that's too narrow.

Basically, it depends on the situation - the underdog wants swingy stats, as it increases his chance to win the assault while the favorite wants as little variance as possible. The "swinginess" of attack stats and saves should be kept in mind when designing units, not merely the average.

In the case of 'stealers, it means that 2+, while still inferior to 2x4+, is closer than the 17% drop given by the expected value indicates. Unless you're in the habit of "charging uphill" with them or getting them attacked in CC of course.

Yes 2x4+ is more "swingy", but that's as often in your favour as not. It just means that individual engagements have more of a chance of being unexpectedly over-effective or ineffective rather than going exactly to plan. I don't see why that's a bad thing.

_________________
http://www.troublemakergames.co.uk/
Epic: Hive Development Thread

And as stat crunching shows above, for the greater majority of the time the extremes of the extra "swing" isn't actually going to turn up.

And in the mean time, it's still a 17% downgrade in the average number of hits generated, to a unit that is generally thought to be underpowered or balanced.

_________________
Currently doing a plastic scenery kickstarter

If you are setting it up so that your stealers will, on average, win the engagement, more swinginess is to your detriment, because it increases your chance of losing the entire thing and the stealers breaking.

Why is this bad? 1) The stealers are worse at winning assaults than the average indicates 2) it's harder to plan when you are less certain of the outcome.

Unless you're in the habit of sending in suicide assaults in the hope of coming up big on the dice of course. Then swinginess is a good thing.

E&C: It's not 17%. More like 10%. The stats also show that the average is unlikely to happen.

A stealer unit with 2x 4+ generates 1 hit.
A stealer unit with 2+ generates 0.83 hits.

That's 17% less.

_________________
Currently doing a plastic scenery kickstarter

...

On average. The average rarely happens. A formation of 6 2+ stealers will generate an average number of hits 40% of the time, a formation of 6 2x4+ stealers will generate it 22% of the time.

Calculating the expected number of hits is very useful, but using it as the be-all, end-all of effectivness when designing a unit is limited and plain wrong.

A combat example: I wish to assault something. I do some quick calculations and figure out that I should do 4 hits on average, my opponent 2, and I got +2 in modifiers. Win by 4, right? Well, it can actually be more likely that one of these two happen than that we get the "average" result: A) I roll well, he doesn't, I lose no units and wipe his formation out. B) He kills 4, I kill 2 and suddenly I'm looking at an even rolloff. My unit breaks with significant losses.

Of course you can argue that I need a better expected difference than +4, but the point is that HOW I get +4 expected has an impact of what my real chance of success is.

Yes, on average the current genestealers will generate 17% more hits than the new genestealers.

That's a pretty important number, because it's the average, most common result that will occur in games.


So is only taking one side of a bell curve and using that to say that a change is smaller than it really is.

_________________
Currently doing a plastic scenery kickstarter

you're both arguing different sides of the same coin. A better single CC value will give you a lower but more certain no. of hits. A worse double CC value will give you a higher but less certain no. of hits, with a greater variation around the mean. Given that GS have terrible saves and are likely to be carrying BMs (especially given Dave's data above), fewer but more certain FS hits may be more useful in planning an assault than more but less certain hits. This could then increase the overall effectiveness of a formation as they are more likely to win an assault, take fewer casualties and stay combat effective for longer (suicide assaults excepted). However, whether or not that is the case will only come out with testing.

The point is, that given that a unit is favored to win, one end of the bell curve is more valuable than the other.

average =/= most common

66% of all results are NOT the average result.

_________________
We are returned!
http://www.epic-wargaming.de/

Nope, not more likely to win the assault, they're less likely to win the assault. They're just less likely to lose the assault unexpectedly...

_________________
http://www.troublemakergames.co.uk/
Epic: Hive Development Thread

I guess my point is that while 2+ is worse than 2x4+, it's not as big a difference as you'd think if you only look at averages.

And more generally that variance should be considered when statting up units.

Personally I think a higher variance and higher average is a better representation of a horde of teeth and claws flying at you!

Remember that variance goes both ways; they're also more likely to win BIG, which balances out the occasions where they fail.

_________________
http://www.troublemakergames.co.uk/
Epic: Hive Development Thread

So what does 'more certainty' mean?

Check out the probabilities of getting 'x' or more hits from 3 units for the 2 options...


So if the attacker's thought process is "so long as i get at least 2 hits, then it's worthwhile attacking; any extra hits over and above the first two are insignificant", then he gains a 4% 'certainty' bump. That's not much. In return he sacrifices an average 0.17 extra hits per unit.

It would have to be pretty exceptional circumstances for me to choose 1x2+ over 2x4+.

Look what happens for 6 units...
(I've included 1x1+, which against 2x4+, is when variance really does come into play)



(cue someone pointing out a mistake in my reasoning and making me look stupid)