| | You are not authorized to post a reply. |
| | Author | Messages | |
E.C. Organizer
galneweinhaw
Commander
2646 Posts
    
 | | 07/17/2006 2:31 AM |
| For your comment.
EDIT:
Some stat analysis to determine whether there is a significant difference between having and not having support.
There is no confirmable advantage of having or not having support over the other.
This does not mean there isn't an advantage, only that there is not a significant enough difference for us to confirm one 19 times out of 20... a statistical subtlety =)
Even for 175pt warband sizes there is no confirmable advantage... tho the sample size is small so time will tell!
details for mathematically inclined and to allow confirmation of my methods =)
Chi-squared test - All Data:<br> Win Loss Totals<br>Observed 49 63 112<br>Expected 56 56 112<br>(|O-E|-0.5)^2 42.25 42.25<br>/T = X^2 0.377 0.377 0.754<br><br>with p=0.05, df=1<br>chi^2 = 3.84<br>0.754 < 3.84 so fail to reject null hypothesis of 50/50 | |
Join the Eternal Skirmish Campaign.
Be who you are and say what you feel, because those who mind don't matter and those who matter don't mind. ~Dr Seuss | |
| E.C. Organizer
galneweinhaw
Commander
2646 Posts
| | 07/17/2006 3:23 AM |
| The minor tweak I would support, which has been recommended before by others, is changing the round down to round up.
... after the test I don't think this is necessary.
| |
Join the Eternal Skirmish Campaign.
Be who you are and say what you feel, because those who mind don't matter and those who matter don't mind. ~Dr Seuss | |
| 
Chozyn
Warrior
213 Posts
| | 07/17/2006 8:07 AM |
| I think the numbers are fairly close and I was involved in a few of those lower level losses, and I can say it wasn't because my opponents were over powered that I lost. The dicebot hates me sometimes... example was against Nirbo this week where I rolled 3-4 crit 1 ones on initiative.
I don't think support is too under powered because the numbers are fairly close. | |
Join the fun of the Eternal Campaign
| |
|
SilentG
Sergeant
616 Posts
| | 07/17/2006 11:13 AM |
| Hmm, the stats seem pretty surprisingly even. I'll throw in my two cents on this, because I don't think these raw stats tell the whole story.
First off, in all the cases where support lost I'd be curious to see if the winner was playing LG. I lost a major support game to LG and it was basically an even match, despite my having +60 points. LG's win rate of like 80% in the last three months needs to be taken into account when considering an issue like support.
Another thing to consider is what units are being taken. 90 Pts of support is meaningless if you bring mediocre to bad units. If you take 90 pts worth of tier one units the only reason for a loss should be a serious dice imbalance (or perhaps a game imbalance - LG). I think a lot of stats get skewed because people don't always take the 'optimal' units. Then when two players who do both take 'optimal' units face off with support their data gets dilluted. The way units are costed (cough - badly) in DDM means that if you aren't taking cost effective units you are essentially wasting support points - which means the data from support wins and losses is really hard to interpret fairly.
The part of support that I think is easiest to change is the activation control advantage. You should only be allowed one additional unit for each level difference between you and your opponent. Maybe extra, unused support points could be used to "enhance" units from your regular band by upgrading them to more expensive units - but they would still be standard units worth VPs. This way no points would be wasted, but you wouldn't get a 4!!! unit advantage on activation everytime. A 4 unit activation advantage should almost on its own guarantee a win. | |
Completed trades: Zarnof, Ehren37, Eric is God, griffrat, cavedweller, Kyrinn, Blade, Humble Minion, Noghri, Ilarue
Reference thread http://www.maxminis.com/forums/topic.asp?TOPIC_ID=9114
Champion of the fiendish, dire milk-cow | |
| E.C. Organizer
galneweinhaw
Commander
2646 Posts
| | 07/17/2006 1:28 PM |
| quote:
Originally posted by SilentG
Hmm, the stats seem pretty surprisingly even. I'll throw in my two cents on this, because I don't think these raw stats tell the whole story.
The nice thing about stats is that alot of your points should be taken care of by the random scheduling.
ie, LG players should be getting support just as often as going against it, and good DDM players will get and go against just as often, same with all of these kind of variables...all on average over the course of 100+ games.
And the more games we play the more reliable the stats should be.
I'm going to do some quick stat analysis to see if the % difference form 50/50 is significant or not. | |
Join the Eternal Skirmish Campaign.
Be who you are and say what you feel, because those who mind don't matter and those who matter don't mind. ~Dr Seuss | |
|
eMpTy Kay
Underboss
1068 Posts
| | 07/17/2006 3:43 PM |
| | I think the point SilentG is making is that LG is currently overpowered in the EC enviroment. So a LG player may be beating someone who had the support, thus skewing the stats to the result of players with support units are losing more than 50% of the time. If that is true, it may be that the Support data would have given close to 50/50 split after such a skewing of the stats were taken into acount. | |
| |
| E.C. Organizer
galneweinhaw
Commander
2646 Posts
| | 07/17/2006 4:17 PM |
| quote:
Originally posted by eMpTy Kay
I think the point SilentG is making is that LG is currently overpowered in the EC enviroment. So a LG player may be beating someone who had the support, thus skewing the stats to the result of players with support units are losing more than 50% of the time. If that is true, it may be that the Support data would have given close to 50/50 split after such a skewing of the stats were taken into acount.
Although this may help us, it would not be accurate (or I think even reasonable =) to assume this.
Since the schedule is random, there should on average be as many LG warlords with support as fighting against it. LG being overpowered is irrenlevant to these stats, since this advantage should also be present when LG plays WITH support.
Again I say that's the beauty of this, all those other factors are averaged out because of the random schedule, allowing us to focus specifically on Spt Troops.
entangled_boson, I have some work for you already......back me up here!!
| |
Join the Eternal Skirmish Campaign.
Be who you are and say what you feel, because those who mind don't matter and those who matter don't mind. ~Dr Seuss | |
|
HailSpork
Sergeant
545 Posts
| | 07/17/2006 5:56 PM |
| | If the level of warlords were randomly determined, that would be true. However, if the LG warlords progress faster than others, it *may* skew the data. If you find out which of those on the chart are LG, *that* could support (or detract) towards the effectiveness of support. | |
CHAMPION OF THE CURRY GOLEM! | |
| E.C. Organizer
galneweinhaw
Commander
2646 Posts
| | 07/17/2006 6:07 PM |
| quote:
Originally posted by HailSpork
If the level of warlords were randomly determined, that would be true. However, if the LG warlords progress faster than others, it *may* skew the data. If you find out which of those on the chart are LG, *that* could support (or detract) towards the effectiveness of support.
But level is independant of whether you will have or fight against support.
EDIT: I should say that this is not necessarily true for very high levels such as ~7+ since there are fewer warlords at this level you may be more likely to play vs support then have it. However little of the data is in this range so it would have minimal effect.
As well, the upper lvls vs lower levels show hardly a difference in results, further showing it has little if any effect. | |
Join the Eternal Skirmish Campaign.
Be who you are and say what you feel, because those who mind don't matter and those who matter don't mind. ~Dr Seuss | |
|
SilentG
Sergeant
616 Posts
| | 07/17/2006 7:16 PM |
| The random scheduling may negate my point about LG, but with so few cases I'd feel better looking at LGs win rate with and without support.
I'm less confident that random scheduling will balance out the impact of "player peculiarities". Player A might try out 80 pts of bad units for support one week and then put in tier one units the next week. It just seems too variable for random scheduling to benefit.
I really think that the bulk of the problem with support is just the flat rate of +4 activations. What are your thoughts on that issue Galnew? | |
Completed trades: Zarnof, Ehren37, Eric is God, griffrat, cavedweller, Kyrinn, Blade, Humble Minion, Noghri, Ilarue
Reference thread http://www.maxminis.com/forums/topic.asp?TOPIC_ID=9114
Champion of the fiendish, dire milk-cow | |
| E.C. Organizer
galneweinhaw
Commander
2646 Posts
| | 07/18/2006 1:32 AM |
| quote:
Originally posted by SilentG
I really think that the bulk of the problem with support is just the flat rate of +4 activations. What are your thoughts on that issue Galnew?
I think the stats show that on average it isn't an issue.
If it was an issue, the result would show players with support winning a significant majority of the games right?
One game or scenario to the next, it may have an impact as many things will, but over the course of the campaign is what is important, and I am happy how things look.
Although I'm not a statistician, I am a mathy kinda guy and I (think I) understand what they mean (still waiting for entangled_boson's confirmation =)
The WHOLE point about looking at statistics like this is so you don't have to worry about irregularities from week to week, and other variables that may effect the outcome of the game, are irrelevant here and allow us to look ONLY at support troops (assuming the sample size is large enough and schedbot doesn't cheat somehow ;)
The reason this works is because Support Troops are an independant variable, they are not correlated with faction, level (except maybe very high ones..but I doubt it), player skill, scenario... anything else.
And if you don't believe me, you can always go back to week 63 and gather the same stats by faction =D (which by the way would be cool and interesting to see...but I've done enough of this already =P | |
Join the Eternal Skirmish Campaign.
Be who you are and say what you feel, because those who mind don't matter and those who matter don't mind. ~Dr Seuss | |
|
Zippy
Underboss
1993 Posts
Whitewater, WI
| | 07/18/2006 1:35 AM |
| | What is the data telling us since V3 was released, or is week 63 just that? | |
There are 10 kinds of people in the world; those who understand binary, and those who don't.
Reference Thread, H/W List, Champion of the Catoblepas | |
| E.C. Organizer
galneweinhaw
Commander
2646 Posts
| | 07/18/2006 1:55 AM |
| quote:
Originally posted by Zippy
What is the data telling us since V3 was released, or is week 63 just that?
Yep 63 was the first week of V3 | |
Join the Eternal Skirmish Campaign.
Be who you are and say what you feel, because those who mind don't matter and those who matter don't mind. ~Dr Seuss | |
| E.C. Organizer
galneweinhaw
Commander
2646 Posts
| | 07/18/2006 2:06 AM |
| quote:
Originally posted by SilentG
I really think that the bulk of the problem with support is just the flat rate of +4 activations. What are your thoughts on that issue Galnew?
I'll add another comment since I was thinking about this on my drive home.
I think I know what you are trying to say. And you are right (if this is what you mean), that someone who knows how to use support has an advantage, whereas those who don't know how to use support would be at a disadvantage, resulting in stats that look like everything is even since the two would average out.
I would argue tho, that this is what we want, because you will face those who suck with support, on average, as much as you will face those who are good with them (if not the stats would be skewed one way)
Unfortunatly there will be times you will go into a match where the odds are stacked against you, be it because of the scenario, your opponent, whatever. But as long as the opposite also happens, and you get the odds stacked FOR you sometimes too, that's all we can really ask for isn't it?
| |
Join the Eternal Skirmish Campaign.
Be who you are and say what you feel, because those who mind don't matter and those who matter don't mind. ~Dr Seuss | |
|
TKort
Sergeant
583 Posts
| | 07/22/2006 10:11 PM |
| Always stacked against me :)
TK.
| |
| |
|
SilentG
Sergeant
616 Posts
| | 07/23/2006 12:42 AM |
| I really don't like the balance you describe Galnew. If I'm facing an opponent who doesn't understand DDM well enough to effectively use support and activation advantages I might as well not be playing. I'm not looking to play games where I lose less than 10% of my forces, but I've gotten a dozen or so of those in the last several months. I've also been on the other side of that fence a few times mostly due to facing substantial support or playing my CE vs a LG opponent.
If I'm playing someone who knows the optimal combinations and the best way to maxmimize the impacts of a serious activation advantage, then odds are they'll be good at everything else too. Without much support this type of game would be great fun - close, tense, and could swing either way depending on good plays and dice rolls. With the massive support numbers that the current formula provides and the flat +4 activations bonus this sort of close, fun game becomes virtually unwinnable.
It sounds like you got the point I was making, but I just don't see playing autowins and virtually impossible wins as a "fun" balance. The most serious point I was trying to pass on was that the easiest way to kind of ratchet down the impact of support would be limiting the activation bonus first. Making it dependant on the level difference is one place to start. I just tossed out one activation per level difference because it sounds effective, but anything that would decrease the bonus acts in general would be good.
quote:
Originally posted by galneweinhaw
quote:
Originally posted by SilentG
I really think that the bulk of the problem with support is just the flat rate of +4 activations. What are your thoughts on that issue Galnew?
I'll add another comment since I was thinking about this on my drive home.
I think I know what you are trying to say. And you are right (if this is what you mean), that someone who knows how to use support has an advantage, whereas those who don't know how to use support would be at a disadvantage, resulting in stats that look like everything is even since the two would average out.
I would argue tho, that this is what we want, because you will face those who suck with support, on average, as much as you will face those who are good with them (if not the stats would be skewed one way)
Unfortunatly there will be times you will go into a match where the odds are stacked against you, be it because of the scenario, your opponent, whatever. But as long as the opposite also happens, and you get the odds stacked FOR you sometimes too, that's all we can really ask for isn't it?
| |
Completed trades: Zarnof, Ehren37, Eric is God, griffrat, cavedweller, Kyrinn, Blade, Humble Minion, Noghri, Ilarue
Reference thread http://www.maxminis.com/forums/topic.asp?TOPIC_ID=9114
Champion of the fiendish, dire milk-cow | |
| E.C. Organizer
galneweinhaw
Commander
2646 Posts
| | 07/23/2006 1:22 AM |
| quote:
Originally posted by SilentG
The most serious point I was trying to pass on was that the easiest way to kind of ratchet down the impact of support would be limiting the activation bonus first.
Unfortunately I must disagree. Currently, we see that those with support are losing more often than winning, so reducing the effectiveness of support would only amplify this.
I think the stats show our current system is good, for what it is trying to do.
In order to correct the problem you are stating, we would need to seperate good players and bad players. Otherwise, doing what you would suggest might make your games vs other good players more even, but it would eliminate any chance of the less skilled players in pulling out an upset if they are using support. | |
Join the Eternal Skirmish Campaign.
Be who you are and say what you feel, because those who mind don't matter and those who matter don't mind. ~Dr Seuss | |
|
| | You are not authorized to post a reply. |
|
| |
ActiveForums 3.7 | You must be signed in to participate in the
games. |