Forum adverts like this one are shown to any user who is not logged in. Join us by filling out a tiny 3 field form and you will get your own, free, dakka user account which gives a good range of benefits to you: No adverts like this in the forums anymore. Times and dates in your local timezone. Full tracking of what you have read so you can skip to your first unread post, easily see what has changed since you last logged in, and easily see what is new at a glance. Email notifications for threads you want to watch closely. Being a part of the oldest wargaming community on the net. If you are already a member then feel free to login now.

Posted By Stu-Rat on 06/12/2007 8:22 AM

Posted By onlainari on 06/12/2007 6:56 AM
Hmm, can you please explain what you do when looking for a shift in battle/comp scores? I don't understand what you're doing.

Did you try doing your shift analysis on the numbers in the excel sheets?

Here's cancon results directly from the TO:

http://www.wargamerau.com/forum/index.php?showtopic=30309

Sure. I'm merely calculating the number of players that have their rank in battle scores lowered due to their comp scores and working that out as a percentage. I'm not applying any modifiers to that shift however, to calculate how much of a change occurs (either individually or together), nor am I counting cases in which the comp score boosts the battle ranking.

Does that seem right or wrong to you?

That technique is not used by statisticians, so I feel that it's the wrong thing to do. Regressions are used heavily by statisticians, so i feel that it's the right thing to do.

Regressions look at the numbers, not the rank. They plot the points and determine if there's any correlation (do high comp equal low battle, for example).

jaghatai I like what you're saying, it makes sense.

So you're focusing on the difference? Between Battle and Comp, I mean.

If so, how are you calculating the effect?

You really need to do a uni course in regression analysis to understand.

We plot all the comp scores and battle scores. It looks like this:



Once you've plotted the results, you want to find the line of best fit. You need an intercept and a slope for that. You run a regression (what I've done in the excel sheets) and you get your line of best fit.

You then do t-stat on the slope to see if it's significantly different from zero. If it is not significantly different from zero, then we cannot say it is not zero. In other words, it could very well be zero, in which case comp has no effect on battle.

This is a worthy goal, but honestly, it looks like a nightmare from here.  Tiny little voices in my head keep saying things like "You have to account for other independent variables, like quality of player and strength of codex".

I can't imagine coming up with meaningful numbers until you've drug those other independent variables out into the open and shot them in the base of the skull.