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.

Any time we have a player-determined soft score (let's say sports), some players are always going to rate it a little higher or lower, because of their personality. Others will rate everyone really low to gain an advantage.
Problem: Isolate the part of the effect that's actual sportsmanship, so tournament results are not effected by the luck of what kind of raters you face.
The first solution would be to try and account for the influence of the rater in your final scoring tally. But then, the chipmunk strategy becomes to rate the bad players high and the good players low. So, we also have to account for success.

Assume X's sportmanship rating of Y is
R=A(X)+B(X)S(Y)+C(X)P(X,Y)+Noise
Where
S=actual sportsmanship
P=performance in game
A,B,C=determined by psychology.
R & P are known. Our goal is to find S. We can solve this with a regression.

I'm not going to bother with the math here, but do you think that that is enough information to foil chipmunking?

I dunno. Your goal is a noble one, but it seems like more work then most TO's would be willing to shell out.

Plus, it borders on results-rigging.

The first rule of mathematical modelling is to simplify as much as possible.
Therefore instead of using linear regression, which relies on many assumptions and may not be suited to this kind of data, you could just remove sports.
Simple is better than clever.

BillTheManiac wrote:Any time we have a player-determined soft score (let's say sports), some players are always going to rate it a little higher or lower, because of their personality. Others will rate everyone really low to gain an advantage.
Problem: Isolate the part of the effect that's actual sportsmanship, so tournament results are not effected by the luck of what kind of raters you face.
The first solution would be to try and account for the influence of the rater in your final scoring tally. But then, the chipmunk strategy becomes to rate the bad players high and the good players low. So, we also have to account for success.

Assume X's sportmanship rating of Y is
R=A(X)+B(X)S(Y)+C(X)P(X,Y)+Noise
Where
S=actual sportsmanship
P=performance in game
A,B,C=determined by psychology.
R & P are known. Our goal is to find S. We can solve this with a regression.

I'm not going to bother with the math here, but do you think that that is enough information to foil chipmunking?

How are you going to find the variable A B and C (or are they constants?)

Anyway, wouldn't it be easier just to ignore Sports like in European tournaments?

Da Boss wrote:
Simple is better than clever.

He does have a point

A, B, C are optimized so as to minimize noise. I could show you the mathematics.

The underlying mathematics can be placed into an Excel spreadsheet. So all you have to do is enter the numbers.
I thought the rule for mathematical modeling was as realistic as possible while making it simple enough to evaluate.

So does everyone here hate soft scores? Or just the people who posted in this thread?

how is this results rigging? I don't get it.

Hate is a strong word.

The issue I have with soft scores is that they are at least partly subjective, yet can be liable to be 'gamed'.

I am dubious that you can teach or enforce sportsmanship by scoring it -- whether that's done by the opponent or according to some tick box list.

Comp is obsolete. 5e rules and codexes more or less force players to take comp-friendly armies.

Painting is an important part of the hobby. It needs consistent judging to make it work. There is the issue of pro painted armies.

Math-modelling is nice for some things
The problem here is that Sportsmanship IS subjective.

An alternate model for comparrison sake would be a Modern Emergency Room. Been to one recently? They ask you to rate your pain from 1-5, 1= mild discomfort, 5= worst pain you ever felt.

For some, a compound fracture of the arm may "seem" only mildly uncomfortable (endorphines and such working overtime.)

For others, A splinter in the finger could be the agonizing.

It is left to the individual being asked to define the actual parameters of the experience.....a.k.a.
Subjective.

Soft scoring could still be left in as a factor in scoring, but its effect on the actual score should be lessened.

I also agree that modelling and painting should be judged more consistantly.... including the use of appropriate bases for models or even having models on the bases to begin with.

I think sports and painting should be seperate prizes.
Both are important but including them in tournament stuff clouds the generalship scores.

Also, there are a lot of problems with excel as far as advanced statistics go. It's not really designed for it. You'd be better off using something more geared towards modelling and stats.

Regression as a technique adds an addtional layer of complexity and may thow up artifacts in your results which have nothing to do with your input. It's generally best to avoid it where ever possible and rely on more direct processes.

All of this said, this could be a fun project and far be it from me to discourage you from messing around with stats. It's just always interesting to discuss the strengths and weaknesses of techniques. As the go, regression is pretty widely applicable, but it has a raft of issues that mean it would be better to avoid it if possible.

I really like this idea. Something similar is used in biostatistics, for gene-expression (the background correction). You could fairly easily fit the model and consider a histogram of the errors to check normality. This doesn't deal with the whole problem however; some people give out all 10s and other people actually grade, which is a problem that this modeling approach doesn't solve.

I'm with Da Boss. Play a tournament to win. Score the tournament on data, not opinions. Have best painting and sports prizes seperatly from the torunament points.

I know that I think it is much more fun to participate in an event when there are multipel ways to compete, either by winning the event, bringing the coolest looking figs, or just being that guy who everyone liked to play regardless of win/lose. This kind of approach rewards all kinds of gamers.

I also enjoy events that track and reward both battle points and win/loss points. I remember coming in 4th or 5th during an event but won a "butcher" award for having the most points gained from objectives and victory points.