PDA

View Full Version : scd probability of tie



bigdjiver
25th-December-2008, 10:26 AM
The grandkids are at the other gran for Christmas, so we had Christmas with them yesterday. So today I am free to give a geeky oresent to anyone that might want it (which may be nobody)

I have criticised other simulations of the scoring which gave a guesstimate of tie probability. As a rust removal exercise I did my own Excel one.

I filled a cell with =simul(.95, .8, 0) and copied it down, repeated in the next column with slightly different parameters as guesses, representing a different pair..

simul is a VBA function with three parameters, probabilities, each varying from 0 to 1. the degree of perfection of the dancer, the correlation between judges with the strictest one, and a "jitters" factor.

This is very quick and dirty programming.

Function simul(skill As Single, judgement As Single, jitter_coef As Single) As Single

Rem /************************************************** **************************************/
Rem /*
Rem /* The dance is split into 100 time intervals
Rem /* During each interval there is a chance that the dancer will make an error
Rem /* This chance depends on the current skill level
Rem /* skill level is the probability of completing an interval without making a scoring error
Rem /* A scoring error loses one point from the severest judge
Rem /* There is a significant chance that the next severest judge will also deduct a point
Rem /* and a lower chance that the other judges will also deduct a point.
Rem /* this chance is simulated as judgment to the power of judge number.
Rem /* If judgement is .9 then there is a 90% chance that the severest judge will deduct a point
Rem /* and around 65% chance (.9^4)that the most generous will.
Rem /* The judges will be less likely to deduct points the lower the contestant is scoring
Rem /* The contestant is assumed to be judge 5, the least able to detect errors,
Rem /* and the most forgiving.
Rem /* However if the contestant does realise their error they will suffer the 'jitters'
Rem /* and be more likely to make another soon after.
Rem /* Their skill level will drop by a jitter factor, which rapidly fades with time passing.
Rem /* jitter fades by half each subsequent time interval
Rem /* However more mistakes will increase the jitter factor.
Rem /* A contestant may be capable of a perfect score, or of a disasterous dance.
Dim current_skill As Single
Dim I As Integer
Dim j As Integer
Dim score As Integer
Dim jitter As Single
judgement = 0.8
current_skill = skill
jitter = 0
score = 40
jitter_coef = 0.3

For I = 1 To 100
If Rnd() > (current_skill - jitter) Then
score = score - 1
For j = 1 To 3
If Rnd() < (1 - judgement) ^ j Then
If Rnd() < score / 40 Then
score = score - 1
End If
End If
Next
If Rnd() < (1 - judgement) ^ 4 Then
jitter = jitter + jitter_coef
End If
Else
jitter = jitter / 2
End If

Next

simul = score

End Function

under par
25th-December-2008, 11:55 AM
So who is going to WIN ????

bigdjiver
25th-December-2008, 01:00 PM
So who is going to WIN ????I refer you, good sir, to the yuletide spoiler SCD thread, that I stumbled into with 3/4 of my mind upon Christmas.:doh: