本文介绍了Score Card的概念，包括客户违约的概率（PD）和非违约概率，以及赔率与得分的关系。转换步骤中，当赔率翻倍时，得分增加PDO（赔率翻倍点数）。通过设定初始赔率和得分，可以计算出A和B的值。举例说明，当赔率为1/60时，得分为600，赔率翻倍到1/30时，得分降低20点。得分公式为：score = 481.89 - 28.85 * ln(odds)。
Score Card Definition (Part 1)
Prob of customer default (PD):
p
(
Y
=
1
∣
x
)
=
p
p(Y=1|x) = p
p(Y=1∣x)=p
Prob of customer not default:
p
(
Y
=
0
∣
x
)
=
1
−
p
p(Y=0|x) = 1-p
p(Y=0∣x)=1−p
Odds, bad/good:
o
d
d
s
=
p
1
−
p
odds = \frac{p}{1-p}
odds=1−pp
Formula for Prob to Score can be expressed as:
s
c
o
r
e
=
A
+
B
∗
l
n
(
o
d
d
s
)
score = A+B*ln(odds)
score=A+B∗ln(odds)
Converting Steps：
set
p
0
p_{0}
p0 when
o
d
d
s
=
Θ
0
odds =\Theta _{0}
odds=Θ0
when Odds double, number of points increase PDO（point of double odds）
formula is:
{
p
0
=
A
+
B
l
n
(
Θ
0
)
p
0
+
P
D
O
=
A
+
B
l
n
(
2
Θ
0
)
\left\{\begin{matrix} p_{0}=A+Bln(\Theta _{0}) \\ p_{0}+PDO=A+Bln(2\Theta _{0}) \end{matrix}\right.
{p0=A+Bln(Θ0)p0+PDO=A+Bln(2Θ0)
Calculate A, B
{
B
=
P
D
O
l
n
(
2
)
A
=
p
0
−
B
l
n
(
Θ
0
)
)
\left\{\begin{matrix} B=\frac{PDO}{ln(2)}\\ A=p_{0}-Bln(\Theta _{0})) \end{matrix}\right.
{B=ln(2)PDOA=p0−Bln(Θ0))
Example
Set
Θ
0
=
1
/
60
，
p
0
=
600
，
P
D
O
=
−
20
\Theta _{0}=1/60，p_{0}=600，PDO=-20
Θ0=1/60，p0=600，PDO=−20
which means: when Odds = 1/60, score is 600, and when odds doubled, 1/30 in this case, score increased by -20 points (higher the odds, more the bad, lower the score)
{
B
=
−
20
l
n
(
2
)
A
=
600
−
B
l
n
(
1
/
60
)
)
\left\{\begin{matrix} B=\frac{-20}{ln(2)}\\ A=600-Bln(1/60)) \end{matrix}\right.
{B=ln(2)−20A=600−Bln(1/60))
s
c
o
r
e
=
481.89
−
28.85
∗
l
n
(
o
d
d
s
)
score = 481.89-28.85*ln(odds)
score=481.89−28.85∗ln(odds)
Relationship between Score and odds
| odds | score | PD |
|---|---|---|
| 1:480 | 660 | 0.2079% |
| 1:240 | 640 | 0.4149% |
| 1:120 | 620 | 0.8264% |
| 1:60 | 600 | 1.6393% |
| 1:30 | 580 | 3.2258% |
| 1:15 | 560 | 6.2500% |
| 1:7.5 | 540 | 11.7647% |
804
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
