Stock market combination characteristic based network loan risk quantitative analysis method
Stock market combination characteristic based network loan risk quantitative analysis method
 CN 104,978,689 A
 Filed: 06/19/2015
 Published: 10/14/2015
 Est. Priority Date: 06/19/2015
 Status: Active Application
First Claim
1. based on a network credit risk quantitative analysis method for securities market assemblage characteristic, it is characterized in that, borrow the threat probability values p of mark r according to following formula computing platform p _{r}with the risk safe probability value p of platform p _{p}:
 p _{r}＝
q×
p _{f2}+(1q)×
p _{s2}；
p _{p}＝
w _{1}×
p _{b1}+w _{2}×
p _{b2}+…
w _{h}×
p _{bh}；
P _{r}be worth larger expression and borrow mark for certain, the side'"'"'s of loaning bill repaying ability is stronger;
The span of q is [0.55,0.9];
p _{f2}=p _{f}×
m;
p _{s2}=p _{s}×
m;
w _{1}, w _{2}w _{h}represent weight;
wherein, the unit by means of standard gold volume is unit;
p _{f}obtain according to carrying out t check analysis to the history annual report data of the listed company borrowing industry belonging to mark, the level of significance of described t inspection is 0.05, the corresponding former probable value of the t statistic calculated in t inspection, if former probable value is greater than abovementioned level of significance, then test of hypothesis is passed through, and now former probable value is converted to risk safe probability value;
Chinese PRB Reexamination
Abstract
The invention discloses a stock market combination characteristic based network loan risk quantitative analysis method. The risk probability value pr of a borrowing mark and the risk safety probability pp of a platform are calculated according to the following expressions: pr is equal to q*pf2+(1q)*ps2, and pp is equal to w1*pb1+w2*pb2+...+wh*pbn, wherein the value range of q is [0.55, 0.9], w1, w2...wh are weight, the higher the value of pr is and the better the loan repayment ability is, pf2 is equal to pf*m, ps2 is equal to ps*m, and h is the number of industries of borrowing marks of all inspected internet debit and credit platforms. According to the analysis method, the risk of borrowing rate can be calculated according to public information of the borrowing mark and the quantized risk safety probability value is given; and the analysis method is accurate and effective.

1 Citation
Loan data processing method, device, computer equipment and storage medium  
Patent #
CN 108,182,633 A
Filed 01/30/2018

Current Assignee

No References
2 Claims

1. based on a network credit risk quantitative analysis method for securities market assemblage characteristic, it is characterized in that, borrow the threat probability values p of mark r according to following formula computing platform p _{r}with the risk safe probability value p of platform p _{p}:

p _{r}＝
q×
p _{f2}+(1q)×
p _{s2}；
p _{p}＝
w _{1}×
p _{b1}+w _{2}×
p _{b2}+…
w _{h}×
p _{bh}；
P _{r}be worth larger expression and borrow mark for certain, the side'"'"'s of loaning bill repaying ability is stronger;
The span of q is [0.55,0.9];
p _{f2}=p _{f}×
m;
p _{s2}=p _{s}×
m;
w _{1}, w _{2}w _{h}represent weight;
wherein, the unit by means of standard gold volume is unit;
p _{f}obtain according to carrying out t check analysis to the history annual report data of the listed company borrowing industry belonging to mark, the level of significance of described t inspection is 0.05, the corresponding former probable value of the t statistic calculated in t inspection, if former probable value is greater than abovementioned level of significance, then test of hypothesis is passed through, and now former probable value is converted to risk safe probability value;


2. the network credit risk quantitative analysis method based on securities market assemblage characteristic according to claim 1, is characterized in that, q and weight w _{1}, w _{2}w _{h}obtaining value method as follows:

1) valued space of q is [0.55,0.9], gets a series of q value by steplength 0.025, composition array, namely 0.55,0.575,0.56 ..., 0.9}; 2) from abovementioned array, a q value is got; 3) according to the proper vector v of the risk safe probability value computing platform of platform, v=[p _{b1};
p _{b2};
p _{bh}], as the proper vector of sample;
4) adopt " melting 360 " grading report to determine the rating level of platform, as sample classification value, train with SVM, obtain model and weight w;
For the model obtained, with the proper vector of sample for input, computation model classification value;5) error in classification of computation model, preserves described error in classification, model, weight vectors w and q;
W=[w _{1};
w _{2};
w _{h}];
6) judge whether to have traveled through step
1) in all q values in array;
If so, 7 are entered);
Otherwise, return
2);7) model that selection sort error is minimum, using w and q corresponding for this model as final value; The computing formula of abovementioned error in classification is; k is the number of internet loan platform.

Specification(s)