Calculate a credit score’s Gini coefficient in a minute
Enter your score bands with their goods and bads. See the Gini coefficient, AUC and ROC curve.
Gini coefficient
Across 28,900 goods and 1,475 bads, the Gini is 68.5% and the AUC is 0.842. The riskiest 20% of cases hold 69.9% of the bads.
ROC curve
The curve runs from the riskiest band. The further it bows above the dashed diagonal, the higher the Gini.
Band results
| Score band | Goods | Bads | Bad rate | Cumulative bads | Cumulative goods |
|---|---|---|---|---|---|
| <520 | 800 | 420 | 34.4% | 28.5% | 2.8% |
| [520,560) | 1,300 | 310 | 19.3% | 49.5% | 7.3% |
| [560,600) | 1,900 | 230 | 10.8% | 65.1% | 13.8% |
| [600,640) | 2,500 | 170 | 6.4% | 76.6% | 22.5% |
| [640,680) | 3,000 | 120 | 3.8% | 84.7% | 32.9% |
| [680,720) | 3,400 | 85 | 2.4% | 90.5% | 44.6% |
| [720,760) | 3,700 | 60 | 1.6% | 94.6% | 57.4% |
| [760,800) | 3,900 | 40 | 1.0% | 97.3% | 70.9% |
| [800,840) | 4,100 | 25 | 0.6% | 99.0% | 85.1% |
| >=840 | 4,300 | 15 | 0.3% | 100.0% | 100.0% |
| Total | 28,900 | 1,475 | 4.9% |
Get the result by email
Next, put the scorecard inside the credit decision. Set its bands in a table and test each change on past applications. Start free
See the formulas
AUC = Σ (Bads ÷ All bads) × (Goods in safer bands + ½ Goods) ÷ All goods Gini = 2 × AUC − 1 Bads in riskiest 20% = Bads in the riskiest bands covering 20% of cases, as a share of all bads
Ties within a score band count half, so finer bands can produce a slightly higher Gini. The Gini measures ranking, not whether predicted bad rates are right.
Calculate a scorecard’s Gini coefficient
The Gini coefficient measures how well a score ranks good from bad. Enter score bands with their goods and bads to calculate the Gini coefficient and AUC, and view the ROC curve.
The Gini shows how well a score ranks, which helps you assess a scorecard. It does not tell you whether predicted bad rates are right, and the cut-off and policy are separate decisions.
Gini coefficient FAQs
How do you calculate Gini from score bands?
Enter each score band with its goods and bads. The calculator works out AUC from the bands and uses it to calculate the Gini coefficient.
What is the link between AUC and Gini?
AUC describes the chance that a random good scores better than a random bad, with ties counted half. The Gini coefficient is calculated from AUC.
What Gini should a credit scorecard have?
There is no Gini value that is right for every scorecard. What a value means depends on the product and the data.
How is Gini different from KS?
This calculator works out Gini from score bands and shows AUC and the ROC curve. Use the KS calculator to calculate KS.
Why does Gini measure ranking, not calibration?
The Gini measures how well a score ranks good from bad. It does not show whether predicted bad rates are right.