Creditcrest Score
Each applicant against your model borrower for the loan type, or a plain score out of 1200.
Includes Intent.
Measured against your own book.
A rating against your own model borrower, updated with every loan that closes.
What you do here. Give a statement and a loan request, and get a rating against your own model borrower.
Score a statementTry it end to endTake the tour- You bring
- A statement and a loan request
- You get
- A rating against your own model borrower, every match shown — or a plain score out of 1200
- Holds the line
- It moves with every loan repaid or written off
An invented statement, $1,500 over 6 months
- Creditcrest Score
- 834
- Risk class
- BB
- Not repaid in full
- 6.2%
- Against the model borrower
- +34
- Measured
- 90% of the weight
Risk class BB. 1 in 25 to about 1 in 12 expected not to be repaid in full.
Each match against the Example Lending Model Borrower
| Measure | This applicant | Model borrower | Ahead of |
|---|---|---|---|
| Affordability · weight 30% · ahead of 66% | |||
| Surplus as a share of income | 20.0% | 12.0% | 73% |
| Debt repayments, with this loan, as a share of income | 7.2% | 15.0% | 73% |
| Headroom under the 10% repayment limit | $65.20 | $60.00 | 53% |
| Income stability · weight 20% · ahead of 55% | |||
| Pay credits in the last 90 days | 6 | 6 | 50% |
| How much the main pay moves | 10.0% | 10.0% | 50% |
| Income that is inferred rather than identified | 0.0% | 5.0% | 65% |
| Account conduct · weight 15% · ahead of 50% | |||
| Dishonours in the last 90 days | 0 | 0 | 50% |
| Days overdrawn in the last 90 days | — | 0 | not measured |
| Other credit · weight 15% · ahead of 63% | |||
| Existing debt repayments as a share of income | 0.0% | 8.0% | 76% |
| Small loans, pay advances or collectors on the statement | 0 | 0 | 50% |
| Earlier loans with this lender · weight 10% · not established | |||
| Earlier loans paid late | — | 0 | not measured |
| Dishonours on earlier loans | — | 0 | not measured |
| Most days past due on an earlier loan | — | 0 | not measured |
| Consistency · weight 10% · ahead of 87% | |||
| Declared against verified income | — | 5.0% | not measured |
| Share of the statement read | 100.0% | 85.0% | 87% |
Suggestion
Lend. The amount and term asked for look affordable.
- Amount
- $1,500.00
- Term
- 6 months
- Repayment
- $166.16 fortnightly
- Surplus after
- $1,006.04 a month
Creditcrest Score compares an applicant with the lender’s borrowers who repaid. It is not a credit decision, it declines nobody, and until it is validated on the lender’s own outcomes it is uncalibrated.
On the plain scale the same statement scores 1093 out of 1200, risk class BB.
Score a statement
How it works
- Each lender has a model borrower for each loan type, named for it: “Example Lending Model Borrower”. Its profile is borrowers of that type who repaid over the last 24 months, shared across lenders as figures only. A business loan is never compared with a small loan’s.
- Each measure is a match against it, counting only the direction that means less risk. The matches give the gap to the model borrower: the bet between the two.
- The applicant’s rating is the model borrower’s rating plus that gap. The model borrower starts at 800, so an applicant is no better than it until shown otherwise. 150 points is ten times the odds, and there is no upper or lower limit.
- Each of the lender’s loans that completes or is written off moves its model borrower’s rating by K × (result − expected), so the scale follows the lender’s book.
- During the loan, every repayment, dishonour, catch-up and payout moves the borrower’s rating. Asking for hardship help never does.
The model borrower an invented lender starts with
- Name
- Example Lending Model Borrower
- Rating
- 800
- Repays in full
- 90%
- Risk class
- B
- Built from
- The prior
At 800, a borrower level with the Example Lending Model Borrower for small loans is expected to repay in full 90% of the time. It started at 800 and 90%, a stated judgement, and has moved with 0 of the lender’s loans since.
| Measure | Middle value | Less risk is |
|---|---|---|
| Surplus as a share of income | 12.0% | Higher |
| Debt repayments, with this loan, as a share of income | 15.0% | Lower |
| Headroom under the 10% repayment limit | $60.00 | Higher |
| Pay credits in the last 90 days | 6 | Higher |
| How much the main pay moves | 10.0% | Lower |
| Income that is inferred rather than identified | 5.0% | Lower |
| Dishonours in the last 90 days | 0 | Lower |
| Days overdrawn in the last 90 days | 0 | Lower |
| Existing debt repayments as a share of income | 8.0% | Lower |
| Small loans, pay advances or collectors on the statement | 0 | Lower |
| Earlier loans paid late | 0 | Lower |
| Dishonours on earlier loans | 0 | Lower |
| Most days past due on an earlier loan | 0 | Lower |
| Declared against verified income | 5.0% | Lower |
| Share of the statement read | 85.0% | Higher |
Until there are 50 repaid loans of the type, each measure blends them with this starting judgement: with n loans, theirs carry n/50 of the weight. The lender’s rating moves by 2 × (result − expected) for each of its loans: a write-off at its own strength about −2, a repayment +0.2.
A rating history
The invented statement’s loan, from assessment to payout. Every change is kept, with its day, its reason and the gap to the model borrower.
| Day | What happened | Rating | Risk class | Plain | Against the model borrower |
|---|---|---|---|---|---|
| 1 Oct 2026 | Assessed Compared with the Example Lending Model Borrower at 800. | 834 | BB | 1093 | +34 |
| 15 Oct 2026 | Repaid on time Repayment 1, due 15 Oct 2026, paid on time. Expected 97%. | 834.3 +0.3 | BB | 1095 +2 | +34 |
| 29 Oct 2026 | Repaid on time Repayment 2, due 29 Oct 2026, paid on time. Expected 97%. | 834.6 +0.3 | BB | 1097 +2 | +35 |
| 15 Nov 2026 | Repaid late Repayment 3, due 12 Nov 2026, paid late on 15 Nov 2026. Expected 97%. | 830.7 −3.9 | BB | 1089 −8 | +31 |
| 28 Nov 2026 | Dishonour The payment of $110.79 made on 26 Nov 2026 was dishonoured. Expected 98%. | 826.8 −3.9 | BB | 1079 −10 | +27 |
| 3 Dec 2026 | Missed repayment Repayment 4, due 26 Nov 2026, not paid within 7 days. Expected 97%. | 819.4 −7.4 | BB | 1059 −20 | +19 |
| 10 Dec 2026 | Repaid on time Repayment 5, due 10 Dec 2026, paid on time. Expected 96%. | 819.7 +0.3 | BB | 1061 +2 | +20 |
| 10 Dec 2026 | Arrears cleared Everything overdue was paid on 10 Dec 2026. Expected 57%. | 822.7 +3.0 | BB | 1071 +10 | +23 |
| 31 Dec 2026 | Missed repayment Repayment 6, due 24 Dec 2026, not paid within 7 days. Expected 96%. | 816.0 −6.7 | BB | 1051 −20 | +16 |
| 5 Jan 2027 | Arrears cleared Everything overdue was paid on 5 Jan 2027. Expected 56%. | 818.9 +2.9 | BB | 1061 +10 | +19 |
| 5 Jan 2027 | Paid out early Paid out on 5 Jan 2027, before the last repayment fell due on 1 Apr 2027. Expected 57%. | 820.3 +1.4 | BB | 1066 +5 | +20 |
| 5 Jan 2027 | Loan completed Repaid in full on 5 Jan 2027. Expected 92%. | 821.0 +0.7 | BB | 1076 +10 | +21 |
Then the loan counts in the Example Lending Model Borrower: expected to be repaid 16% of the time and repaid, so its rating moves +1.7 to 801.7. Repaid, but more than 7 days past due at some point: it counts in the rating, not the profile.
What moves a rating
Each event is a match against what a borrower at 800 does. K is 9 for the first repayment and shrinks as matches accumulate. A repayment is late if paid within 7 days of its due day and missed after that; a payout counts once it has stood 7 days.
| Event | Result | Expected at 800 | K | Change at 800 | Plain points |
|---|---|---|---|---|---|
| Repaid on time A repayment paid by its due day. | Win | 95% | 9 | +0.5 | +2 |
| Repaid late A repayment paid within 7 days after its due day. | Draw | 95% | 9 | −4.0 | −8 |
| Missed repayment A repayment still unpaid 7 days after its due day. | Loss | 95% | 9 | −8.5 | −20 |
| Plan payment kept A payment under a hardship arrangement, made as agreed. | Win | 95% | 9 | +0.5 | +2 |
| Dishonour A payment that was dishonoured and reversed. | Loss | 97% | 5 | −4.4 | −10 |
| Arrears cleared Everything overdue paid, after a missed repayment. | Win | 50% | 9 | +4.5 | +10 |
| Paid out early Paid out before the last repayment fell due. | Win | 50% | 5 | +2.3 | +5 |
| Loan completed Repaid in full. | Win | 90% | 9 | +0.9 | +10 |
| Loan written off Principal given up by the lender. | Loss | 90% | 9 | −8.1 | −50 |
| Loan written off with hardship relief Principal given up by the lender as hardship relief. A write-off, counted apart. | Loss | 90% | 9 | −8.1 | −50 |
Nothing inside a hardship arrangement lowers a rating: a repayment due during one is not a match, a plan payment missed is not a match, and one counted before an arrangement was recorded is given back.
The plain score
A lender may turn the comparison off. Each measure is then read against a fixed scale, the same weights give a score out of 1200, and the same events move it by fixed points. It is compared with nobody.
| Measure | Value: share of its marks |
|---|---|
| Surplus as a share of income | 0.0%: none · 10.0%: half · 25.0%: all |
| Debt repayments, with this loan, as a share of income | 5.0%: all · 20.0%: half · 40.0%: none |
| Headroom under the 10% repayment limit | $0.00: none · $50.00: half · $200.00: all |
| Pay credits in the last 90 days | 0: none · 3: half · 6: all |
| How much the main pay moves | 5.0%: all · 20.0%: half · 50.0%: none |
| Income that is inferred rather than identified | 0.0%: all · 25.0%: half · 60.0%: none |
| Dishonours in the last 90 days | 0: all · 1: 40% · 3: 10% · 5: none |
| Days overdrawn in the last 90 days | 0: all · 5: half · 20: none |
| Existing debt repayments as a share of income | 0.0%: all · 15.0%: half · 40.0%: none |
| Small loans, pay advances or collectors on the statement | 0: all · 1: 40% · 2: 10% · 3: none |
| Earlier loans paid late | 0: all · 1: 30% · 2: none |
| Dishonours on earlier loans | 0: all · 1: half · 3: none |
| Most days past due on an earlier loan | 0: all · 14: half · 60: none |
| Declared against verified income | 5.0%: all · 20.0%: half · 50.0%: none |
| Share of the statement read | 50.0%: none · 75.0%: half · 95.0%: all |
Risk classes
Each rating is a chance the loan is not repaid in full, and each class a range of that chance. A typical small loan borrower, repaying in full 9 times in 10, is B.
| Class | Not repaid in full | In words | Rating | Plain points |
|---|---|---|---|---|
| AAA | under 0.5% | fewer than 1 in 200 | 1002 or more | 1185+ |
| AA | 0.5% to 1% | 1 in 200 to 1 in 100 | 957–1001 | 1160–1184 |
| A | 1% to 2% | 1 in 100 to 1 in 50 | 911–956 | 1130–1159 |
| BBB | 2% to 4% | 1 in 50 to 1 in 25 | 864–910 | 1100–1129 |
| BB | 4% to 8% | 1 in 25 to about 1 in 12 | 816–863 | 900–1099 |
| B | 8% to 15% | about 1 in 12 to about 1 in 7 | 770–815 | 670–899 |
| CCC | 15% to 30% | about 1 in 7 to about 3 in 10 | 713–769 | 410–669 |
| CC | 30% or more | about 3 in 10 or more | below 713 | below 410 |
| D | In default (90 or more days past due) or written off. A loan’s class, never an applicant’s. | |||
Notches (+ and −) are shown once the score is validated on the lender’s own outcomes. Before that they would be false precision. On the plain scale a class is fixed points out of 1200, not a chance of default. These are the lender’s own risk classes, not a credit rating.
Model card
| Status | Uncalibrated. It is validated once the lender has 100 repaid and 30 defaulted loans. Until then it runs in shadow beside the lender’s policy. |
|---|---|
| Model borrower | A starting judgement, not data: the middle value and spread each measure is expected to have among borrowers who repay. Replaced by repaid borrowers of the same loan type, shared across lenders, once there are enough. |
| Starting judgements | A borrower at 800 repays in full 90% of the time, pays 95% of repayments on time and has 3% of debits dishonoured. |
| Components | Affordability 30% · Income stability 20% · Account conduct 15% · Other credit 15% · Earlier loans with this lender 10% · Consistency 10% |
| Never used | age beyond the 18+ eligibility check; sex or gender; postcode or address; whether income is a government benefit; a hardship arrangement or request; vulnerability; answers to the gambling questions; character inferred from where someone shops. |
| Validation | AUC, Gini, KS and the default rate in each risk class against the range it stands for, on the lender’s own outcomes, with population stability. |
| Decides | Nothing. A lender’s policy may refer on it; it never declines on its own. |
For a program
POST /v1/score scores a statement against the model borrower you send, or on the plain scale; POST /v1/score/model-borrower counts completed and written-off loans in it; POST /v1/score/history reads a loan’s ledger into its rating history. You keep the model borrower and the histories; nothing is stored here.