...
However, if a built-in scorecard is used, it is the TurnKey Lender who keeps it efficient. To do so, we are using an AI-based algorithm that is trained to improve the scoring efficiency on an ongoing basis. Let's take a closer look at each of the stages.
Assign Risk Score to Customer Data
Scorecard management is available in the System → Decision engine → Scorecards section. When creating a custom scorecard, you can manually assign scores to specific details.
...
In the Custom card, if the value is numerical, you can add/remove ranges (“categories”) for each piece of customer information and assign a specific value to each of them. If the value is one of many, the card will include all options available for your company.
Some of the common customer details used in scoring include, for example, number of dependents, education, company size, job title and employment duration, property in possession, credit bureau score, number of previously fully paid loans, etc.
Total Risk Score and Risk Segments
Now, when a loan application is processed, the system will calculate and add up all the risk scores (defined manually or specified in the built-in card). After that, the system looks at the score, and defines what risk level it corresponds to.
...
These details are managed in the “Risk segments” tab. There are five risk segments allocated by the risk level from low to highest + default. If the client decides to use a custom scorecard, they must specify score ranges for each risk level. They can also define the probability of dafult, expected distribution, note-down the odds and recommendations, and define what should happen to the loans withing this segment (automatically approved, automatically rejected or sent for further manual review). As it’s been said before, for customer cards, these values lie solely on the client.
AI Magic of the Built-In Cards
For built-in card, TurnKey Lender runs analysis of the scoring success (big data analysis) and the is continuously improving the risk segment allocation and its details, as well as, in case of need, notifying TurnKey Lender team about the need to create additional categories to improve the scoring capabilities.
...
When using this built-in card, the clients can modify the recommendations and further actions on the application (auto-approval, auto-reject or manual review). However, probability of default, expected distribution and odds are calculated based on the existing data analysis (cannot be edited by a client if built-in card is used) and are updated as the system conciders previously scored customers.
Further Data Analysis
As the system analyzes results of the scoring, they become available in the Scoring reports. These help you see how well the built-in card is doing its job or assess the quality of your custom scorecard and improve it if necessary.
...
Conclusion
This way, TurnKey Lender clients can undertake to use their own cards and improve them on their own (using Scoring Reports as a reference point) or use our built-in scorecard, that is not visible to the clients (due to copy-righted information safery concerns related to the algorirthms used to improve it) but is updated by TurnKey Lender and is continuously improved using machine-learning AI algorithm.
...