Measurement Of Risk Analysis Customised With Predicta

Every business enterprise has to take precautions, in order to secure itself in the event of unforeseen circumstances. Risk analytics is one such process that can be used as a precautionary measure, which is also increasingly being opted for.

The latest risk measurement tool to enter the market is Predicta, which has been jointly developed by Cranes and Fractal. In an interview with Biztech2.0, Mohamed Asif, Head, Banking and Financial Services Practice, Cranes and Nirmal Palaparthi, Chief Architect, Fractal spoke at length about Predicta.

What has Predicta to offer in Risk Analytics?

Nirmal – Predicta is a tool that will measure risk in the internal rating square. It’s like a wizard that will take the user through the entire modeling process and is also Basel II compliant.

What makes Predicta Basel II compliant?

Nirmal – The primary requisite of Basel II is an internal rating space approach, statistically measuring risk. Banks need to start assessing and responsibly lending in order to protect themselves from consequences that arise from having a high rate of defaulters.

Asif - Internal data needs to be with banks, thus the complete Predicta process is inbuilt. Currently third party agencies are assisting banks, but Basel II wants them to do this by themselves for more accurate data.

How flexible is this product?

Nirmal – It is very customizable, and there is no need to change the API at all. The custom model is built on the basis of the data. As of now, we have 40 Logical Data Models only for credit cards and loads of space to customize it. It’s based on open architecture and uses collective intelligence.

Did you have a pilot run?

Asif- We haven’t had a pilot run yet, but we are looking out for a Beta customer. The whole modeling process is packaged in one, thus our target companies are those who want to build their own scorecard.

How different is Predicta from Fmetrics?

Nirmal - Fmetrics is a generic scorecard, which is used when one does not have data or means to aggregate the data. Predicta however can build a custom model. Typically a custom model is 30% more accurate than generic model. However, if Fmetrics input is bundled in Predicta, then one can have a base platform so that the model is more robust.

What are the other channels through which this product can be used?

Nirmal - Internet is one really big possibility, but banks may not want to be completely transparent to their branches on the output, as reengineering can make a loophole of leaking the data and a possible security threat. But what can be done is introducing a batch mode processing, which involves sending a couple of applications, which will come in an approved or rejected format from the head office. Alternatively, the Embedded PMML format (Predictive Modeling Markup Language), which is an XML-based language, can be used to securely get the information from Head Office to its branches as well.

Asif - Implementation through SAAS is possible and we will be looking at a number of other channels as well. Retails banks have the capacity for large investments and might opt for a complete enterprise installation; however SMBs are primarily lending oriented, so they can have it the SAAS way.

Why start with BFSI?

Asif - Cranes has identified this as a target segment because of the spends and given our experience in statistics and fractal in analytics it would be a winning combination. Maximum spends in analytics is coming from BFSI, of course Retail and Telecom are spenders and we do have plans for them in the pipeline as well. We will be looking at partnerships, wherein we partner with specific domain experts and develop specific products.

What are the future plans for Predicta?

Nirmal – We haven’t heard of any other product such as Predicta so far. The risk analysis tools that currently exist are mostly generic. We are focusing on the APAC region for now and Predicta will definitely have more customisations in the future versions. We are looking for IP-based Collaboration to ensure business continuity during crisis.

Asif – We are working on improving the usability of the product right now. Predicta supports application score carding, but we intend on including behaviour score carding and collection score carding as well. The focus is on micro finance segment, with generic score cards specifically for micro and new business enterprises.
What’s the total cost of ownership?

Asif - No decisions made on that front yet but will probably be like a license model, however minimum investment would be around $20,000. There is no competitor for Predicta thus can’t quantify on that.