AI Modeling

It’s a question we get from nearly every client with a credit model in production. Once you’ve launched a model, how often should you revisit it? Is there a fixed schedule you should follow? Or is it only necessary when something breaks? As with many modeling questions, the answer is: it depends. There are some clear principles we use to guide retraining cadence — and some concrete signs that it’s time to act. What we mean by retraining Let’s start by clarifying what we mean

“How many features should I have in my model?” It sounds like a simple question. But like so many modeling decisions—especially in the credit space—the honest answer is: it depends . We often hear this question from internal teams and executives alike. It comes up when a team is building its first in-house model, when it's looking to upgrade an existing scorecard, or even when trying to explain why their current model looks the way it does. At Ensemblex, we've built models fo

Most lenders we work with already have a sizable pile of data—application fields, bank data, bureaus, device signals, platform behavior. When building a credit model, not all data is good data. Some data actually carries long-term risk not just to your model, but to your entire business. Here’s a quick guide to evaluating your data. Here’s a snapshot of what we often see on our first call with a lender: Application data : Self-reported income, employer name, product type, pur

If you already have smart people on your team—and you’re not looking to fully outsource your modeling work—why bring in an outside partner? We’ve partnered with all sorts of modeling teams, from small startups creating their first model to established lenders and banks with mature modeling talent and fully built-out systems. Across all of them, we’ve learned how to make our presence a multiplier, not a crutch. Here’s how. 1. We meet you where you are We don’t sell prebuilt mo

One of the most common questions we hear about credit model modernization is this: “We use a lot of underwriting rules—how can we safely reduce or replace them?” Or more bluntly: “Our model isn’t doing much. Most of the real work is happening in the knockouts.” Many lenders start with a simple rules-based credit policy, planning to grow out of it over time. But shedding the knockouts is easier said than done. What starts as a short-term solution often turns into a long-term l

In short: Yes, we can. But not without you. Potential clients, especially in emerging markets, often ask us how well we understand the peculiarities of their market. It’s a good question. A successful model is built for the context in which it will operate: the unique customer dynamics, regulatory constraints, and data ecosystems. Otherwise, while it might look attractive in testing, it’s likely to cause headaches in production. We’ve now worked on most continents (we’re stil

Do you want to accurately and consistently segment risk, therefore enabling your entire credit strategy? Then yes, you need to monitor your model! We see robust monitoring save our clients real money all the time: A shadow scoring test flagged PSI anomalies arising from a difference in a vendor's data at month-end (a quirk that wasn't visible in the development data set). We were able to make adjustments to the model in production. A live model suddenly received drastically d

"If it ain't broke, don't fix it." Lenders often push back when we suggest exploring a new model build. It's fair—model builds require resources, and it can feel silly to fiddle with an underwriting model that "works," especially if origination volumes are on track and losses seem manageable. But at Ensemblex, we know that "works" often means "leaves money on the table." What Does It Mean for a Model to "Work"? In technical terms, a credit model is effective if it "slopes ris

What Exactly Is a Leaky Variable? A leaky variable is any feature in your training data that contains information you won't have at decision time. The most extreme example would be using default status to predict default. That creates a perfect model in development with zero real-world utility, as the model will simply learn to predict the outcome with itself. Of course, any serious data scientist would catch an error that massive. But leakage can be subtle: Post-application

Gradient Boost Models (GBMs) have become the go-to tool for many credit modelers for good reason. GBMs can unlock meaningful lift in predictive accuracy, helping lenders better distinguish between high- and low-risk applicants, expand safe approvals, and reduce losses. But with great power comes great risk. At Ensemblex, we’ve spent years developing, testing, and monitoring GBM credit models. And while we remain strong advocates for their use in the right context, we also kno