At Titan Funding, we learned FICO scores don't always tell the story. We saw people with strong FICO scores default while their day-to-day cash flow was all over the place. The most useful data for us was real-time cash flow analysis. By looking at recent spending habits, we could spot trouble weeks before it started and call customers early instead of waiting until it was too late.
The best alternative data signal for predictive performance of charge-off before FICO is consistent with cash flow, and not gross income level alone. FICO shows how someone paid in the past; cash-flow data indicates if they can pay now. When inflows go wobbly, balances hover near zero or fixed expenses begin to elbow discretionary spending aside, risk is rising — even if the credit score hasn't budged. That decay appears months before a missed payment ever makes it onto a bureau report. From an investor's standpoint, this is what the appearance of trouble looks like in real estate long before a default. A property doesn't fail because the rent looked good last year; it fails when the operating cash no longer covers debt service, taxes, insurance, and utilities. Fintech loan originators who rely on monitoring real-time cash-flow behavior are effectively underwriting the balance sheet of today rather than yesterday's credit history. And since it has that advantage of timing, you can intervene earlier — change limits, price risk correctly or avoid charge-offs with action before stress becomes irreversible.
I would say looking at the velocity of new financial commitments is something that traditional lenders don't really consider in my experience (apart from perhaps private banks and specialist lenders who can take a broader, more bespoke view), and is a space where modern fintech lenders can improve risk assessments. Mainstream banks have very rigid underwriting processes that don't tend to look at the direction you're trending in - it's more of a snapshot of your current situation (typically past 3 months), and that's it. If fintech lenders are able to analyse a bigger picture, reliably, then they can factor risk more accurately - and recent velocity of new financial commitments is a great example of something to look at. Imagine an individual that has recently cut back from 10 active BNPL commitments to 5 in the past 6 months - on paper, they would look higher risk than someone that's gone from 0 to 4 BNPL commitments in the past 2 months - and yet I know who I would rather lend money to!
Of all the alternate data signals out there, the main signal has been cash-flow volatility of the borrower's primary deposit account specifically due to immediate income disruption or an increase in the difference between inflows and outflows needed to support the borrower's lifestyle. Cash-flow volatility indicates current financial hardship while being FICO is only a lagging measurement and based on previous borrowings. When lenders see many irregular payroll deposits or increased overdraft frequency each month or continued increases of "essential spending", they are often seeing the first signs of default prior to seeing it show up on a credit report. Charge-off's are largely a fnancial liquidity issue, not a moral issue like people do not just stop caring about paying back a loan. A loss of liquidity causes a borrower to not be able to meet their debt obligations. Cash-flow volatility at the account level identifies a lack of liquidity early on and is less likely to be "worked around" than many of the other alternative data signals available.
One of the things that I have realized is that, according to the data, real-time cash flow information is the biggest factor in predicting defaults. While FICO scores show information about the past few months, bank transaction data reveals exactly how much money is in an account right now. By examining daily balances and ongoing bill payments, lenders are able to identify when a borrower is going through a liquidity crisis weeks before the borrower misses the payment. Lenders are now able to manage risks more effectively with better information than they had previously. To me, knowing the actual cash buffer provides lenders with the best gauge of the financial health of a borrower at that moment. It ends the uncertainty of lending and changes it to a much more scientific and accurate process.
A customer's history of utility and telecommunications payments is the best early warning indicator of when they may be experiencing financial difficulties. The reality is most people will choose to pay their cell phone or electricity bill before they would pay their credit card bill. Therefore, if a customer begins making late payments on their utility bills, it is an obvious red flag that they likely will be charged off on a loan very soon afterward. It provides many insights into the borrower's financial discipline on a more frequent basis than that provided by a borrower's monthly FICO score. By monitoring for these minor behavioral changes, lenders can prevent losses from occurring.
Real-time cash flow data represents the optimal predictive signal of charge-offs previously available before FICO scores. FICO is a trailing product based on monthly bureau up-loads whereas transaction-level bank data provides you an "active" window into the borrowers raw cashflow today. Fintechs tap into that signal to scan for early alerts of distress, including too fast a drawdown of cash reserves or a series of overdrafts or choppy patterns in deposits and income. By detecting these behavioural changes as they occur, the technology allows the lenders to predict a new default weeks earlier than when a payment has technically been missed bypassing what can be up to a 30-to-90-day lag in reporting defaults on traditional credit files.
Transaction data on cash flow is the best alternative signal, as it gives us a real-time look at a borrower's financial health, while traditional FICO scores are lagging indicators. Open banking allows fintechs to analyse daily bank account activity, enabling them to spot "pre-delinquency" indicators such as an sudden loss of income or rapid depletion of liquidity weeks ahead of a missed payment being officially reported to credit bureaus. Whereas scoring generally will not adjust until a 30 days delinquent status is achieved, cash flow takes into account immediate changes in the amount of discretionary money and net income. This level of granularity enables lenders to detect financial distress and potential charge-offs much more accurately and quickly than static bureau data.
The most predictive indicator of whether someone will stop paying his or her loan has proven to be their bank transactions. That is what is known as "cash flow data". Because it's real-time money moving. A credit score reflects only what has already occurred. By examining how much of the paycheck remains after bills, lenders can spot money problems and borrowers in trouble early. That helps them discover creditworthy borrowers who have no credit history. It also helps them identify people with high credit scores who are actually beginning to run short on money. This makes making loans much safer and more precise.
In my work with data, I've seen how someone's mobile and social habits can flag risky borrowers before a credit score does. It reminds me of side projects where drops in app engagement always meant users would quit soon after. Lenders could use that same thinking. If you're experimenting, test whether these digital signals can improve your current models, but you really have to watch out for privacy and fairness issues.
It's actually cash-flow volatility that is the best predictor of charge-offs." And unlike a snapshot FICO score, which shows how a borrower has handled credit in the past, real-time banking data would reveal how healthy that person's finances are right this second. Lenders miss such lights flashing in the borrowers' living rooms and "low-balance alerts" on bank accounts showing financial hardship. A sharp uptick in stock-market volatility typically signals financial distress weeks before an actual missed payment. This is important because it captures "life events" that credit bureaus which tend to be slow in reporting such things have not reported yet like loss of job or a medical emergency. By spotting these patterns early, fintechs can provide their own proactive assistance programs. This decreases the probability of outright default, and protects the lender's portfolio more than traditional scores alone.