Key Takeaways
- Fannie Mae's July 1, 2026 data drop is aggregate, historical loan-level data for lenders and investors, not a consumer tool, so no borrower can run a personal file through it.
- Announcement SEL-2026-04 allows current use of VantageScore 4.0 and future, phased use of FICO Score 10T; 10T is not live at every closing, and the Classic FICO tri-merge is still standard in 2026.
- FICO 10T is trended. It reads roughly 24 months of balance direction, so a file with rising balances may, directionally, score lower than under Classic FICO.
- The variable you can actually control is your own balance slope; the scoring model your lender pulls in a transition year is something to ask about, not assume.
The July 1 X-Ray of a File Like Mine
On July 1, 2026, Fannie Mae did something quietly seismic. It published historical credit-score data covering FICO Score 10T on loans it acquired from April 2013 through September 2025, plus additional VantageScore 4.0 data on loans from April 2023 forward. More than twelve years of back-data, spanning thirteen calendar years, dropped on a Tuesday and aimed at lenders and investors rather than at me. I read the announcement the way you would read an X-ray of a body part you have been favoring for months, because my file has a wear pattern.
My card balances have been creeping up. Not dramatically, not into missed-payment territory, but month over month, the trend line points the wrong way. And the models the mortgage world is bracing for happen to read exactly that: the trend, not just the snapshot. What follows is an autopsy of a rising-balance file like mine, performed against what the July data actually says, and, just as important, what it does not. I cannot run my own numbers through Fannie's dataset. Neither can you. But I can read the wear pattern honestly.
What Fannie Actually Published, and What It Isn't
Start with the record, because the internet already got it wrong. What Fannie Mae released is aggregate, loan-level historical data, a research file for market participants. It shows how FICO Score 10T and VantageScore 4.0 would have looked across pools of loans the enterprise already acquired. It is emphatically not a consumer tool. There is no portal where I punch in my details and watch my 10T score render. A borrower cannot run a personal file through it, and any post promising you can is selling a fantasy.
The Real News Is SEL-2026-04, Not the Data Drop
The data drop grabbed headlines, but the operational news sits in a Selling Guide announcement, SEL-2026-04. It permits current use of VantageScore 4.0 and future use of FICO Score 10T for loans delivered to Fannie Mae. Read that sentence twice, because the two halves are not the same. VantageScore 4.0 is usable now. FICO 10T is approved but on a future, phased footing. The change is effective immediately on paper, yet it rolls out through a limited lender rollout first, so the plumbing and the operational readiness can catch up before broad availability.
One model now, one model phased
SEL-2026-04 allows current use of VantageScore 4.0 and future use of FICO 10T for loans delivered to Fannie Mae. 10T is coming, not everywhere yet, and whether it touches your loan depends on your lender being switched on.
"FICO 10T now scores every mortgage application, so my rising balances just cost me at every closing."
2026 is a transition year. At most closings the Classic FICO tri-merge is still standard, and 10T rolls out lender by lender.
Why It Matters
SEL-2026-04 approves FICO 10T for future, phased use while VantageScore 4.0 is approved for current use. Whether 10T touches your file depends on whether your specific lender has switched it on, which is a question to ask rather than assume.
2026 Is a Transition Year, Not a Switchover
tri-merge
A mortgage credit report that pulls scores and history from all three bureaus, Equifax, Experian, and TransUnion, so underwriting can read across them.
Why a Trended Model Makes a Rising-Balance File Nervous
trended credit data
Scoring input that reads the pattern of your balances and payments over time, roughly 24 months, rather than only the amount owed on a single query date.
How the two models read the same file
| What it reads | Classic FICO | FICO 10T (trended) |
|---|---|---|
| Time window | A snapshot on the query date | Roughly 24 months of history |
| Balance direction | Largely ignored | A core input |
| Two files, same balance today | Score about the same | Can score differently by direction |
| What a rising file signals | Barely registers under thresholds | Reads as heading the wrong way |
This is general, directional mechanics, not a promise to anyone's individual file. As a rule of thumb the industry has floated, a file whose card balances have been rising month over month may score somewhere in the range of ten to thirty points lower under the newer trended models than under Classic FICO. The mirror image is the encouraging half. A file whose derogatory marks are twenty-four-plus months in the past, sitting under a clean recent profile, may actually read more favorably, because the trend is now working for it rather than against it. None of that is certain, and no one can hand you a number for your specific file. It is the shape of the mechanism, and the shape is what you plan around.
My Wear Pattern, on the Table
So let me put my own file on the table, clearly as an illustration and nothing more. My balances are not high in absolute terms. My utilization would pass a snapshot test on most days. But the trend line is the tell. Over the last year and a half, my revolving balances have drifted upward almost every cycle. A slightly bigger statement balance in month two than month one, a little more in month three. Under Classic FICO's snapshot, that barely registers as long as I stay under the thresholds. Under a trended read, it is the whole story: a file being asked, month after month, which way is this person heading, and answering upward.
What the Autopsy Means Before You Apply
Let me be scrupulous about the line between what the July data proves and what I am inferring. The published datasets prove that Fannie now has a deep historical read on 10T and a shorter one on VantageScore 4.0, and that the enterprise is confident enough to approve both, one for current use, one phased. They do not prove anything about my file, or yours, because they are aggregate research data, not a personal scoring engine. Everything I have said about a rising-balance file scoring lower is directional industry mechanics, not a readout from Fannie's numbers. The dataset is a map of the terrain, drawn from loans already made. It is not a mirror. So the autopsy's practical finding is unglamorous. I cannot control which model my lender pulls, but I can control the slope of my own balances, and I have roughly twenty-four months of history that a trended model will read. Bend the curve down, carry it long enough to matter, and stop treating utilization as a single day's ratio.
Disclosure
Credit scores depend on the specific model a lender uses and on your unique credit profile. Scoring outcomes vary by model, by lender policy, and from file to file, and no data drop or rule of thumb can predict a number for any individual borrower. This article is educational and is not a promise of any particular credit score or lending decision.
Action Items
Frequently Asked Questions
1. Can I run my own credit file through Fannie Mae's July 2026 FICO 10T data?
- No. What Fannie Mae published on July 1, 2026 is aggregate, loan-level historical data for lenders, investors, and analysts. It is not a consumer tool, and there is no portal where a borrower enters personal information and receives a 10T score. Any site claiming you can run your own file through it is misrepresenting what the dataset is.
2. Is FICO Score 10T now used at every mortgage closing?
- No. Announcement SEL-2026-04 permits future, phased use of FICO Score 10T for loans delivered to Fannie Mae. The change is effective on paper but rolls out through a limited lender rollout for operational readiness before broad availability. At most 2026 closings, the Classic FICO tri-merge is still standard, so whether 10T touches your loan depends on your lender.
3. What is the difference between VantageScore 4.0 and FICO 10T in 2026?
- Under SEL-2026-04, VantageScore 4.0 is approved for current use on loans delivered to Fannie Mae, while FICO 10T is approved for future, phased use. So in 2026 the newer model most likely to touch a real application is VantageScore 4.0, not 10T. The two are distinct models and do not treat every file the same way.
4. Why would a rising-balance file score lower under the new models?
- FICO 10T is a trended model. It reads roughly twenty-four months of balance and payment direction, not just a single snapshot. As a general, directional rule the industry has floated, a file whose card balances have risen month over month may score somewhere in the range of ten to thirty points lower than under Classic FICO. That is a description of the mechanism, not a promise about any individual file.
5. What does trended credit scoring actually mean?
- A trended model looks at the pattern of your balances over time rather than the amount owed on a single query date. Two files with the identical balance today can score differently if one has been paying down and the other has been climbing. The direction of travel across about two years becomes part of the score.
6. Is Classic FICO going away in 2026?
- Not in 2026. It is a transition year, not a switchover. The Classic FICO tri-merge is still standard at most closings, while VantageScore 4.0 is now approved across Fannie, Freddie, and FHA loans and lenders use the models variously. There is no single national date when the legacy model goes dark.
7. What can I actually do if my file has a rising-balance wear pattern?
- The variable you control is your own balance slope, not which model your lender pulls. Because trended models read roughly twenty-four months of history, the practical step is to bring balances down and carry them lower across enough statement cycles that the direction of travel visibly changes. It is the same file hygiene that reads well under either model.