Migration Model Accuracy Report: Tampa, Jacksonville, Charlotte

Our team at CorridorIQ has been building a machine learning model that predicts county-to-county migration across the United States. Our model generates these predictions in real time, before any official data confirms them.

The challenge is that IRS migration data, the closest thing to ground truth, takes two to three years to arrive. By the time it's published, the market has already moved on. Our model was built to close that gap: predict the patterns now, verify them when the IRS catches up.

On March 20, 2026, the IRS released the actual 2022-2023 county-to-county migration data. Our predictions for this same period had already been generated and timestamped in our database before the IRS published anything. As any eager research team would, we ran the comparison immediately across three markets: Tampa, Jacksonville, and Charlotte.

Across all three markets, we correctly identified 52 of 60 top-20 feeder counties, an 87% hit rate. Cook County, Illinois, our #1 predicted feeder into Tampa, was off by one household. We said 492. The IRS says 493.

---

Key Findings

• Across 30,009 evaluated corridors, the model correctly identified 52 of 60 top-20 origin counties for Tampa, Jacksonville, and Charlotte (86.7%), all without seeing any 2022-2023 data during training.
• Cook County, IL was Tampa's #1 interstate feeder. Predicted: 492 households. IRS actual: 493. Error: 0.2%.
• In a separate test, six cities were hidden from the model during training. When we asked it to predict their migration patterns, it averaged 87.5% accuracy.

---

The Test

The model evaluates 200,000 corridor pairs across 1,500-plus destination counties. None of the 2022-2023 IRS data was available when these predictions were generated.

We scored ourselves strictly: for each market, we compared our predicted top 20 feeder counties against the actual top 20 from the IRS data, then counted how many we got right.

---

Tampa's Top Feeder County: Cook County, IL, Off by One Household

Our #1 prediction for Tampa's interstate feeder was Cook County, Illinois. We predicted 492 households. The IRS published 493.

Those 493 households brought $54 million in adjusted gross income to Tampa with an average household income of $109,000. This is the profile of Cook County's outbound population:

In 2022-2023 alone, interstate movers brought $1.72 billion in adjusted gross income into the Tampa market across 19,067 households. Knowing which counties are sending that demand, before the IRS confirms it, is the advantage.

---

Tampa: Predictions vs. Actuals

The full top-10, side by side:

#	County	Predicted	IRS Actual	Error
1	Cook County, IL(Chicago)	492	493	-0.2%
2	Queens County, NY	359	349	+3%
3	Los Angeles County, CA	294	316	-7%
4	Kings County, NY(Brooklyn)	375	312	+20%
5	Harris County, TX(Houston)	360	309	+17%
6	Suffolk County, NY(Long Island)	299	292	+3%
7	San Diego County, CA	311	271	+15%
8	Fulton County, GA(Atlanta)	272	252	+8%
9	Fairfax County, VA(DC Metro)	248	246	+1%
10	Wake County, NC(Raleigh)	172	172	0%

---

Three Markets, Side by Side

Jacksonville: The Georgia Cluster

Jacksonville's #1 interstate feeder was San Diego County. But the more revealing result was next door. Jacksonville sits directly on the Georgia state line, and six Georgia counties ranked in its actual top 20 feeders: four from the greater Atlanta metro area, Savannah on the coast, and Camden County on the Florida border. The model identified all six without being told to look at neighboring states. It learned the pattern from the data.

A third market, Charlotte, North Carolina, came in at 80% accuracy. For example, we predicted 1,259 households would move from York County, South Carolina to Charlotte. The IRS says it was 1,389.

---

Six Cities the Model Had Never Seen

The results above compare predictions we made before the IRS data existed. As a separate test, once the 2022-2023 data was released, we updated the model so it could learn from another year of actual migration patterns. During that update, we held out six new metros entirely: cities the model would never see in training. The goal was to test whether it could predict migration for markets structurally different from Sun Belt destinations. For example, we chose markets such as Boston, Salt Lake City, Philadelphia, and Minneapolis, among others.

Market	Accuracy	Origins Caught
Boston	90%	18 / 20
Salt Lake City	90%	18 / 20
Philadelphia	90%	18 / 20
Columbus	90%	18 / 20
Denver	85%	17 / 20
Minneapolis	80%	16 / 20
Average	87.5%	105 / 120

Four of six hit 90%, for cities the model had never seen.

---

What We Learned

The model's biggest miss was Manhattan. It predicted Manhattan as Tampa's #2 feeder when the actual rank was #12. Still in the top 20, but off on the ranking. We suspect this is a post-COVID artifact: the training window included 2020-2021, when New York experienced a historic outbound surge that has since normalized. The model weighted that momentum signal higher than it should have for the 2022-2023 period.

The updated model, now trained with 2022-2023 data, corrects for this. It's one of the advantages of a system that improves every time new IRS data is released.

---

Why This Matters

Every year, hundreds of thousands of households relocate across state lines. Our hypothesis is that solving the problem of where people are moving right now unlocks tremendous value for real estate professionals. That is why we built our platform to provide proactive intelligence rather than waiting on stale data published years after the fact.

We map the patterns that are happening right now. Our data tells you which counties are sending buyers to your market today, not three years from now. With that intelligence, you can:

• Retarget your ad spend toward the counties that are actually sending buyers to your market.
• Build referral relationships with agents in those origin markets.
• Create content that speaks to the buyer profiles actually showing up.
• Win more listings by showing sellers exactly where buyer demand is coming from.
• Master the outbound markets by understanding the economic forces driving people to move.

This turns guesswork into calculated, real-time intelligence.

In 2022-2023 alone, the IRS confirmed 54,828 interstate households relocated across Tampa, Jacksonville, and Charlotte. This represented $4.7 billion in adjusted gross income and an estimated $175 million in buyer-side commissions.

---

What's Next

Our model has been stress-tested against seven years of IRS data and continues to learn from each new release. What's live on our platform now shows predicted migration patterns for 2026, layered with real-time signals from our nowcasting layer.

Migration corridors shift weekly. Our users are already acting on this data to target the right origin markets before anyone else. If you want to see what your market looks like, we'd love to show you.

---

Data points in this post are sourced from the IRS SOI county-to-county migration data (2022-2023) released March 20, 2026. Our predictions were generated prior to this release and timestamped in our database. Questions about methodology: zgreene@corridor-iq.com.