Where the transfer portal moves players: the 2026 state board

Recruit Intel Research · finding RI-RF393 · 2026 transfer portal cycle · data pulled 2026-09-01, board re-derived 2026-09-05

Through the 2026 transfer portal cycle, 4,227 committed college transfers have both an origin program and a destination program that resolve to a US state. That is 90.1 percent of the 4,691 committed transfers on record, enough to draw a real state by state map of where the portal is actually moving players.

Texas leads every state in net gain: 416 transfers in against 336 out, a net of plus 80. Pennsylvania is the opposite story, the biggest net loser in the country at 104 in against 164 out, a net of minus 60. Those two numbers, one state pulling players in faster than any other, one state losing them faster than any other, are the whole shape of this cycle.

Where the portal is gaining players

Net flow by state, transfers in minus transfers out, among 4,227 state resolved 2026 portal moves. Top five gainers, ranked by net.

Texas and California both sit inside the top three gainers while also running some of the highest gross volume in the country, 752 combined moves for Texas and 391 for California. That is the signature of a big, high turnover recruiting state: a lot of players leave, a larger number arrive, and the state comes out ahead on net without ever looking calm. Tennessee tells a tighter version of the same story at 208 in against 158 out.

Mississippi and Arizona tie for the fourth best net gain in the country at plus 18 apiece, on very different volumes: Mississippi 126 in and 108 out, Arizona a tighter 68 and 50. Neither is trying to be the busiest state in the country. Both are simply keeping more of the players who pass through than they lose.

Where the portal is losing players

Net flow by state, transfers in minus transfers out, among 4,227 state resolved 2026 portal moves. Top five net losers, plus a tie for fifth, ranked by net.

Pennsylvania and North Carolina are not small football states. Between them they account for 316 committed transfers arriving and 432 leaving, well above the volume some net gainers post in both directions combined. What they lack is the offsetting inflow Texas and California get. North Carolina brought in 212 players, a real number, and still finished 56 in the hole because 268 left.

New York is the cleaner case. Fifty five in, ninety out, a smaller total market than Pennsylvania or North Carolina but the same one way pull. Oregon sits alone at fourth, 75 out against 53 in, 41.5 percent more leaving than arriving. Fifth place is a tie: Maryland and Michigan both finished at net minus 16, at very different volumes. Maryland is a small market, 21 in and 37 out, 76.2 percent more leaving than arriving, nearly double. Michigan runs one of the bigger books in the country, 118 in and 134 out, and still comes out a net loser.

The quiet middle

Most states are not on either extreme. Virginia ran 155 transfers in against 152 out, a net of plus 3 on real volume. Louisiana posted 138 in against 132 out, plus 6. Colorado came in at 94 and out at 88, plus 6. Kansas and Nebraska each finished at exactly minus 1. None of that is a headline, and that is the point: a state can process well over a hundred portal moves in a single cycle and still land almost dead even, because its inflow and outflow are roughly matched rather than because nothing is happening.

Where Nevada sits

Nevada, the market Recruit Intel is built around, is a genuine net importer this cycle at the state level: 44 transfers in against 30 out, a net of plus 14. That is not the quiet middle. Nevada ties Hawaii for the seventh best net flow of 49 states, just behind Wisconsin's plus 16 and ahead of Indiana, Arkansas and Kentucky, a real signal for the state's programs, not a flat line. The state total combines both Nevada programs: the Nevada Wolf Pack alone are a modest net importer at 22 in against 17 out, a net of plus 5, and UNLV adds another 22 in against 13 out, a net of plus 9.

How we counted

Source: the portal_transfers table, Recruit Intel's dedicated NCAA transfer portal dataset, fed from structured recruiting data feeds rather than inferred from roster changes. Population: every 2026 cycle row with status committed, which matches the platform's own established definition of a completed move. Signed and enrolled are also recognized completed-move statuses in the schema, but zero of the 4,691 rows in the current 2026 population carry either label. Three other statuses are excluded because they are not completed moves: entered and transfer, both players who are in the portal with no destination program resolved yet, and withdrawn, players who pulled back out. The transfer bucket holds 133 rows. The entered and withdrawn buckets are live inventories that move every time a player commits or pulls out, so this piece does not put a count on them.

Each row's origin program and destination program were joined to their state. A row counts toward a state's inflow if that state is the destination, and toward outflow if that state is the origin, and a row is only counted at all when both ends resolve to a state. A move where both ends sit in the same state counts in both columns for that state, which is standard for this kind of flow accounting: 341 of the 4,227 resolved moves, 8.1 percent, were exactly that, an in state transfer with no net effect on either column.

Coverage: as of 2026-09-01, 4,691 transfers carry a committed status for the 2026 cycle, and 4,227 of them, 90.1 percent, have both ends resolved to a state. The remaining 464 are missing a state on at least one end, most because the destination program has not yet been linked to a specific college record, a small remainder because the program itself does not carry a state on file (non-traditional or junior college programs). That gap can move the exact magnitudes in this piece by a few transfers per state as those links resolve, though the top and bottom of the board have held steady through repeated checks.

This finding was first built on 2026-08-21 and independently re-derived on 2026-09-01 against live data before this piece was allowed to publish. That 2026-09-01 pass also caught and fixed a counting bug: the original method credited a state with a transfer whenever either end of the move resolved to a state, even when the other end did not, which let a small number of partially resolved rows count on one side only. Every number in this piece now counts a transfer for a state only when both the origin and the destination resolve to a state, the same 4,227-row population this piece has always claimed. That fix moved 13 states nationally by one to three transfers each. Among the states named in this piece, Texas, Tennessee, Mississippi, North Carolina and Louisiana each shifted by one or two transfers, and Michigan's corrected total surfaced the tie for fifth on the losers board above. Every other state named here, including Nevada's state total, was unaffected by the fix.

Separately, in the ten days of ongoing portal activity between the 2026-08-21 build and the 2026-09-01 re-derivation, portal_transfers grew from 5,202 to 5,258 total 2026 rows. Of the six states the finding's own audit trail tracked across that gap, net flow moved by zero to three transfers each. Texas was the top gainer both times, Pennsylvania and North Carolina were the two biggest losers both times, and New York's net landed at exactly minus 35 both times.

What this board does not measure: it tracks where college programs are located, not where a player grew up or played high school football, so it answers where programs gained or lost portal talent rather than where players themselves moved from. It is also limited to the college transfer portal. It has no connection to high school transfers, which Recruit Intel tracks separately and under a different set of gates.

Correction, updated 2026-09-05. Two row counts have been cut from the how we counted section above: the count of rows carrying an entered status and the count carrying a withdrawn status. Neither was recorded in RI-RF393, and both are live inventories that change every time a player commits to a destination or pulls back out, so both were removed rather than restated at a value the finding does not carry. The board itself was re-derived from RI-RF393's own method against live data on 2026-09-05, by two independent routes, and registered in the finding. One cell moved: North Carolina now reads 212 transfers in and a net of minus 56, against the 213 in and minus 55 published on 2026-09-01, a difference of one transfer, and the combined Pennsylvania and North Carolina inflow above moves with it. North Carolina is still the second worst net loser on the board. Every other state named in this piece reproduces cell for cell, including Tennessee at 208 in and 158 out, Louisiana at 138 in and 132 out, Virginia, Colorado, Kansas, Nebraska, the Nevada state total and both Nevada programs. That same re-derivation puts Florida at net minus 17, which would place it fifth on the losers board and move the Maryland and Michigan tie to sixth. An earlier version of this note, published 2026-09-04, reported drift in the Tennessee and Louisiana outflow counts and placed Arkansas level with Nevada. It was computed from a faulty read of the portal table, all three claims were wrong, and they have been withdrawn along with the Arkansas edit that note prompted.

Follow @RecruitIntelHQ on X for the research as it publishes.