The number that ends most portal arguments

When a Power 4 player enters the transfer portal, the assumption in most conversations is that he lands at another Power 4 program. That outcome is real, but it is not the safe bet the assumption implies. We pulled every portal move by a Power 4-origin player from 2022 through 2025 with a recorded destination, 7,576 of them, and sorted each one by the level it landed at.

  • Back to Power 4: 49.9% (3,777 of 7,576)
  • Down to Group of 5: 33.1% (2,511 of 7,576)
  • Down to FCS: 15.7% (1,186 of 7,576)
  • Below FCS, D2 plus D3 plus JUCO plus NAIA combined: 1.3% (102 of 7,576)

Read the top line again: 49.9%. A Power 4 player who enters the portal lands back at the Power 4 level slightly less often than a coin flip lands heads. It is the single most common outcome, but it is not the default outcome, and just over half the pool moves down a level or further.

One in six ends up at the FCS level

The 15.7% FCS figure is the number most portal coverage skips. Of the 7,576 portal entries in this population, 1,186 left a Power 4 program and landed at an FCS school, roughly one in every six. That is more than eleven times the share that dropped below FCS entirely (102 entries, 1.3%, split across D2 with 44, JUCO with 38, NAIA with 17 and D3 with 3). The FCS landing spot is not a rare edge case in this population, it is the third most common outcome after P4 and G5.

Why this holds regardless of when a reader arrives

Portal activity is not spread evenly across the calendar. It concentrates in windows, and most of a given cycle's moves get made during them. But the underlying split does not reset with each window: it is a four-year population, every portal cycle from 2022 through 2025, not one window's worth of moves. A player, family or staff member reading this at any point in the year is working from the same historical base rate as someone reading during an active window. When the moves happen does not change what the odds look like once a full cycle is counted.

What this means for a player weighing the portal

Three practical reads follow from the split.

First, treat P4-to-P4 as close to even odds, not the expected outcome. Just under half of this population repeated their level. Just over half did not.

Second, an FCS landing spot is common enough that it belongs in the conversation before a player enters the portal, not as a surprise after. One in six entries in this population ended up there.

Third, dropping below FCS entirely is genuinely rare in this population, 1.3%, but it is not zero. For a Power 4 player, the portal's downside case is usually a level or two down, not a fall out of the FBS and FCS pipeline altogether.

How we counted

This piece publishes Recruit Intel research finding RI-RF381. Population: every row in the Recruit Intel portal_transfers table with origin_level equal to P4 and a transfer year between 2022 and 2025, restricted to rows carrying a recorded destination_level. That population is 7,576 portal entries (some players appear more than once, having entered the portal in more than one year in this window; zero duplicate (player_id, transfer_year) rows exist among the 6,574 entries carrying a canonical player match, and a name-level check run across the full population today finds 26 duplicate (player_name, transfer_year) groups, 27 rows, 0.36% of the population, several of which are plausibly distinct players sharing a common name rather than duplicate records). A separate 308 P4-origin entries in the same window carry no recorded destination and are excluded from the percentages above, roughly 3.9% of the full 7,884-entry P4-origin population in the window.

RI-RF381 was originally derived on 2026-08-13. We re-ran the same query live on 2026-09-01 before publishing this piece: the population moved to 7,548 known-destination rows (of 7,856 total P4-origin entries, still 308 unknown) as ordinary roster and dedupe corrections landed. The live shares: P4 3,753 (49.7%), G5 2,509 (33.2%), FCS 1,185 (15.7%), below FCS 101 (1.3%). Every share held within two tenths of a point of the figures reported above, and none of the reported outcomes or their ordering changed.

One caveat carries over from the underlying finding. Conference-level tags (P4, G5, FCS) were not individually spot-checked against mid-window realignment moves, so a program that changed level during 2022 to 2025 could carry an inconsistent tag on a small number of rows. The scale of any such effect has not been quantified and is flagged here rather than smoothed over.