# Real-World Decommit & Follow-Coach Rates (Default Config Research)

Deep-research report (2026-07-14, adversarially verified: 25 claims checked with 3-vote verification, 19 confirmed, 6 refuted) answering: *how often do committed recruits decommit after a coaching change, and how often do they follow the departing coach?* Purpose: evidence-backed **default percentages** for the tool's config system.

## Contents
- [Verified evidence](#verified-evidence)
- [What has NO quantitative evidence](#what-has-no-quantitative-evidence)
- [Refuted claims (do not encode these)](#refuted-claims-do-not-encode-these)
- [Recommended default config values](#recommended-default-config-values)
- [Caveats](#caveats)
- [Sources](#sources)

## Verified evidence

### 1. HC change ≈ 2.5× the odds of decommitting (HIGH confidence — peer-reviewed)
The only peer-reviewed model of FBS decommitment (Bigsby, Ohlmann & Zhao, *Decision Support Systems* 116, 2019; 2016 class, 2,644 recruits via 247Sports data):
- A recent **head-coaching change is the single largest risk factor**: **+148% odds** (odds ratio ≈ 2.48, p < 0.001 in all five models).
- Baseline monthly decommit probability ≈ **5.2%** → ≈ **12%/month** with a recent HC change.
- Baseline all-causes decommit rate: **20.3% of recruits decommit at some point** in a cycle (up from 12.4% in 2012 — and ~20% again in a 247Sports audit of the 2024 P4 class, so stable into the portal era).
- The study encodes **head-coach changes only** — nothing on coordinators or follow-the-coach.

### 2. Mid-season HC firings: ~37.5% short-term decommit rate, huge variance (MEDIUM)
Sept 2025 firings ~2–2.5 months before early signing day (UCLA, Oklahoma State, Arkansas, Virginia Tech): **27 of 72 committed 2026 recruits decommitted within weeks** (UCLA 10, OkState 8, Arkansas 7, VT 2). Tempo: UCLA lost 7 of ~22 (~32%) within **48 hours**. This is an acute, in-season-firing scenario — treat as an **upper bound**.

### 3. Distance from signing day is the strongest modifier (MEDIUM)
- **Two cycles out:** Penn State's entire 2027 class (incl. a 5-star) decommitted **within ~1 hour** of Franklin's firing — **100%**.
- **One cycle out:** Oklahoma's 2023 class lost **~43% in one day** after Riley left for USC.
- **Current cycle (weeks from signing):** Oklahoma's 2022 class lost **~17% in two days** (more followed by signing day).
- **Floor is real:** Florida had **0 decommits for ~6 weeks** after Napier's Oct 2025 firing. → Roll per-recruit probabilities; never force a fixed exodus.

### 4. Timing: risk RISES toward signing day (HIGH)
Decommit odds increase ~1%/day closer to signing day; 71% of decommits happen in the final six months, 25% in the last month (late poaching). Hazard ramps up until signing, then drops to zero.

### 5. Follow-the-coach: real but a small share (MEDIUM — single case study)
Riley→USC is the best-documented case: he contacted 5-star ex-commit Malachi Nelson within a day (Nelson committed to USC two days after decommitting; later signed). But USC's first early-signing class had **7 signees, of which exactly 1 (~14%) was a former Oklahoma commit**. Follow-coach conversion concentrates in **elite recruits the coach personally recruited**, and only when the new school is a peer-or-better destination.

## What has NO quantitative evidence
Set these heuristically and expose as user-tunable (they are already in our config design):
1. **OC/DC or position-coach change** decommit rates (no study or media count exists).
2. **Primary recruiter departure** effect (asserted often, never quantified).
3. **Re-sign-with-same-school** rate after a change-driven decommit.
4. **Star-rating gradient** — claims that 5-stars bail first/fastest were *refuted* in verification.

## Refuted claims (do not encode these)
- "Elite recruits decommit first/fastest after an HC change" — 0-3 / 1-2 votes, contradicted by the sources themselves.
- "Notre Dame's internal promotion (Freeman) held decommits to ~9%" — failed verification; the low-end anchor for smooth transitions is soft.
- Specific Penn State 2026-class decommit counts (7, or 10, in ~2 weeks) — sources disagree; don't cite a precise number.

## Recommended default config values

| Setting | Default | Evidence anchor |
|---|---|---|
| HC fired/departed → decommit chance per committed recruit | **30%** | Between the 2.5× academic multiplier and the 37.5% acute-firing aggregate; carousel-timing in-game ≈ 1–2 months pre-signing |
| HC change, smooth transition variant (e.g. promoted coordinator), if we can detect it | **12%** | Direction supported, exact number heuristic (refuted-claim caveat) |
| OC change → decommit chance, **offensive** recruits only | **10%** | No data — heuristic ⅓ of HC effect |
| DC change → decommit chance, **defensive** recruits only | **10%** | No data — heuristic ⅓ of HC effect |
| HC + coordinator both changed | use HC rate (don't stack) | No data |
| Follow-coach chance (among recruits who decommitted because of that coach's departure, new school meets prestige/space criteria) | **15%** | ~14% share in Riley→USC case |
| Follow-coach bonus for high-interest recruits (departing school was #1 on their board / gem-tier interest) | **+25%** (→ ~40%) | Elite-target concentration in Riley case |
| Follow-coach when new school is lower prestige | **~0–5%** | Direction only |
| Decommitted recruits keep old school in top list | always on (mechanic, not probability) | Re-recruiting is documented (Nelson/USC pattern); rate unquantified |
| Per-recruit random roll (not class-wide quota) | always | Florida 0% / PSU 100% variance |

All of these belong in the user-editable settings UI with these as shipped defaults.

## Caveats
- The peer-reviewed baseline predates the early signing period (2017), portal (2018), and NIL — likely conservative for the modern era.
- The 2025 case studies are acute *mid-season firings*, an upper-bound scenario vs. orderly December transitions; the in-game carousel is an offseason event, so the true analog sits between the two.
- Nearly all case-study counts come from sports media snapshots (counts grew afterward; denominators sometimes inferred; Arkansas confounded by earlier defensive-staff firings).
- Follow-coach rests on one well-documented case (Riley→USC).

## Sources
Primary: [Bigsby/Ohlmann/Zhao, DSS 2019 (publisher)](https://www.sciencedirect.com/science/article/abs/pii/S016792361830160X) · [full PDF](https://www.biz.uiowa.edu/faculty/kangzhao/pub/DSS_2019_B.pdf) · [SI Commitment Project 2011](https://www.si.com/more-sports/2011/08/05/commitment-trends)
Case studies: [CBS Sports Oct 2025 (4-school aggregate)](https://www.cbssports.com/college-football/news/college-footballnewscollege-football-recruiting-coaching-changes-nil-deals/) · [ESPN Sept 2025 fired-coach FAQ](https://www.espn.com/college-football/story/_/id/46295312/college-football-coach-fired-september-happens-roster-faq) · [ESPN PSU 2027 collapse](https://www.espn.com/college-sports/recruiting/football/story/_/id/46574278/high-profile-27-recruits-decommit-penn-state-firing) · [247Sports PSU tracker](https://247sports.com/college/penn-state/article/tracking-penn-states-recruiting-and-roster-activity-following-the-firing-of-head-coach-james-franklin-256663803/) · [ESPN OU post-Riley](https://www.espn.com/college-football/story/_/id/32745121/oklahoma-loses-two-more-espn-300-commits-wake-football-coach-lincoln-riley-departure) · [SI OU tracker](https://www.si.com/college/oklahoma/football/oklahoma-recruiting-tracker-where-recruits-stand-after-lincoln-rileys-usc-move) · [SI new-hire recruiting Dec 2021](https://www.si.com/college/2021/12/22/new-coaching-hires-in-college-football-recruiting) · [Yahoo Nelson flip](https://sports.yahoo.com/5-star-qb-recruit-verbally-commits-to-lincoln-riley-and-usc-a-day-after-decommitting-from-oklahoma-232323202.html) · [ESPN Florida zero-decommit note](https://www.espn.com/college-football/recruiting/story/_/id/46758805/college-football-recruiting-rankings-update-coaching-change) · [247Sports 2024-class decommit audit](https://247sports.com/longformarticle/college-footballs-decommitment-reality-what-the-2024-recruiting-class-says-about-if-commitments-stick-or-not-234573678/)