About & Methodology — CollegeAthleteInsider

C. B. Zakarian

C. B. Zakarian is an independent analyst who writes about college football and basketball — the parts of them that can actually be measured. He builds every model, chart, and calculator here himself from public data, shows the working, and never invents a number. When the data can't answer a question, he says so. Expect ratings rebuilt from scratch, season-long census work, and a plain-English read on the NIL-era rules.

I started this because most of what gets written about college sports falls into one of two buckets. Either it's a number with no context — a team is "elite" because it ranks 4th in something, and you're left to take that on faith — or it's a take with no number at all. I wanted a third option: show the metric, show how it's built, and be honest about where it stops working. That's the whole site.

I'm not affiliated with the NCAA, a conference, a school, or a data company. This is a personal project. When I publish a rating or a chart, I built it myself from public data, and I'd rather walk you through the math than ask you to trust a black box.

How I work

Every statistic here traces back to a specific, repeatable data pull. Each data-driven piece has a companion Python script that recomputes the article's figures from the cached source responses every time the site is built, and warns loudly if a published number stops reproducing — several of those scripts run more than fifty checks against the prose. The articles name the script and the dataset behind each number so you can see which computation produced what.

Be clear about what that is and isn't: it is my verification, not yours. The scripts and the cached API responses sit in the build repository on my machine and are not published, so nothing here asks you to take a download on faith — and nothing here is something you can re-run from this site. What you can check is the method. Every data piece states its source, its retrieval date, its sample, its filters and its formula, and the Python tutorials print their code in full on the page. If a figure doesn't survive that description, email me and I'll fix it or explain it. The one rule I won't bend: I never make up a number. If I can't source a figure, it doesn't go in. There are no "roughly" stats here standing in for ones I couldn't find.

The public sources I lean on:

CollegeFootballData.com — a free, community-run API (you'll need your own key) for football schedules, results, and advanced ratings like SP+.
sportsdataverse — the open-source hoopR (men's) and wehoop (women's) college-basketball data, built from public sources.
ESPN's public endpoints — schedules, scores, and box scores for worked examples and trends.
NCAA.org and NCAA.com — official records, rules, and the NET ranking.
Bart Torvik's T-Rank — a free, public basketball ratings site I reference for context.

Which seasons you'll actually see: the football census pieces — margins, home field, overtime, bye weeks, the FCS buy game, week one — all measure the 2024 season, because 2024 is the last full football season I hold a complete cached pull for. There is no 2025 football in that cache, so nothing here measures the 2025 season, and I'd rather say that than quietly leave you to assume otherwise. The 2026 pieces run off separate, dated pulls of the season now in progress, and the basketball work runs as recently as the 2025-26 season. Every data article states its season and its retrieval date in the first paragraph or the sources block; none of them claims to be the most recent season available.

What I'm skeptical of

Two things get my guard up faster than anything else in this field. The first is recruiting hype. Star ratings are a real signal in aggregate, but the certainty people attach to a single 18-year-old's ceiling is mostly theater — and I won't profile or rank individual high-school recruits, who are minors, regardless. The second is the small sample. Half of what passes for analysis in October is noise: a stat off six games, a "clutch" record built on three close finishes, a turnover margin that's mostly luck and will regress. When I think a number is too thin to mean anything yet, I'll say so instead of dressing it up.

What I won't do

I don't republish proprietary or paywalled datasets — no KenPom tables, no recruiting or NIL valuation figures lifted from services like 247Sports, Rivals, or On3. When those are relevant I'll mention and link them, but the analysis itself is built only from genuinely public data. And I don't tout: no picks, no plays, no "best bet of the week," no sportsbook sponsorship, and no affiliate links to a book. Point spreads and prices do appear here — a spread is the market's own estimate of a game, and turning one into a win probability is standard sports statistics, which is what the win-probability calculator does. They appear as inputs to a model or as a claim being measured, never as a recommendation to wager. This is analysis, not a tout sheet.

A note for athletes and families

Some of what I cover — NIL, revenue sharing, the transfer portal, roster limits — lands directly on athletes' lives, so I try to explain it plainly and without the hype. But nothing here is legal, financial, or eligibility advice. The rules change fast and vary by school and state; for your own situation, talk to your compliance office. The full disclaimer spells this out.

One name on the work

My name is on every piece here because I'd rather stand behind the work than hide behind a brand. There is no masthead to appeal to and no house style to blame: if a chart on this site is wrong, there is exactly one person to email about it, and that is the point. The method is the same on every page — public data, a stated sample, a stated formula, a real chart, no invented numbers — and where a question can't be answered from public data, I say so rather than reaching for a number that sounds right.

Get in touch

Corrections, questions, and good-faith arguments are all welcome. Reach me at contact@collegeathleteinsider.com or through the contact page. If you think a number here is wrong, tell me — and point me to the data. I'll fix it or explain it.