We grade ourselves against the line. And we publish our losses.

FanFave forecasts game outcomes before tip-off, in writing, committed publicly where it cannot be edited after the fact. Then we publish what happened — including the bets that lost and the games the model says “no edge — pass.” All of it, scored, with equal weight. No highlight reel, no “8-2 last week.”

---

The honest part is the baseline. We do not grade ourselves on whether the team won. We grade ourselves against the closing line — the hardest public number there is — because beating “did they win” is easy and beating the market is the only thing that means anything. Most days, the honest answer is pass.

Each forecast ships with its full reasoning: the factors, their weights and confidence bands, the situation it assumes, and the cases we could not rule out. Inspect it. Question it with your own LLM. We are not asking you to trust the pick — we hand you what you would need to prove it wrong.

Why this looks different from everything else in your feed:

• We publish the losing bets. A handicapper who only shows winners is the one to walk away from.
• We pass, a lot. When the math does not clear the line, we forecast nothing — and we show those days too.
• We grade against the closing line, not the final score. Beating the market after the vig is the only honest bar; most picks do not clear it, and we say so.
• No black box on the outcomes. You do not need to see how the engine works to check whether it has been right. The record will be public; recompute it yourself.

We are not selling picks. We are building a record you can audit. If that is not what you are after, we are not for you.