MLB NRFI / YRFI predictions — Monday, August 31, 2026

NRFI is the most popular niche bet in baseball, and most of what's written about it is streak chatter — this team's hit a run in eight straight first innings, that starter always settles in late. None of it comes with a ledger. We built this board to be the opposite: one first-inning model, a probability and a fair price for every game, set beside the market's line and graded against the real first inning — every day, in public, misses included.

What an NRFI bet actually is

No Runs First Inning: you're betting that both halves of the 1st go scoreless. YRFI is the other side. That framing hides how narrow the bet really is. Only the first three or four hitters in each order usually bat, and the arm they're facing is the one player you know for certain hours in advance. No bullpen. No pinch hitters. No manager decisions. It's the top of each lineup against the opposing starter, twice, and then it's over — usually inside twenty minutes.

That's also why it settles faster than any other baseball bet. By the bottom of the 1st, you have your answer.

Why the first inning can be modeled at all

Most of a baseball game is chaos you can't see coming — the sixth-inning reliever, the pinch hitter, the ball that clips the chalk. The first inning is the one slice where the personnel is fully known: a confirmed group of leadoff hitters against a named starter, in a known park. Small, closed matchup — the kind a model can actually get its arms around.

We project each side's first-inning runs from exactly that matchup: how the top of the order has actually hit against this kind of arm — handedness matters enormously — set against what the starter tends to give up early, with the park folded in. Two run projections, one per half-inning, become one probability that the inning stays scoreless, and that probability becomes a fair price. The exact weighting is ours, and it re-tunes weekly against real first innings — same discipline as every other model on the site (how MatchWiz works).

Small and closed doesn't mean predictable. One hanging slider beats a great projection. The model prices the inning; it doesn't script it.

Reading the board

Every game row shows the same four things: our NRFI probability, the fair odds that probability implies, the market's posted line, and the gap between the two. The gap is the whole story. A game where we say 58% NRFI and the book prices it like 57% is the market agreeing with us — nothing to see. The interesting rows are the ones where our fair price and the book's price disagree by real margin. That disagreement — not the raw probability — is what a lean means here.

The receipts

Every first inning gets graded — no exceptions, no curation. Across 534 graded games this season, the actual first inning has stayed scoreless 48% of the time, and our higher-probability call has been on the right side 53% of the time. Those two numbers live on the board itself and update as innings settle.

Two honest notes on those numbers. First, calling the higher-probability side correctly isn't the same as beating the price — most NRFI markets are priced close to true, so a right call at a short price can still be a break-even bet. Second, when our strongest disagreements with the market have graded out with an edge, you'll see it above; when the sample is thin or the edge fades, you won't. The board doesn't argue either way. It just shows the ledger.

Where the edge is — and isn't

Here's the part most NRFI content won't tell you: this is a hard, efficiently priced market. Roughly half of major-league first innings are scoreless, the books know each game's true rate within a point or two, and the vig sits on top of all of it. Blanket-betting NRFI every night — or riding whatever streak is going around — is a slow leak, not a strategy.

If there's something to act on, it lives in the few games where a disciplined model genuinely disagrees with the price — a lineup whose top three are quietly miserable against this arm slot, a starter whose early-inning numbers don't match his name value. That's why we publish fair odds instead of picks: the gap between our price and the book's is the only number here with any teeth. We treat this board as transparency, and when a first-inning edge earns a sustained, validated record, it surfaces on the Plays of the Day board the same way every play does — the backtest decides, not a person.

How to actually use it

• Compare prices, not vibes. Find the game you care about, read our fair odds, and set them against the price your book is offering. If the book is better than fair, that's worth knowing. If it's worse, that's worth knowing more.
• Cross-check the starter. A strong NRFI number usually rides on the arms. The strikeout board shows how each starter projects against the full lineup tonight — a pitcher the model likes for whiffs is usually a pitcher it trusts early.
• Roll back the archive. Every past date is one click away, graded. Before you trust any first-inning model — ours included — read a few weeks of its receipts.

The traps

• Streak logic. "Eight straight YRFIs" is trivia, not signal. First innings are near-independent events; the streak tells you what happened, not what's priced wrong tonight.
• NRFI parlays. Stacking several short-priced, near-coin-flip legs multiplies the book's hold on every leg. The math gets worse with each add, not better.
• Ignoring the price. A 60% NRFI game at −180 is a worse bet than a 52% game at +105. The probability means nothing without the odds next to it — which is exactly why the board shows both.
• Name-brand starters. The market prices reputations fast. An ace's NRFI line already carries his name premium; the model only cares what the top of the order does against his actual profile.