NRFI, measured: 2,237 MLB first innings and their first-pitch prices on Polymarket

Bet no run in every MLB first inning on Polymarket and you lost money. No run scored in 49.4% of 2,237 games. In the 2,234 with a first-pitch price (three postponed games had none), the no-run side cost 51.4 cents on average and came in 49.3% of the time, the gap the chart below draws.

That gap cost $1 on NRFI every game 3.95% before fees and 6.38% after Polymarket's sports taker fee, the only after-fee return in this study whose interval excludes zero.

The sample is 136 games from September 4, 2025 to September 23, 2025 and 2,101 games of the 2026 season from March 26, 2026 to September 12, 2026, every MLB first-inning run market that settled Yes or No. Of the closed markets for games before September 11, 96.8% have a settled result in 0xinsider's database; the other 72 are not counted.

These are small markets. Polymarket reported $10.4 million in volume on the 2026 NRFI markets, a median of $1,833 a game, against $1.27 billion on the same season's MLB moneylines. Polymarket counts each share traded at its $1 face value, so the cash that changed hands was less. Large-trade alerts are too few to price them, 73 buys since April 2, so this study prices every game at first pitch instead.

For the price, we read each market's history from Polymarket's public price API and took the no-run share's last price before first pitch, a median of 0.9 minutes before the start. On 10 games Polymarket listed a later start than MLB did, and we used MLB's time, because a price read after first pitch already knows part of the inning. The three postponed games are left out of the pricing, which leaves 2,234 priced games.

No run scored in the first inning of 1,104 of 2,237 games, 49.4% with a half-width of 2.07 points. The 2026 season came in at 49.5% over 2,101 games, and September 2025 at 47.8% over 136.

From April on, the monthly rate stayed between 46.5% and 51.6%, inside half-widths of five to eight points. March's 59.6% rests on 57 opening-week games. In every month measured, a first inning was close to an even bet to stay scoreless.

The away team scored in the top of the first in 26.4% of games, and the home team scored in the bottom in 32.7%. Both teams scored in 8.4%. Runs come from MLB's linescores, which split the inning by team.

So a YRFI bet usually wins on one team's half, and more often on the home team's. The data counts runs; it does not say why the bottom half scores more.

Of the 1,133 games with a first-inning run, 507 had exactly one, 312 had two, 168 had three, and 146 had four or more. A scoring first inning averaged 2.08 runs.

An NRFI bet loses the same way on one run or five. The size of a first inning matters more for the first-five-innings total and the full-game total, which Polymarket lists on the same game.

By team, the share of games with no first-inning run ran from 40.9% for the Minnesota Twins to 58.3% for the Miami Marlins. Each rate rests on about 150 games and carries a half-width of about 8 points.

To test whether those differences are real, we simulated 5,000 seasons on the same schedule in which every team's half-innings score at the league rate for that half. The simulated team no-run rates spread at least as wide as the real ones in 36.5% of seasons. How often each team allowed a first-inning run spread a little wider than chance, matched by 10.2% of seasons; how often each team scored in its own first inning spread narrower than chance.

So a season of team NRFI records cannot tell a real no-run team from a lucky one. Starting pitchers, lineups, and parks may matter game by game; a team's season rate does not isolate them.

The no-run price at first pitch averaged 51.42 cents, with a median of 51.5 and 90.8% of games priced between 45 and 60 cents. The market treats most first innings as close to even.

The price does sort games. Where the no-run side was priced under 50 cents, no run scored in 44.4% of 748 games; where it was priced at 50 cents or more, 51.7% of 1,486. That gap is wider than the two half-widths together.

In every price band, though, no run scored less often than the price said, by 1.1 to 9.1 points. Scored as a forecast, the first-pitch price beat a flat 49.3% guess for every game by 0.15% on the Brier score, with an interval that includes zero.

Across 2,234 priced games, no run scored 49.3% of the time against an average price of 51.42 cents: 2.14 points short, with an interval from minus 4.20 to plus 0.04. The 2026 season alone came in 1.96 points short.

In money, $1 on NRFI in every game at the first-pitch price returned minus 3.95% before fees. Polymarket's sports taker fee today is 0.05 × price × (1 - price) per share, about 2.4 cents per dollar at these prices, which takes the NRFI return to minus 6.38%, an interval from minus 10.40% to minus 2.07% that excludes zero.

The same $1 on YRFI returned plus 4.85% before fees and plus 2.28% after the fee, with an interval from minus 2.28% to plus 6.61%, and neither figure counts the spread a market order pays. The run side was the better of the two bets; after costs, its edge is inside the noise.

We also took each market's earliest price in the 24 hours before first pitch. In 358 games the no-run price fell 3 cents or more by first pitch, from 53.87 to 48.22 cents on average, and no run scored in 47.5%. A no-run bet at the earlier price came in 6.39 points behind, an interval that excludes zero; at first pitch the same games were within a point.

In 445 games the no-run price rose 3 cents or more, from 49.61 to 55.32 cents, and no run scored in 53.3%. A bet at the earlier price came in 3.65 points ahead, an interval that includes zero, and at first pitch 2.06 behind. In the 1,431 games where the price held within 3 cents, both prices came in about 2.5 points behind.

In both directions, the last day's move pointed toward the result. That is closing line value applied to NRFI: a bet the price later moves against has more often been a losing one. The closing line value guide covers the idea.

Do not bet NRFI by default. In the 2,234 priced games the no-run side cost about two cents more than it paid back, and after the fee $1 on it every game lost 6.4%. Split by the first-pitch price, it lost in every band, as the chart below shows, though no single band is large enough to rule out chance on its own.

A price is a better guide than a team's record. Games priced under 50 cents for no run had fewer scoreless first innings than games priced above it, while team records spread no wider than chance. Watch the move too: an NRFI price that falls 3 cents after you bet has been a sign the bet was wrong.

Cut the costs you control. A limit order that rests on the book and fills pays no taker fee, which on a 50-cent bet is 2.5 cents a dollar.

It does not model starting pitchers, lineups, weather, or parks. It measures results and the prices the market set, and the team table describes past games only.

The price is the last point in Polymarket's public price history before first pitch. A buyer pays the ask, which sits above that price by up to half the spread, plus the taker fee, and the fee here is today's schedule applied to every game. The earliest-price comparison uses a price a trader could have taken a day out; this study does not know who took it or at what size.

Markets that settled on Polymarket without a recorded result in 0xinsider's database, 72 of 2,281 closed markets for games before September 11, are not counted. The sample is September 2025 and one season of 2026, and another season's run scoring can move the base rate.

Polymarket lists "Will there be a run scored in the first inning?" for nearly every MLB game. Yes pays if either team scores in the 1st inning and No pays if neither does, so No is the NRFI bet, short for no run first inning, and Yes is YRFI. A postponed game keeps the market open until it is played. A game canceled with no makeup settles 50-50.

Until April 2, 2026, the market ran the other way. It was titled "NRFI: Phillies vs. Brewers", and Yes paid when no run scored.

Anyone reading older prices has to check which format a market used: a Yes price of 60 cents meant a 60% chance of no run under the old title and a 60% chance of a run under the new one. This study reads each market by its own rules, and the chart below shows the first inning came out near even under both.

We matched every settled market to MLB's official linescore for the game. The settlement agreed with the first-inning runs in 2,237 of 2,237 games, including 3 postponed games followed to their makeup dates.

Large money leaves NRFI alone. Over the months the large-trade feed carried tens of thousands of big MLB moneyline buys, it carried fewer than a hundred on NRFI, as the chart below counts. If you want to follow large money into a first inning, there is almost none to follow, so the price at first pitch is the signal to read.

The MLB board lists live and upcoming baseball games on Polymarket, led by MLB: each side's price and its 24-hour move, live runs and the inning, the game's two most traded props, and the top five holders on each side with their grades and baseball win rates. The table view adds the favorite's 1-hour move.

Open a game for its event page. On Pro, its Markets section lists each market family, the first-inning run market among them, with a volume-by-family bar; without Pro it shows sample data. The large-trade feed shows recorded large trades admitted by the feed rules, live on Pro and 24 hours behind without it. The MLB betting guide covers every MLB market Polymarket lists and how each one settles.

The study's numbers come from four published runs: a read-only query against the production database at 22:28 UTC on September 13, 2026, a linescore check against the public MLB Stats API, a price history pull from Polymarket's public CLOB API, and an analysis script that joins the three exports and reproduces every table. All of them are at github.com/0xinsider/research, under polymarket-sports-markets, in the files whose names carry the word nrfi.

The universe is every MLB first-inning run market on Polymarket settled Yes or No for a game that started before September 13, 2026, each read by its own rules.

Read again on October 11, 2026 from 0xinsider's database, without the linescore check, the same window holds 74 more settled markets, close to the 72 the study found closed without a recorded result, and the no-run rate barely moves. The games since came in under even too, on a sample too small to move the base rate. The chart below sets the three side by side.