October 26, 2026
Most of the writing about sports books on Polymarket focuses on the moment of maximum uncertainty — the hours when a line is still moving and the order book is genuinely two-sided. I find the opposite case more instructive. When a cluster of related markets has already collapsed toward the boundary, the small residual differences between them stop being noise and start being a readable map of what the market thinks is still physically possible.
The snapshot I'm working from here covers eight books on a single Kansas City–Seattle matchup dated October 26, 2026. Every one of them had converged. That sounds like a non-story. It isn't — the arithmetic between those converged prices is one of the cleanest teaching examples I've logged this month.
Here's what the eight books looked like together:
Start with the two books that look like opposites but aren't. Chiefs −3.5 pays if Kansas City wins by four or more. Seahawks −2.5 pays if Seattle wins by three or more. Add the prices: 0.1% + 99.6% = 99.7%. The missing 0.3% is the middle — every outcome from a Chiefs win by three down to a Seahawks win by two, including a tie.
That residual is the whole point. Two spread books on the same game never have to sum to 100% unless their thresholds are perfectly adjacent, and the gap between them is a directly readable probability for an outcome band that has no market of its own. This is the single most useful habit in any multi-book polymarket analysis: stop reading each line in isolation and start differencing adjacent thresholds.
Apply the same logic up the board. Seahawks −3.5 at 99.5% minus Seahawks −7.5 at 99.5% gives roughly zero for a margin of four through seven. Minus Seahawks −9.5 at 98.7% leaves about 0.8 points of probability for a margin of eight or nine. And −9.5 minus −10.5 leaves 0.2 points for a margin of exactly ten.
Notice what's missing: the key-number structure. In a genuine pregame market, the three-point and seven-point bands carry visible mass, because NFL margins pile up there. Here the four-through-seven band prices at essentially nothing while eight-or-nine carries real weight. That's the signature of a board that has stopped forecasting a distribution and started tracking a single developing state. Converged books flatten key numbers; live books don't. It's a fast diagnostic for telling which kind of market you're looking at.
The liquidity column is where I'd point a newer reader. Intuition says depth should concentrate in the tightest, most contested book. It doesn't. The moneyline carries $707K of liquidity at 100.0% — a price with no residual uncertainty at all — while Seahawks −2.5 at 99.6% carries just $85K. Meanwhile the widest, least certain rungs hold more: $88.5K on −7.5, $76.7K on −9.5, $73.3K on −10.5.
Within the spread family, then, depth is roughly proportional to remaining doubt, which is exactly what you'd hope. The moneyline's $707K is a different animal — that's exit liquidity for a book that did $5.8M of turnover in 24 hours. Depth at the boundary is mostly about the volume of positions that still need unwinding, not about disagreement over the outcome.
The moneyline shows $5.83M of 24-hour volume against $5.51M of total volume. A 24-hour figure larger than the lifetime figure is a reporting inconsistency, not a real phenomenon, and I'd treat the precise numbers as approximate. I'd rather name that than quietly build a conclusion on top of it.
The change figures look dramatic: Seahawks −10.5 up 77%, −9.5 up 73.7%, the Over down 49.5%. Those are relative moves on small bases. A book travelling from 55% to 98.5% is a 77% relative change and a 43.5-point absolute change — two very different-sounding descriptions of the same event. Near the boundary the distortion gets worse: a move from 0.2% to 0.1% is a 50% drop and a one-tick move.
Whenever I screen movers, I convert everything to absolute points before ranking. Any polymarket analysis built on percent-change leaderboards will systematically over-weight cheap tails and under-weight genuine repricings in the 40–60% range, where the real information usually lives.
Converged clusters go into the methodology notebook, not the watchlist. There's no research question left on the outcome. What's reusable is the technique: difference adjacent thresholds to extract unpriced bands, check whether key numbers still show structure, read liquidity as a proxy for unwind demand rather than conviction, and normalise change columns to points. I apply that same sequence to election ladders, rate-decision stacks and anything else that ships as a family of related thresholds.
None of this is a trade recommendation, and execution on this project is currently off. It's a framework for reading prediction market odds across a board instead of one line at a time.
If you want these snapshots as they're logged — with volume, liquidity and catalyst notes attached — the free watchlist runs in our Telegram channel. It's a research feed shared openly with fellow traders, not a signal service. Come read along.
Because their thresholds usually aren't adjacent. Chiefs −3.5 pays on a Kansas City win by four or more; Seahawks −2.5 pays on a Seattle win by three or more. Everything between those thresholds — a Chiefs win by one to three, a tie, a Seahawks win by one or two — belongs to neither book. In this snapshot that uncovered middle was worth about 0.3%.
Not necessarily. Displayed prices round, and a book showing 100.0% may still be trading at the minimum tick above zero on the other side. What it does mean is that the market has stopped expressing meaningful uncertainty, so the remaining activity is mostly position unwinding rather than price discovery.
Large turnover creates large open positions, and those positions need somewhere to exit. The $707K of depth on the moneyline reflects $5.8M of 24-hour volume needing an off-ramp, not disagreement about the result. Within the spread family, depth tracked residual doubt more closely — the widest lines held more than the tightest ones.
I'd convert to absolute percentage points first. Relative change exaggerates moves on low-priced books — a one-tick drift from 0.2% to 0.1% registers as a 50% move — while understating substantial repricings near the middle of the range. Ranking by points keeps the screen pointed at markets where real information arrived.
No. This is observational analysis of a public market snapshot, published as a methodology example. Trader execution on this project is off, and nothing here should be read as advice to enter a position.