October 18, 2026
Most odds screens show you one number: the current price. The more useful column, in my experience reading these boards, is the pair of change figures sitting next to it — what moved in 24 hours, and what moved over seven days. When those two point the same direction, you're usually looking at a trend. When they point opposite ways, you're looking at a book that's arguing with itself.
Today's screen has three good examples of that argument, plus one clean counterexample. Here's the polymarket analysis I'd run on each before adding anything to a watchlist.
The two headline books on Brazil's presidential race are moving in opposite directions on opposite clocks.
Flávio Bolsonaro sits at 81.3% Yes, up 4.1 points in 24 hours but down 6.6 points over the week. The mirror book, Lula, prints 19.5% Yes — down 4.0 on the day, up 8.0 on the week.
So the weekly tape says the race tightened meaningfully. The daily tape says roughly half of that tightening came back out in one session. That's not a contradiction so much as a signal about what kind of information is driving the book: a slow, multi-day repricing (polling drift, campaign news absorbed gradually) interrupted by something sharper and more recent.
Add the two Yes prices: 81.3 + 19.5 = 100.8%. That 0.8-point overround is tight for a two-sided political pair, which tells me the two books are being arbitraged against each other fairly efficiently. Combined turnover is real — roughly $1.64M across both in 24 hours against $48.3M of lifetime volume — and liquidity is deep on both sides ($664K and $841K respectively).
The research prompt here isn't "who wins." It's narrower: which clock is right? If the weekly drift reflects something structural, the daily bounce is noise. If the daily bounce reflects a dated catalyst, the week was the noise. Dated catalysts in an election cycle are knowable in advance — that's the thing worth checking before the number matters.
The most striking turnover ratio on the board belongs to a small, young market: "Will there be no Hormuz agreement?" at 90.5% Yes.
Total lifetime volume: $513,013. Volume in the last 24 hours: $245,519. That's roughly 48% of everything this book has ever traded, done in one day — and the price moved -4.8 points while the 7-day change reads exactly 0.0.
I treat turnover ratio (24h volume ÷ total volume) as a rough measure of how settled a price is. A mature book like a national election pair turns over a few percent a day; a 48% ratio means the price you're looking at was essentially established this session. Add the negative framing — this is a question where "Yes" means *nothing happens* — and you have two separate reasons to read the quote carefully rather than at face value.
Note also the deadline: November 2. At 90.5%, the book is pricing roughly a 1-in-10 chance of an agreement materialising inside about two weeks. Liquidity is thin by comparison at $172,545 — the shallowest book in today's set — which is exactly why a single active session can shift it nearly five points.
For contrast, look at "Will the US confirm that aliens exist before 2027?" — 5.1% Yes, up 1.0 on the day and 2.0 on the week. Same direction, small magnitude, no disagreement.
What makes this book worth noting isn't the probability. It's the size: $42.1M in lifetime volume, the largest total on today's screen by a wide margin, against only $395K traded in 24 hours. That's a turnover ratio under 1% — the opposite profile to the Hormuz book. An enormous, slow, long-dated tail that accumulates volume over months and reprices in single-point increments.
With roughly ten weeks to the deadline, a 5.1% price implies something in the neighbourhood of a 0.5% chance per remaining week. The practical question on a book like this is almost never the physics — it's the resolution language. "Confirm" is doing enormous work in that sentence, and on tails this long, definitional risk tends to dominate event risk.
None of the above is a trade recommendation, and trader execution on this project is currently off. What I'm doing is cataloguing prediction market odds by behaviour rather than by headline: which books are mature and slow, which are fresh and jumpy, and which ones have a day/week sign conflict that deserves a catalyst check before anyone treats the quote as consensus.
A simple three-column screen does most of the work — price, turnover ratio, and whether the two change figures agree. Run that across any board and the markets that actually need research separate themselves from the ones that are just sitting there.
I post these observations as they develop, free, in our Telegram channel — timestamped notes on movers, liquidity shifts, and upcoming resolution dates. If you want the running watchlist alongside the fellow traders who read it, that's where it lives.
It usually means a slower multi-day repricing has been interrupted by newer information, or that one of the two moves was driven by a small number of large orders rather than broad flow. It's a prompt to check whether a dated catalyst landed recently, not a signal in itself.
Dividing 24-hour volume by total lifetime volume tells you how much of a book's price discovery happened recently. A ratio near 50%, like the Hormuz deadline book today, means the quote was essentially set in one session. A ratio under 1%, like the long-dated disclosure market, means the price has been stable across a long accumulation period.
Overround is the sum of all outcome prices above 100%. Brazil's two headline books sum to 100.8%, meaning the pair is tightly arbitraged and the implied probabilities are close to internally consistent. Wide overrounds suggest the two sides aren't being traded against each other efficiently.
Long-dated tails stay open for months, so volume accumulates even at low activity levels. They also attract traders on both sides — some buying lottery-ticket upside, some systematically selling the tail — which keeps turnover steady without moving the price much.
No. Everything here is observational research on publicly visible prediction market odds — prices, volumes, liquidity, and deadlines. Trader execution is off on this project, and nothing in this post should be read as advice to enter a position.