Trading Behavior
Where contracts trade on the probability axis, how trade sizes differ by venue, and when unusually large prints appear.
Volume and size mix
How a venue’s volume splits across trade sizes, and how much of it moves in large blocks. Pick a venue and a segment here — both controls govern every chart in this section.
About this page
The top chart counts only the contracts from trades at or above the threshold you pick; the ribbon below shows the full size mix for the same window. OG/Crypto.com isn't shown here — only daily totals are available for it, not individual trades.
Large-trade share radar
Flags the days when the big-block share ran well above its recent normal — not just busy days, but unusually top-heavy ones.
Flagged anomaly days
How this is calculated
Each raw trade is placed into a size bucket based on contracts_traded. The composition chart uses each bucket's share of daily contracts, not share of trade count. The spike radar compares the selected threshold's contract share with the previous 30 trading days and flags days at least 3x above baseline with at least 1M large-trade contracts.
Largest individual trades
The single biggest prints a venue has published — different notions of "big," since raw contracts, the larger side's dollar stake, and what the taker specifically put up don't always pick the same winners.
Largest trades in small markets
The same rankings, but restricted to trades that were unusually large for the specific market they happened in — a print that ate a huge share of everything that market traded, not just a big number in isolation.
How this is calculated
On Kalshi, a trade qualifies when it was at least 100,000 contracts and at least 20% of that market's entire lifetime volume in that one print, excluding parlays. Parlays are left out here for two reasons: a parlay combo is by construction its own tiny market, so almost any parlay trade looks like a huge share of a thin one — and the per-combo volume totals behind that ratio aren't reliable at that granularity. Parlays still show up in the table above, which has no market-share requirement. Because the denominator (the market's lifetime volume) can keep growing for still-active markets, a trade can drop out of this list over time even though the trade itself never changes.
On the competitor venues the denominator is a window, not a lifetime. Those tapes only run as far back as collection does — from a few weeks on the newest venues to about two years on ForecastEx — so "% of market" means share of what that market traded while we were watching it. Numerator and denominator are both inside the window, so the figure is still bounded by 100%, but a market that was already busy before collection started will read as more concentrated than it truly was. The 20% share floor is Kalshi's; the contract floor is set per venue, since these books are one to three orders of magnitude smaller and a flat 100,000 would empty the table.
A market must also have traded at least 20 separate times to appear at all. Without that rule the table degenerates on venues whose market identifier is close to one-per-trade — Novig's median market carries a single print, and its first build returned trades sitting at exactly 100% of "their market", which is true and tells you nothing. It is the venue-neutral form of Kalshi's parlay exclusion.
Volume by probability
The share of each venue's observed volume traded in each five-cent price bin.
Coverage differs by venue. Kalshi uses the taker's side; ForecastEx uses the yes leg; DKeX, Polymarket US, and Underdog use the named leg; ProphetX uses its published home-side price. DKeX and Underdog parlays are both excluded, because a combination price is the price of a whole basket rather than one market’s probability; DKeX’s are on DKeX · Parlays.