Approach

How a quote gets made.

Where a quote’s price comes from, how large it may be, what a new contract has to answer first, what takes a quote down, and the conduct it is held to. The research notes carry the long versions.

Fig. 1 Illustrative simulation: a fair-value path, our two-sided quote around it, and simulated futures that each resolve to 0 or 100. Not market data.

A quote starts as a probability.

Each quote is built around fair value: our estimate, in cents, of the chance the event happens. The bid and the offer sit either side of it.

Proprietary algorithms
Fair value comes from proprietary algorithms built on the desk’s own research. How they work is not published.
A low-latency quoting system
Quotes follow fair value closely, and come down fast when it moves.
Related contracts must agree
Where exactly one outcome in a set pays, the prices should add up to about 100¢. On a strike ladder, a higher rung can’t be worth more than a lower one. The same question on two venues should carry the same price once fees and settlement terms are allowed for. Quotes are checked against all three.

Read the long version

What a two-cent spread pays for 5 min

What the width either side of fair value pays for, and why one width can’t serve every price.

Sized to the loss at settlement.

Event contracts don’t drift to their final value. They jump to 0¢ or 100¢, often within a minute of the news, and the limits on what we hold are set on the jump.

Worst case first
A limit is a loss at settlement. A YES bought at 62¢ can lose 62¢ a contract (Fig. 2), and that times the size is what counts, whatever the price did last week.
One event, many contracts
Contracts that hinge on the same fact are one risk under several names. Positions are netted by event, and the limit applies to the worst outcome across all of them.
Releases and breaking news
There is no fixed playbook. It depends on the contract, and on how far its price can gap in a single moment: a release, an injury, a headline. Where that gap can be large, we widen, cut size or step out entirely.
Controls that fail closed
Price bands, loss limits per contract and per event, a limit across the whole book, quotes pulled when an input goes stale and a kill switch, among others (Fig. 3). They run apart from the strategy, and when one can’t confirm that trading is safe, quoting stops.
A YES contract bought at 62 cents drifts until resolution, then jumps to 100 cents (a 38 cent gain) or to 0 (a 62 cent loss). 100¢ 62¢ 0¢ YES +38¢ NO −62¢ BUY YES AT 62¢ RESOLUTION
Fig. 2 A YES contract bought at 62¢ ends at +38¢ or −62¢. There is no in-between, so we size risk to the jump, not to the drift.
Six controls, smallest scope first Schematic · not the live book

Limits on what can be held

  1. 01 Pre-trade price band

    Looks at
    Each order, before it is sent.
    When it trips
    An order priced outside a band around fair value is refused.
  2. 02 Per-contract loss limit

    Looks at
    One contract.
    When it trips
    The worst loss at settlement on that contract is capped.
  3. 03 Per-event loss limit

    Looks at
    Every contract that settles on the same event.
    When it trips
    Positions are netted, and the worst outcome across them is capped.
  4. 04 Whole-book limit

    Looks at
    Every open event at once.
    When it trips
    The total across the book is capped.

Switches that stop quoting

  1. 05 Stale-input pull

    Looks at
    Every quote built on an input that has stopped updating.
    When it trips
    Those quotes are pulled automatically until the input recovers.
  2. 06 Kill switch

    Looks at
    All quoting.
    When it trips
    Quoting stops, and stays off until a person turns it back on.
Fig. 3 Each board is the same made-up book: three events, each column one contract, each dot one resting order. Lit dots are what a control looks at; rings are quotes it has taken down. These six are the main controls; there are others. To switch them on and off one at a time, try the control stack in Sizing for the jump.

Read the long version

Sizing for the jump 5 min

Netting by event with a worked example, the minute a release lands, and a control stack you can switch on and off.

Before we quote a new contract.

Seven questions a new contract answers before its first quote. One it can’t answer is enough to leave it alone.

  1. Who resolves it, and from what source?

    The rules name a source: an official publication, or a named measurement at a stated time. We read the rule text itself. If two careful readers could take it two ways, that is the answer, and the contract waits.

    More on Resolution source
  2. At what hour, and on whose clock?

    The settlement time decides when the risk ends and which readings count. A contract that settles on a figure published the next morning is open overnight, however settled the evening looks.

    More on Settlement
  3. How far can the price jump in one moment?

    A release, an injury or a headline can move a price most of the way to 0¢ or 100¢ at once. The size of that jump, more than the quiet days around it, decides the size we can carry into it.

    More on Jump risk
  4. What else settles on the same fact?

    Other rungs of a ladder, other outcomes in a set, the same question on another venue. They count as one event with the new contract: positions are netted, and the limit applies to the worst outcome across all of them.

    More on Netting
  5. Do the prices around it agree?

    A new rung or outcome is priced to fit the ladder or set it joins. If the set as listed doesn’t add up, that is a question about the rules before it is a quote.

    More on Coherence
  6. What is scheduled before it expires?

    Anything on the calendar between now and settlement is a moment the price can gap. For each one we decide, contract by contract, whether to widen, cut size or step out.

    More on Scheduled release
  7. Can we size it to zero?

    The kill switch can take every quote down at once. Stopping quotes doesn’t close what is already held, so a position only goes on if its worst case at settlement fits inside the limits.

    More on Kill switch

When something breaks, quoting stops.

News breaks overnight and some markets never close, so the systems are built to stop quoting on their own when something upstream fails, without waiting to be asked.

Each venue on its own terms

We quote on Kalshi and Polymarket US. Tick sizes, fees, rate limits and settlement rules differ between them, and each venue is handled on its own terms.

Time budgets on inputs

Each input a quote is built on is expected to update within a set interval. When a feed lags or a venue degrades, the quotes that depend on it are pulled automatically until it recovers.

Reports to venues

We report to venues on the liquidity and quotes we provide in their markets.

Logged and kept

Every order and fill is logged and kept, along with funding movements, so any trade can be reconstructed when a venue or a reviewer asks.

The rules we trade under.

Prices are only useful if people trust the process behind them. These are the rules we hold to on each venue we trade on, in full.

  1. 01

    We trade within the venue’s rulebook.

    Each venue’s rules, and the laws and regulations that govern it, come before any trade. Where a rule and a trade disagree, the trade doesn’t happen.

  2. 02

    We send only orders we’d be content to see filled.

    No wash trades, no spoofing, no layering, and no trading with ourselves to make volume.

  3. 03

    We trade only on information we can account for.

    No material non-public information. If we can’t say where a piece of information came from, we don’t trade on it.

  4. 04

    We keep our hands off the outcome.

    We never try to influence an event we trade, or the process by which a contract resolves.

Questions

A market maker that shows its work.

Venues, counterparties and reviewers: questions about anything on this page go through the contact page, and we reply within two business days.