The math

Start with the ticket cash flows

Expected value is a probability-weighted estimate. For a standard house-side ticket, before fees:

The estimate is useful only to the extent that the probability is useful. Actual profit depends on the resolved result and costs. A market midpoint is not a known true probability.

Multiplying independent probabilities

If selected outcomes are independent and their probabilities are accurate, the probability of all of them winning is their product. Using 60% for each leg gives:

The engine rejects detected related legs and checks the books behind its marks. Remaining correlations and pricing errors can still affect the estimate.

Margin is a markup

Pregame applies a margin to the estimated parlay fair value. It is a percentage of that value, not a percentage of the collateral at risk.

A leg-count target is a margin for the whole parlay. It is not an instruction to add that same margin to every leg. Autopilot’s final margin can also reflect its cost floor and applicable adjustments.

The same EV, another way

For 60 fills with these same economics, total estimated EV is $30 before fees. Multiplying a 5% markup by $39.50 of capital at risk would give the wrong answer. Expected values add across tickets; shared outcomes still affect the range of actual results.

The same prices can produce different nights

Suppose 40 different tickets each have the example economics: $10.50 premium, $50 payout, and an estimated 20% hit probability. The book collects $420, with eight payouts expected under that model.

A shared winning pick can cause several tickets to pay together. More fills do not necessarily diversify the book, and repeated trading does not remove model error. Read the shared-outcome view alongside the individual tickets.

What the limits bound

Limits should be understood in dollars and in time. Per-ticket and total open risk describe committed capital. Daily fills describes a count. None of those fields is a promise about daily profit or maximum daily loss.

Choose limits that reflect losses you can bear. Check confirmed settled net as the session progresses, and remember that separate tickets can share the same underlying risk.

Shared outcomes change the book

Multiplication assumes independence. For related events, the joint probability can be higher or lower than the simple product. The direction and size depend on the relationship.

Pregame rejects detected duplicate, same-game, same-event, and shared-team legs within a request. Across separate tickets, concentration still matters. One result can affect many tickets.

Small price errors can be large relative errors

Neither error is automatically harmless. The trade’s size determines the dollars involved. A $10 premium at a 2¢ quote buys $500 of possible payout and requires $490 of house capital before fees. The fair-price band and capital limits constrain different parts of that exposure.

Price and fill rate belong together

A wider margin can produce more estimated value per accepted trade and fewer accepted trades. The useful comparison is expected net value per fill multiplied by the likelihood of getting that fill, using comparable requests and costs.

Autopilot learns from resolved auctions and uses your target win rate subject to its economic floor. The pricing panel labels its source and fallback state. Manual control uses your authored base margin instead.

One ticket, end to end

On the Sweat room, capital at risk and premium describe the position. Live chance to pay is an estimate. Confirmed settlement determines the recorded result.

What to check at a glance

• Engine state: has the desk confirmed it is quoting or stopped?
• Available capacity: how much capital is committed, and how close is it to your limits?
• Pricing source: is the displayed target learned, a fallback, or your manual value?
• Concentration: which shared outcomes affect the most tickets?
• Data freshness: are prices and settlement checks current?
• Confirmed results: how does settled net compare with your estimates and costs?