Skip to content

The risk layer for event contracts

Predictive intelligence and risk underwriting tools for market makers and institutions.

Explore the platform
THE PROBLEM

Volume is here, liquidity is not

Most contracts barely trade because no desk can price, hedge, net or warehouse event risk at size, so books stay thin and institutional capital stays out.

—
24h volume
—
Open interest
—
Trades per minute
Kalshi and Polymarket
THE SOLUTION

A risk-adjusted value, never a bare point

Senthos gives market makers and institutions independent predictive intelligence for every contract, built from quotes, TWAP, correlated flow and cross-event covariance, so desks can underwrite, margin, net and hedge their exposure.

Illustrative inputs feeding one risk-adjusted band around a venue quote; the band widens as the inputs deteriorate.
DESK APPLICATIONS

Margin and financing

The same intelligence that marks every contract runs the desk around it, from the collateral a position needs to how its inventory is financed.

Cross-venue netting

Portfolio margin

Today every leg posts its full worst case, venue by venue. Net the correlations across the whole book and the collateral falls to the residual portfolio risk.

Illustrative before and after: the venue's worst-case lock beside Senthos margin, with the book and each leg's venue listed alongside, advancing from one market held on two venues to opposing positions to a deep book, freeing 8, 38 and 60 percent of collateral.
Bilateral and tri-party

Swaps and repo

Finance inventory or swap event exposure, bilaterally or through a tri-party agent, with Senthos marking collateral hourly and setting haircut and margin.

Repo flow on Senthos marks: the engine marks the inventory and sets the haircut, the contracts sit as collateral with the lender or a tri-party agent, cash moves against them and returns with the repo rate; swaps run on the same marks.
RISK WORKFLOW

One valuation basis for pricing, risk and execution

Marks, factor exposures and backtests are computed from one versioned market snapshot. Each mark retains its model version and inputs, so pricing, risk and execution work from the same figures.

01Pricing correlation

Priced as a correlated book

Pairwise correlation is estimated across listed markets and refreshed as markets are added, so each contract is priced with the contracts it moves with.

Illustrative pairwise correlation matrix across seven event markets: lower triangle with the coefficient in each cell on a diverging scale, terracotta for negative and the accent colour for positive, with one pair read out at a time.
02Factor risk

Factor exposure across the book

Map positions to common macro and event factors, aggregate net exposure, and identify concentration that contract labels do not reveal.

Factor risk visualization
03Point-in-time capture

Full-depth books, kept as of every moment

Every order-book level, trade and settlement on Kalshi and Polymarket is recorded with the venue's timestamp and never revised. The reference data each market settles on is retained alongside it, so any mark can be reconstructed exactly as it stood at the time.

Illustrative captured order book: seven levels each side of a Kalshi market with price, size, spread and mid, stamped with the venue, the depth captured and the capture time.
04Backtesting and flow

Backtests on the books as they stood

Strategies are replayed against the order books as they stood, fill by fill, through to settlement, so slippage is measured against actual depth.

Illustrative backtest replay: the cumulative profit and loss of a strategy replayed on the captured full-depth books, trades and settlements, against the same strategy priced at venue mid, with spells of adverse flow shaded, correlated-flow events marked, and fills, hit rate, adverse-fill share and drawdown following the replay.
05Exogenous pricing

Priced from the data behind the market

Each market is priced from its underlying data, whether rates, economic releases, weather or polling, so a mark exists even where the venue has little or no trading. The model version and inputs behind every mark are retained.

Illustrative model mark distribution on a zero-to-one yes-price axis, with the mark read out as it moves.
INTERFACES

Consistent outputs in every interface

Every interface reads one set of marks and risk figures, so a number on the console is the number returned by the API.

Risk API

Access marks and market state programmatically, live or as of any point in time.

Python SDK

Bring marks and exposures into research and models.

MCP server

Give agents permissioned access to the same figures, within limits the desk sets.

Web console

Review marks, exposures and exceptions in one place.

CONTACT

Join the private beta

We are onboarding a limited number of market-making, trading and risk teams.

Interested? Contact us