Guru's Verification engine ensures consistency, confidence, and trust in the knowledge your organization shares. Learn more.

Pricing Models - Overview Price Formation

Compass Pricing is a managed service for our clients.

Rate formation is a vital part of MFX Compass. We don't use aggregation
We can stream to clients and venues any combination of
• client’s existing pricing
• aggregated liquidity
• Compass's own pricing
Internally however, we make pricing and hedging decisions against prices formed from listening directly to ECNs, using the same sorts of models as top tier banks, with detailed analytics of every aspect of rate formation. These models place a heavy emphasis on Predictivity
The performance of the hedgers show that the aggregate top of book is regularly wrong, and our internal reference rate is a better benchmark for determining which trades should be allowed through to the risk book

Compasses pricing model forms its own enriched TOB price based on the information it has consumed from a number of different LPs quotes from the market. The model then runs through each of the pre defined pricing nodes and makes a decision on how the price should be adjusted based on the configuration of the node.

Prices are formed in Compass by:

The pricing model takes into account a number of different LPs price quotes
Filters ensure latent or stale data, anomalies, crossed rates etc. are removed
A clean order book is produced from the filtered LP quotes which then produces unskewed reference price using market interpretation.
HFP signals are then used to predict changing market conditions and make decisions on how to adjust spread and adapt skew

high frequency skew => PCA combination of trading signals.
Risk-reducing skew

Models reference a bespoke constructed base spreads timetable, spreads are then adjusted according to

Volatility
Liquidity
Exceptionally high risk
Market directional movements
Scheduled news
Benchmarking

The model runs through each pricing node to adjust prices in line with market conditions and defined parameters
Safety => When a price is indicative we broker the flow.

Firewalls can be set to ensure the rate is protected in unusual market conditions

Can be temporary or until manual intervention
Volume or PnL limits
Price spike
Price barrier e.g. pegged

Net Open Position limits one-sided indicative on NOP-increasing side
Sanity check

Depth and published quantity of models can be controlled

filters

Filters remove anomalous quotes in constructing a filtered market data order book source.

A stale quote presenting itself as the best bid or offer presents latency arbitrage opportunities and results in the formation of an incorrect order book. This renders pricing vulnerable.

Crossed market quotes are also removed from the filtered market data order book. A summary of the filters can be found in pricing.filters and hovering over the tooltip (‘i’ icon)

Signals

The signals are High Frequency Predictors that utilise available market data to predict market directions within the next few minutes.

Compass provides an adaptive signal composition, which is a method of combining multiple signals that adapt to changing market conditions and avoid the requirement to calibrate or hand tune.

The application of signals can be used in a variety of areas:

Price Formation
Distribution - tailoring rates to:

Attract liquidity in aggregated environments
Defend against aggressive flow
Internalise decision making

Hedging - withdrawing passive hedging when the trade being filled is predicted to go against us

pricing nodes

Once the filtered market data order book is constructed, the order book is processed by a series of individual pricing nodes. Each node contains specific pricing logic and will update the price if the conditions within the node trigger a change.

These include:

Arb protection - prevents inadvertently publishing a price crossed with the market.
Wide spread suppression
Minimum spread protection
Widening