Pricing Models - Signals
Most signals are from market data, these can be used to identify breakouts, mean reversion and more. Signals are adaptively weighted for the current conditions. As part of this, we use adaptive signal composition, which is a method of combining multiple signals that adapts to changing market conditions and avoids the requirement to calibrate or hand tune signal composition. The historic output of these signals can be reviewed in the Compass dashboard.
Signal application in pricing enhances model predictivity and provides protection against arbitrage and other nefarious trading strategies. Also see Predictivity Review,Different signals have different applications in pricing:
Signal Configuration
Configuration at pricing.adjustmentSignalParameters
Signals can be configured to be attractive, defensive, or both.
Attractive
Attractive signals result in tightening of spreads to become more favorable to clients in order to attract flow.
- BID: if the signal thinks the price is going up it will increase the bid to attract more sellers i.e. it will increase what we are prepared to BUY for.
- OFFER: if the signal thinks the price is moving down, the offer will decrease to encourage more buyers i.e. it decreases what we're willing to SELL at.
By tightening spreads and offering more favourable prices to counterparties ahead of a move, the model is able to attract buyers/sellers to trade with us and skew our position towards the side that will benefit us in the future.
e.g. we increase what we are willing to buy for, expecting market prices to increase in the near future. This allows us to onboard a long position which can be offloaded at a better price after the prices move.
Defensive
Defensive signals make our rate less attractive to traders by widening out the side we are expecting to move against us and therefore avoiding onboarding risk.
- BID: If the signal expects prices to decrease, it will decrease the bid. i.e. we decrease what we're willing to BUY at.
- OFFER: if the signal expects prices to go up, it will increase the offer i.e. we increase what we would SELL for.
By making the prices less favourable to traders, we can discourage buyers and sellers from trading with us and so decrease the likelihood of onboarding new risk and hence reduce our exposure going into a move.
Signal WIDENING CAP
Configuration at pricing.adjustmentSignalWidthLimiting
WidthLimitingPricingNode is used to limit the impact of signals on an order book.
Widening is benchmarked against the worst price in an order book. There is no point going too much wider than that
the reference order book used to discover the worst price is either
- filtered source LP market data order book
- source order book pre-signals
It prefers filtered source LP market data order book, but if there is less than the configured normalQuantityMultipleRequired available in that order book it uses the source order book pre-signals