# OpusAMM

## A new way to conceive market making

Especially in the crypto space, market making platforms have always faced limitations due to the fact that only the internal supply and demand curve is visible with a satisfactory degree of reliability. This is caused  by the famous **Blockchain Oracle Problem**, which prevents protocols to safely access external data (and in many cases, even on-chain ones).

OpusAMM solves this problem by **using the Blockchain technology for the purpose it was originally created for**: giving a standard in data transfer and storage. Instead, the market data are fetched **externally,** using the most renowned tools and brokers (Reuters, ICAP, Refinitiv...) to achieve the maximum reliability and compliance. Then, the market making operations are delegated to **Artificial Intelligence**, which is a much better suited technology to deal with it.

## Machine Learning in market making

#### The objectives of market making

Market making amounts to set buying and selling prices for a given set of assets. In order for this to be profitable, a set of **objectives** must be met:

* The bid/ask **spread must be** **broad** enough to cover for potential losses.
* The bid/ask **spread must be tight** enough to make the prices attractive for as many clients as possible.
* The **inventory must be balanced** so that the risk of owning it stays within acceptable boundaries.
* The **arbitrage opportunities** must be eliminated or seized depending on whether they are against or in favour of OpusDigital.

#### The factors to consider

In order to meet the objectives, many factors need to be taken into consideration, among which:

* The current **market,** such as volatility, interest rates, market sentiment et cetera.
* The **risk** of the various portfolios, typically in term of greeks.
* The **counterparty,** which can be at a higher of lower risk.

#### Machine Learning as a solution

Reinforcement Learning is applied to optimize the AMM pricing decisions. This has a double advantage:

* The possibility of using the most advanced **trading algorithms** to have a competitive edge on the market.
* Increasing automation allows to **dramatically cut costs,** which in turn allows to be more competitive and apply tighter spreads to clients.