Transcript:
Dr. Anna Becker Speaking:
There are several definitions of algotrading and many types of algotrading. Before AI computerization, algotrading was called systematic trading, and traders would sit next to monitors full of charts, news, and other information, make decisions when to buy and to sell, and then send trades through one of the trading platforms. But then, algotrading came along, which replaced both parts, algo and trading.
Computer algorithm is now looking at assets and deciding when to buy and when to sell. And once decision is made, the trading, the execution is done automatically by directly connecting to exchanges. It is very important to understand each trading decision, each buys and sell, and each order execution is done by a computer program without human interference.
There are many pros to such an approach. There are no emotions, no human mistakes. We can trade thousand of assets, make trades without delays, et cetera, et cetera. But also, there is a small con. AI and computers don’t have gut feelings. They do only the things that we expose them to. Our main effort is to expose our AI technology and help it with as many ideas for trading and examples as possible.
I want to give you a feel for what algorithmic trading world looks like. On the right side, are the algorithms. We start with the decision. Where to look for trading opportunities, what assets to trade, like crypto pairs, Bitcoin, Ethereum, or maybe fiat, or maybe commodities. Assets contain opportunities. We look for what affects the markets; prices, volumes, who trades this asset, patterns, market liquidity, economy, news, tradist psychology. Then we decide on trading frequency. Will it be one millisecond or one month or anything in between? Once we made our setup, we’ll look for the model. Sometimes models are called black boxes, but our Endotech policy is to have fully transparent boxes that we understand the reasoning behind each buy and sell behind it.
First and foremost comes the idea. How to look for opportunity. What does opportunity look like? Is it a trend? Is it a resistant support belief? The fundamental economy structure has it affects the prices, options formula, seasonability, some mix of technical analysis indicators, price correlations.
The idea is the king in modeling. But to build upon the idea, to make a durable trading model, we need a framework. This starts with setting a goal for the model. As we said, the technical goal is to buy low and to sell high and to do it over and over again.
Another goal is to make returns. But as I’m explaining over and over again, these goals are too simplistic and are not helping us in modeling the markets and trading. We need the framework in order to create durable models that will stay fit and profitable over time. Instead, we have the following list of goals with constraints.
Then comes the exciting part. Our AI technology starts to train our models to be fit and stable. It tests, and optimizes them. And together with quant operators, decides on the best ones. Once Algo is ready, we go to the execution part. On the execution side, we need to make sure that it is scalable. That model can actually trade on the capital for all of the account is subscribed to them without having a reduced performance. On the execution part, in addition to the order management, we’re also taking care of portfolio management, money allocation, risk management. In the end, algotrading, as we define it by Endotech, is the systematic taking of trading opportunities while mitigating risk.
