Paolo's Blog Random thoughts of a Junior Computer Scientist

Modeling Financial Markets as MDPs

We know that games can be modeled using Markov decision processes and thus be solved using Reinforcement learning algorithms. Can the same reasoning be applied to the financial market? Can financial markets be treated as a game where the utility function is to make more money as possible? Recent papers told us so. Even though financial mark... Read more

Time series forecasting using CNNs

Usually, CNNs are linked to the whole field of computer vision, and that’s because in the past 10 years the best breakthrough in Computer vision is due to the use and improvement of the CNN architecture. Today we are going to see how CNNs can be also applied to the problem of time series forecasting. First and foremost we shall introduce wha... Read more

Building a NG strategy with a Random Forest Regressor

Is it possible to teach an ML model to trade? The short answer is yes. But how well the model will perform with data it has never seen before? That’s the big problem with machine learning and also with trading strategies. As we don’t want a trained model to overfit the training data we also don’t want a trading strategy to overfit the in-sampl... Read more

Coding an ERC20 token smart contract with Vyper

As I’m getting to know more and more crypto space and, especially the Ethereum ecosystem I’m kinda wondering how those smart contracts work. This idea of a programmable blockchain sounds so cool to me! So I’ve decided to get to know to build on of those smart contracts starting from the silliest thing that one person could do: building his own c... Read more

Basics of Reinforcement Learning: MDPs

After introducing Reinforcement learning, a question that may arise is: How are we going to represent those Reinforcement learning problems? Well, mathematics provides us with all the tools that we need, and, for this purpose, it gives us a framework for modeling decision-making in situations where the outcomes are partially random and part und... Read more