Synthetic data generation

The generation of realistic symulated data is of considerable importance in a large set of reaserch areas. In finance and economics, the use of simulated data goes from senario generatio for streess tests, asset pricing and risk control to name a few. The following article intruduces a novel algorithm to generate artificial data applied to the S&P500.

Surrogate Monte Carlo

Math formulation of trading strategies

Time-series momentum (trend-following)

Proposed prototipical trend-following strategy which average performance and standard deviation admits a close form solution.

Initial results were presented at R/Finance 2013 held at UIC.

Latest version was presented at QCMC 2015 held at the Max Planck Institute for complex systems.

The presentation slides with few interactive plots created using Shiny can be found at this link QCMC15.

Final article has been accepted for publication in Quantitative Finance. The most recent preprint has been substancialy extended, edited and updated.

Momentum

Traditional momentum involves a portfolio of assets therefore cross-correlation between assets is important. We are looking into formulations that describe such correlations and their significance.

Sample of older published research

• Stochastic resonance and the trade arival of stocks. The preprint can be found here

• Virtual Volatility. The preprint can be found here

• Stochastic volatility of financial markets as the fluctuating rate of trading: An empirical study. The preprint can be found here

• Exponential distribution of financial returns at mesoscopic time lags: a new stylized fact. This work was also presented at the University of Leiden Lorentz Center Workshop Volatility of financial markets: theoretical models, forecasting and trading

• Temporal evolution of the “thermal” and “superthermal” income classes in the USA during 1983–2001