Giovanni Trombetta

SIAT Member. Head of Research & Development in Gandalf Project

Head of Research & Development in Gandalf Project, Giovanni Trombetta is an Electronic Engineer, a trading system developer and a quant trader. He has many years of programming experience with several languages and trading platforms and uses his knowledge in the field of genetic algorithms and neural networks to create automatic trading systems and financial models.
In 2009 he has contributed to the drafting of the book “Visual Trader II: Implementing Winning Strategies” (Trading Library). In 2009 he developed “Gandalf” a data mining software to look for statistical inefficiencies on the historical prices of stocks, futures, currencies and ETFs. In 2012 he founded the “G.A.N.D.A.L.F.” project, within which he led the research and development team, specialized in the application of artificial intelligence to the world of quantitative finance.
He is an appreciated trainer, an associate in S.I.A.T. (the italian branch of I.F.T.A.), speaker to several important financial events in Italy (Investment and Trading Forum, TOL EXPO of Borsa Italiana), cooperates in educational projects with banks, brokers and IT Companies.

 

 

ABSTRACT

Artificial Intelligence and Trading Systems.
Robustness and validation of genetic trading systems

Today modern Machine Learning approaches make it possible to sculpt extremely timely trading systems, reducing the development times and by identifying, with extreme accuracy, statistical inefficiencies to leverage on the different financial instruments. In particular genetic algorithms, borrowed from the Artificial Intelligence world, are particularly suitable for discovering hidden rules, not immediately evident with more traditional approaches. Delegating completely the Backtesting and Testing steps to the machine, hides a number of pitfalls. The overfitting of the series noise must be contained through a rigid and reliable operating protocol. The “traditional” trading systems and the “genetic” trading systems will be compared, giving particular emphasis to an innovative technique for validating genetic trading systems.