STAGE Spatio-temporal classification of China’s stocks based on Rényi difference matrix

Problematic
At the end of this research, we must be able to know what China’s stocks might be governed by similar dynamics and
on which time scales. By dynamics we mean:
- What is the degree of non-Markovian process for each stock and for given timescales?
- What is the degree of predictability of each stock on given timescales?
- What is the difference of dynamics'behaviors (i.e. complexity measures by the Rényi entropy) between each
stock and for given timescales?
- What type and group of stock provide the less information (in the sense of Shannon) on its future behaviors
and for given timescales?
Answer, Solution
By being able to answer these previous questions, we should be able to provide strong quantitative answers regarding
fundamentals topics in quantitative research namely:
- On what individual or group of stocks from China’s Index, “volatility models”, “noise models or “flow
trading strategies might be suitable to apply, and regarding which timescales?
- What individual or group of stocks from China’s index may exhibit on high frequency timeframe, volatility
behaviors with less predictability, weak memory process, and therefore might be more suitable for long term
investment strategies rather than “flow trading strategies”?
Perspective of future work
The outcome of this research can potentially lead us to work on a more ambitious question namely:
- Is it possible to develop a new approach to the Portfolio Allocation Theory, not based on correlation matrix
that does not take into account the underlying behaviors of stock volatility but on complexity matrix?
- Is the “Complexity Matrix approach more competitive to the well-known “Random Matrix approach to
Portfolio Management?