Masafumi Nakano

Generalized exponential moving average (EMA) model with particle filtering and anomaly detection

We propose a new exponential moving average (EMA) model in a state space framework.We develop 3 anomaly detectors with a particle filter used for investment decision.We implement investment analysis with our method by using global asset price data.Our scheme outperforms practically well-known strategies including standard EMAs. This paper proposes a generalized exponential moving average (EMA) model, a new stochastic volatility model with time-varying expected return in financial markets. In particular, we effectively apply a particle filter (PF) to sequential estimation of states and parameters in a state space framework. Moreover, we develop three types of anomaly detectors, which are implemented easily in the PF algorithm to be used for investment decision. As a result, a simple investment strategy with our scheme is superior to the one based on the standard EMA and well-known traditional strategies such as equally-weighted, minimum-variance and risk parity portfolios.Our dataset is monthly total returns of global financial assets such as stocks, bonds and REITs, and investment performances are evaluated with various statistics, namely compound returns, Sharpe ratios, Sortino ratios and drawdowns.

Creating investment scheme with state space modeling

A unified approach to create investors desirable portfolio.A new interpretation for state-space model to attain various investment objectives.Particle filtering ensures the general applicability of our scheme.Numerics: creating alpha over S&P500 and well-performing mean-variance portfolio. This paper proposes a unified approach to creating investment strategies with various desirable properties for investors. Particularly, we provide a new interpretation and the resulting formulations for state space models to attain our investment objectives, which are possibly specified as generating additional returns over benchmark stock indexes or achieving target risk-adjusted returns.Our state space models with particle filtering algorithm are employed to develop expert systems for investment strategies in highly complex financial markets. More concretely, in our state space framework, we apply a system model to representing portfolio weight processes with various constraints, as well as the standard underlying state variables such as volatility processes. Further, we formulate an observation model to stand for target value processes with non-linear functions of observed and latent variables.Numerical experiments demonstrate the effectiveness of our methodology through creating excess returns over S&P 500 and generating investment portfolios with fine risk-return profiles.

Fuzzy logic-based portfolio selection with particle filtering and anomaly detection

This paper proposes a new fuzzy logic (FL)-based expert system with particle filtering and anomaly detection to create high-performance investment portfolios. In particular, our FL system selects a portfolio with fine risk-return profiles from a number of candidates by integrating multilateral performance measures including Sharpe, Sortino and Sterling ratios. The candidates consist of various mean-variance portfolios with multiple time-series models estimated by a particle filter and anomaly detectors. In an out-of-sample numerical experiment with a dataset of international financial assets, we demonstrates our expert system successfully generates a series of mean-variance portfolios with satisfactory investment records.

State Space Approach to Adaptive Fuzzy Modeling: Application to Financial Investment

This paper proposes a new state space approach to adaptive fuzzy modeling under the dynamic environment, where Bayesian filtering sequentially learns the model parameters including model structures themselves as state variables. In particular, our approach specifies the state transitions as meanreversion processes, which intends to incorporate and extend the established state-of-art learning techniques as follows: First, the mean-reversion levels of model parameters are determined by applying some existing learning method to a training period. Next, filtering implementation over test data enables on-line estimation of the parameters, where the estimates are adaptively tuned for each new data arrival based on the obtained reliable learning result. In this work, we concretely design a Takagi-Sugeno- Kang fuzzy model for financial investment, whose parameters follow autoregressive processes with the mean-reversion levels decided by particle swarm optimization. Since there exist Monte Carlo simulation-based algorithms called particle filtering, our methodology is applicable to a quite general setting including non-linearity, which actually arises in our investment problem. Then, an out-of-sample numerical experiment with security price data successfully demonstrates its effectiveness.