Ahora bien, podemos usar un Modelo de Markov para intentar predecir los mercados? O al menos para mejorar nuestro conocimiento sobre las probabilidades de cambio del estado del mercado?
Veamos unos estudios sobre el tema:
- Tesis doctoral, Yingjian Zhang B.Eng. Shandong University, China, 2001: PREDICTION OF FINANCIAL TIME SERIES WITH HIDDEN MARKOV MODELS
- https://www.cs.sfu.ca/~anoop/students/r ... thesis.pdf
The end of the 400-day training and testing period generates a return of 195%.
- Universidad de Stanford: Stock Forecasting using Hidden Markov Processes
As a result, we can verify that the Double HMM can predict better than the Single HMM and its transition matrix verifies the effectiveness of Double HMM.
However, the volatility index data can be a good candidate to extract the economic situation because it can gives us direct estimation of variance. As we mentioned in the introduction, the macroeconomic data is also directly connected to the economic situation too. By testing the effectiveness of those candidates, we can improve the predictability more. Moreover, if we
can construct a choosing algorithm that chooses the most effective candidate by learning, we might build efficient trading machine.
- Universidad de Berkeley: Analysis of Hidden Markov Models and Support Vector Machines in Financial Applications http://www.eecs.berkeley.edu/Pubs/TechR ... 010-63.pdf
However, as we see from the results, we were able to train the SVM to accurately predict the S&P500 up to 70%!
(Ver imaganes adjuntas)
- Desde South Korea el libro: Forecasting Change Directions for Financial Time Series Using Hidden Markov Model
: http://link.springer.com/chapter/10.100 ... 02962-2_23
Experiments showed that our method forecasts the change directions of financial time series having dynamic characteristics effectively.
- Desde India y Oman: Stock Market Trend Analysis Using Hidden Markov Models
Six optimal hidden state sequences are generated and compared. The one day difference in close value when considered is found to give the best optimum state sequence.
One day difference in close value when considered is found to give the best optimum sequence. It is observed that at any point of time over years, if the stock market behaviour pattern is the same then we can observe the same steady state probability values as obtained in one day difference of close value, which clearly determines the behavioural pattern of the
- Desde China: Using Hidden Markov Model for Stock Day Trade Forecasting
The HMM model is a very population model in the speech recognition domain. This work approve the HMM model have a very well forecasting in the stock recognition domain. Especially, the TAIFEX futures from the opening price recognized the closing price.
However, this work has a very conspicuous by the statistics testing of significance. The HMM model is beater then Random Walk model and the Modified Trading model. But this work only has a blemish in an otherwise perfect thing, which the risk is same with the Modified Trading model. Additionally, this work approve that long-term trend of the HMM model will be a positive slope
line. In other words, the revenue will be increasing.
In clusion, the HMM model is a feasible and stable model in the TAIFEX futures day trading. The HMM model will have positive revenue in the long-term time series.