Conclusions

In this project, we first build a simple hind sight market timing model using Dynamic Programming to get an optimal timing solution between two stocks. Then we build an online market timing model using HMM and test the accuracy of its prediction on several different real stocks. Also we come up with a timing strategy to do investment using the prediction of the model. We test the sensitivity of the model to different parameters and use two different evaluation methods: STD and Accuracy, to evaluate our model. The experiment shows in many cases, this model beats the simple "buy and hold" strategy.

Future Work:

We should have done more experiments with the timing strategy part, testing on more stocks, and evaluate the performance by comparing to other benchmarks than "buy and hold". But the experiments are too slow, so we have not done it as completely as we expected.

However, we are satisfied that we learned a lot from this project.

Referencs:

[1]   http://en.wikipedia.org/wiki/Market_timing 

[2]   S. Shi and A. S. Weigend. Taking Time Seriously: Hidden Markov Experts Applied to Financial Engineering. In Proceedings of the IEEE/IAFE 1997 Conference on Computational Intelligence for Financial Engineering, pages 244-252. IEEE, 1997.