HMM_predictor

• Use Viterbi method or Comparison method to predict, seeing hmm_model.py.

• Make some trading decision via the prediction, seeing trading_strategy.py.

• plotting.py gives some plotting functions, like plotting the change of asset and the point of buy and sell.

Key Concepts:

• States: The system can be in one of several possible states, but the true state is not directly visible (hidden). These states evolve over time according to a Markov process.

• Observations: While the actual states are hidden, we observe some data that is probabilistically dependent on these hidden states. These are the observable outputs.

• Transition Probability: This defines the probability of moving from one hidden state to another.

• Emission Probability: This defines the probability of observing a particular output given the current hidden state.

• Initial State Probability: The probability distribution over which state the model starts in.