作为特殊的条件随机场模型的隐马尔可夫模型是大奖章基金辉煌业绩背后的秘密武器之一，本文从金融程的角度出发，将隐马尔可夫模型引入到投资领域来预测个股未来的涨跌情况。假设涨和跌的股票各自都存在一种明确的模式，都分别可由一个HMM模型来描述，那么如果一个股票在表征上涨模式的HMM模型上的观测条件概率越大，说明该股票实际上涨的概率也越大。基于HMM模型构建了个股的HMM因子，并在沪深300成分股中通过在2016年到2018年历史回测的实证分析取得了不错的超额收益，从而说明条件随机场引入到量化选股中具有一定的预测能力。
As a special Conditional Random Field model, the Hidden Markov Model is
one of the secret weapons behind the brilliant performance of the medallion
fund. From the perspective of financial engineering, this paper introduces the Hidden
Markov Model into the investment field to predict the future rising and falling
situation of individual stocks. Assuming that each of the ups and downs has a
clear pattern, which can be described by an HMM model, if the probability of a
stock's observation on the HMM model characterizing the rising pattern is
greater, the probability that the stock actually rises is also the bigger.
Based on the HMM model, the HMM factor of individual stocks was constructed,
and the empirical analysis of the historical back testing of the Shanghai and
Shenzhen 300 constituent stocks from 2016 to 2018 achieved a good excess return,
which indicated that the conditional random field introduced into the
quantitative stock selection has a predictive ability.
Zhang, L. and Ji, Q. (2010) Image Segmentation with a Unified Graphical Model. IEEE Transactions on Pattern Analysis and Machine In-telligence, 32, 1406-1425. https://doi.org/10.1109/TPAMI.2009.145
[6]
Zhang, L. and Ji, Q. (2011) A Bayesian Network for Automatic and Interactive Image Segmentation. IEEE Transaction on Image Processing, 20, 2582-2593. https://doi.org/10.1109/TIP.2011.2121080
