In the age of Big Data, Financial markets worldwide accumulate tones of data; we need a new method to build a frame so that new techniques like machine learning could have a chance reshape the foundation of the future market. This paper proposes a new frame to take the small step to this big future. This paper presented a new architecture of comprehensive system to define similarity could capture the almost every promising K-line patterns of the stock price with the power of big data techniques, this system allows us to generate patters, predict the possibility of correlated events attached, it is a necessary component in the field of finance technology. With this new tool, we predict direction of stock prices as a test. In order to define the K-line similarity, first we proposed a new coding system to every possible shape of K-line; then we coding the series of K-lines, with the decoding technique and inference approach algorithm we could find the transfer possibility of every possible pattern. Possibility turns to be the “knowledge” of this system. Naturally,” ARIMA with GARCH effect” model and Naïve Predict Model were chosen as the benchmark to test the feasibility of the system. Many specific patterns were thought to be magic in revealing the future of asserts, and the very shapes or patterns were based on experience of experts. In this study I use the data from Chinese stock market, by setting a series of basic method as benchmark. Many evidences show massive patterns search method based on K-line similarity match should be promising, the general predict power of pattern search is better using in massive predict rather than a single assert prediction. This tool is a new path to study patterns and events from K-line series, give a complete frame to train the deep learning networks like Generative Adversarial Network. By using this frame well, “Technical Analysis” will be reshaped.