期货是现在投资组合中一个重要的部分，其自带保证金，双边交易，T+0 交易的特点使其格外适合量化交易，但目前市面上的期货量化策略框架较为单一，相似度很高，很多几年前表现优秀的量化策略，近些年来表现都很一般，因此市场需要一些不同框架的量化策略，来扩大资金容量，增加量化策略组合稳定性。论文选择使用期货价格时间序列匹配的方法来预测期货价格变化，并以此指导投资行为。期货价格按时间排序形成的序列是一个典型的时间序列，人们经常通过对时间序列的分析来揭示事物内部规律，预测未来变化发展。本文中采用时间序列匹配的方法来预测期货价格变化，在时间序列去噪上，使用了感知重要点（Perceptually Important Points，PIP）特征提取法。对于提取出的样本点尝试了多种不同的相似性定义：欧式距离和动态时间弯曲（Dynamic Time Warping，DTW）等。基于这些投资策略，我们在期货品种聚氯乙烯，玉米和聚乙烯上进行了实测。实验结果表明，在零延迟的交易系统上，这些品种的年化收益率均能达到 90%，这已经显著优于市面上大部分的投资策略。

Futures is now an important part of the current investment portfolios. Characterized with Margin,,bilateral trade,and Transaction plus 0 days, makes it particularly suitable fbr quantitative trading. However, the simplex framework and high similarity makes many quantitative strategies that performed well a ffew years ago have performed just so-so in recent years. Therefore, the market needs some quantitative strategies of different frameworks to expand the capital capacity and increase the stability of quantitative strategy portfolio.This thesis chooses time series matching with futures price to predict the change of futures price and guide the investment behavior. Formed by the time sequence,the futures price is a typical time series.Through analyzing the time series,People can reveal the internal laws of things and predict the future development or changes. In this thesis, time series matching method is used to predict the change of futures price, and Perceptually Important Points (PIP) feature extraction method is used for time series denoising.A variety of similarity definitions have been tried for the extracted sample points: Euclidean Distance and Dynamic Time Warping (DTW).Based on these investment strategies, we tested on futures PVC, corn and PE.The experimental results show that on a zero-delay trading system, the annualized return rate of these varieties can reach 90%, which has been significantly better than most investment strategies on the market.