The Estimation Method Of Number Of Trades In Chinese Futures Market Based On High Frequency Data

As an important financial tool to discover prices and avoid risks,China's futures market plays significant role in the construction of Chinese socialist market economy system.In recent years,China's futures market is in a period of rapid development,and the total volume of transactions has increased significantly year over year.In the past two years,China's futures market has sped up the pace into the international market and gradually made an impact internationally.Meanwhile,investors increasingly value more fine granularity of the data as the information technology and data science develop rapidly.Unlike the order by order data provided by some international markets and China's stock market,the most detailed data provided by China's futures market is high-frequency snapshot data by intervals of 0.25 to 0.5 seconds.In the process of order by order data to snapshot data,some information is discarded or lost,including number of trades.This information has proved to be very valuable in recent years and is a good indicator of market activity.This paper focuses on proposing a variety of methods for the first time to use snapshot data to estimate the number of trades over a past period of time.Specifically,based on the poisson distribution hypothesis of the order flow,multiple estimators are established,using the high-frequency snapshot data.And the monte carlo method is applied to generate the simulation order flow according to the hypothesis.And It is found that the estimated value can converge to the real value.These can verify the validity of these methods.After further testing the methods by running on the order by order data of Chinese A shares,it is found that the estimator is highly correlated with the real value although the high-frequency data of A shares market does not exactly apply to the poisson distribution hypothesis.Furthermore,in this paper,the original poisson distribution hypothesis is modified,explanatory variables are added and a new model is established.The method of two fitting model are developed and got verified in A shares market.we also empirically tested the prediction ability of number of trades on price volatility in China's futures market,and found that the introduction of this data into GARCH model could not only improve the model,but also better predict the volatility than the trading volume data.As far as we know,the multiple estimation methods proposed in this paper fill in the blanks for relevant theories.The methods help in filling in the missing data caused by slicing data of China's futures market.And,the estimated value coming from this unique estimation method doesn't heavily rely on the trading volume which is highly correlated with the estimated value.That's the reason the methods can estimate not only the number of trades,but also the average volume of each order.The methods effectively improve the transparency of the market and provide new methods and perspectives for measuring market activity and liquidity.