A Trading Strategy Based on the Lead–lag Relationship Between the Spot Index and Futures Contract for the Csi 300

A Trading Strategy Based on the Lead–Lag

Relationship Between the CSI 300 Index and Futures

-Xianrao Zhu

1. Introduction

As we know that there is a relationship between the futures price and underlying asset:

[pic 1]

Also stock index such as S&P 500 and CSI 300 and index future should follow this rule. Otherwise there will be arbitrage opportunity. Abhyankar’s (1998) finding of a lead-lag relationship between FTSE 100 spot and futures of 5 to 20 minutes. Therefore, can we explore an investment strategy by using this relationship? Emerging market like China is a good object because as the financial market is not mature, few people are considering this relationship and the strategy based on this.  Hence we choose CSI 300(300 stocks traded in the Shanghai and Shenzhen stock exchanges) as the sample. We choose every 10 minute data because higher frequency causes higher transaction cost but lower frequency causes this investment strategy disappeared.

The process of this strategy is first, develop a model that has the best good of fitness. Second, use different trading strategies based on this model and see the result.

1. Data selection

The newest data can better reflect the relationship between index and index future. Therefore, I selected every ten minute CSI 300 prices from 9:40 to 15:00 on April 07, 2016. There are 24 observations. (The reason I do not select data pool is that there is no place to download the data for free so I had to input the data manually.) Similarly, I also selected every ten minute one month CSI 300 future from 9:40 to 15:00 from Feb. 23, 2016 to April 06, 2016. There are 1536 observations. The reason I choose one month future is because the nearest future has higher liquidity and therefore, can better reflect the real price in the market. As there are some data missed, I replaced it by using previous one minute price instead. Then I take log of them. We get lnSt and lnFt.

1. Econometric analysis, methodology and results

As we want to find out the relationship between St and Ft and Lead–Lag relationship, the VAR (Vector Auto Regression) model is a good model to use.

The equation of the VAR which has the spot returns as dependent variables may be expressed as:

[pic 2]

Where St is the spot price for CSI 300 and Ft is the CSI 300 index future price and Vt is the residual term.

Now we need to decide the lag length. The result of information criterion is shown below:

From the table we can tell that AIC, SC and HQ criterion tell us 6 periods of lag is the best number of lags.