Can I take mean of correlation coeficients for equally spaced data sets?

Historic market (Cash) prices and future contract prices are available for last 4 years. I have found correlation between Jan'11 market price with Jan'11, Feb'11 and March'11 future contract prices separately. I am planning to do the same for all months till Dec'14. So I will have 3 coefficients per month, 36 coefficients per year and 144 coefficients for 4 years. Now I want to integrate the data, that means 1) To find only 3 coefficients for each month using total 12 coefficients of January of 4 years. Should I just take mean? Please let me know. 2) How can I incorporate effect of external factors like weather, volumes traded?

Please let me know if I am using wrong approach.

• You wrote "I have found correlation between Jan'11 market price with Jan'11,Feb'11 and March'11 future contract prices separately." So that's 3 correlation coefficients right there? What are you trying to do here? – Taylor Apr 22 '15 at 5:54
• @Taylor I want to establish relations between spot price and future price.I am building a model which would aid me in future contract buying decision? e.g. Today,stock is trading at certian price and I want to make a decision whether to buy May'15,June'15 future contracts or not.Please let me know if my approach is wrong. – user2122922 Apr 22 '15 at 6:44
• What kind of futures? How are your data sampled (daily, monthly)? Are you trying to model the spread or are you just trying to forecast future prices. Keep in mind the second one is pretty difficult :) – Taylor Apr 22 '15 at 6:54
• @Taylor 1) Monthly Future contracts used for hedging 2) Data is daily sampled. 3) I want to forecast future prices. I wish I could post some spreadsheet data for better understanding.But I don't know how to put image/formatted text here. – user2122922 Apr 22 '15 at 7:20
• @Taylor Feel free to suggest any other testing method to attain the objective. – user2122922 Apr 22 '15 at 7:24