python: How use the S&P 500 index to predict japan stock,namely timezone issue

Question

I want to use American stock index, such as S&P 500 index(open, close...) to predict japan stock daily close price or other with machine learning. I found that there is timezone between japan and New York timezone, there are about 13 hours difference, I mean Wednesday, 5 July 2017, 19:12:00 in Tokyo time, the New York time is: Wednesday, 5 July 2017, 06:12:00. I mean if I want to use the S&P 500 index which is American index to predict japan stock daily close price, I think I can't use the same day(date) data of s&p 500 and japan stock data to predict the next day close price since there are time difference between Japan and USA. For example, if want to predict 7/6 Japan stock close price, I can use the 7/5 japan stock price data for features, and I can't use the 7/5 S&P 500 index data for features, I should use the 7/4 S&P 500 index data for predicting 7/6 stock price. I want to ask:

(1). Should I use the yesterday or today data of S&P 500 data to predict tomorrow japan stock price.

(2). How to processing the missing value of S&P 500 data which some date(I mean the same date of japan stock price) is missed. How about Friday S&P 500 index.

Thanks!

Answer 1

To forecast something you can use any variable that is known before the result you are trying to forecast.

For example the Wednesday close of S&P and the Wednesday close of Nikkei can be used to predict the Thursday Open and/or Thursday Close of Nikkei. When S&P closes on Wednesday it is about 5 am Thursday in Japan so the Thursday business activity in Japan has not begun (or finished) yet. So a prediction would be meaningful, it would be looking forward in time.

In general you can use today closing index value of S&P 500 to predict tomorrow Japan stock price.