At first, make the code working on your machine.

You need to finish the following tasks in this assignment:

Suppose x0 is the current day price of bitcoin and x1, x2, x3, x4, and x5 are the price prediction for the next five days, using the model in predict.py. We may use these values x0, ..., x5 to design buy/sell rules:

• Buy rule: if x0 < (x1+x2)/2 < (x3+x4)/2 < x5, then place a buy order at the price of the next day;
• Sell rule: if x0 > (x1+x2)/2 > (x3+x4)/2 > x5, then place a sell order at the price of the next day.

To test the above buy/sell rules, in predict.py, you create two accounts: cash and coins. Initially, cash=$1,000,000 and coins=0. For each buy order, if your cash account allows, buy upto $100,000 worth of bitcoins, add them into your coins account, and subtract the value of these coins from the cash account. For each sell order, sell one third of the total bitcoins and add the values of these coins into your cash account. You can place at most one order a day. Use prices for all trading and ignore trading fees.

Test the above trading strategy for the prices of bitcoin from the 200 consecutive days (in the test set) and report your result (how many trades, changes of values in your cash and coins accounts).

• The previous trading strategy is one of many trading strategies. In this task, we will use the average price of a week (7 days) instead of daily prices, to train the neural work and predict the average prices for the next five weeks. Pleas repeat the previous task, except that in the description, "day" is replaced by "week". For the test, use the last 12 trading weeks. Compare your result with the previous task with a short discussion.

Please submit everything required in the ICON dropbox for Assignment 8 before the deadline.

Thank you!