Although it is important to test any trading model over periods of both upward and downward market movements, I was not able to do so over an entire month period. For the past several months, the cryptocurrency market index has consistently risen. A graph of the total market capitalization of all cryptocurrencies is shown below. The market cap increased from roughly $11 billion on September 1st, 2016 to $36 billion on April 30th, 2017. Since I only had 1 year of trade data, every testing period I could have used either experienced an increase in the market index or did not have enough data before that period to effectively train the models. Even though a full month of negative price movement was not tested on, there were still periods of several days that experienced a significant downturn. An example of a downturn can be seen between April 3rd and April 12th in the profitability graph on the previous page. Fortunately, each tested model actually performed better over this period than over any other part of the month. This is likely because the trading algorithm can ‘short’ cryptocurrencies, effectively having a negative position, to make money whenever the market becomes temporarily overpriced.