There's not a buy or sell every 15 minutes just the opportunity IF the EMAs crossed during that period. Below is a graphical representation of my most recent trades.
The blue line is the short EMA, the red line is the long EMA, and the orange line is the price.
A) Buy @ 143 USD
B) Sell @ 181 USD
C) Buy @ 188 USD
I understand that there's not a buy or sell every 15 minutes.
What I was saying is that when you bought at 143 or 188, it immediately could have dropped after that, immediately triggering the sell and a loss, your system claims "it is not likely to based on this mathematical relationship..."
This could happen at any of your buy points... your claim is that based on long term vs short term averages you are finding good buy points.
Sure, the system mitigates losses intelligently, the arbitrary portion is claiming a long-term and short-term exponential average crossover can divine future behavior of bitcoin valuation...
for example, it would be interesting to test this system on the 30$ -> $2 crash. How much would it have gained/lost if you applied it to a pre-bubble + post bubble burst. What would the returns be in the years after that? etc.
The sell at 188 happened because there was a drop in the price causing the short EMA to pass back under the long EMA, but I do understand what you are saying. If that downward price fluctuation was just a temporary price drop due to a large sell or whatever reason then the EMA may sell unnecessarily. But what you are referring to is called whipsaws.
I do agree I would like to see how highly volatile price changes would affect the ROI. I can run simulations to see how much I would have made using historical data.
(Or close to) 
Just buy and hold would have outperformed you easily. 
