The trading strategy uses three simple moving averages. An SMA is an average of the values of n number of points before the current transaction.
The three simple moving averages are then broken down into a fast moving trend and a slow moving trend.
A buy signal will be generated when both the fast moving and slow moving trends are bullish (or above 0) and the previous fast moving trend was bearish (or below 0).
A sell signal is generated when the fast moving trend becomes bearish (or drops below 0).
The chart below shows the buy signals (marked B). Note how the buys signals started to occur with a temporary rise in the Bitcoin price.
The chart below shows the sell signals (marked S). Bitcoin has experienced a bearish market in the beginning of 2018 (reflected in the initial downward trend of the price in the price chart above). The strategy does identify much more sell signals than buy signals. The Bot ignores a sell signal when not in a trade.
Below is a chart that shows the closing price for Bitcoin on the Luno.com Exchange. I have lined the timing up with the above SMA.
The chart above shows a bearish market in the first part of 2018 with limited buy signals in the SMA graphs further up.
As more price data comes in over time, machine learning will be applied (using R) to find the best SMA combination to generate buy and sell signals.