Quote from: rail233 on January 09, 2014, 09:38:10 PM

Butter Bot's back-testing capabilities are great, but the results are optimised for USD which can make it difficult to compare to a buy-hold strategy.

I've put together a simulation which mimics Butter Bot's trading strategy to hopefully give everyone a better idea of 2 things:

The "best" short and long EMA parameters
The actual increases you would have seen versus buy-hold

Thus far I've only run the simulation on btc-e trades. The 1 min trade frequency simulations take a long time to complete and given my Macbook wasn't a particularly willing participant with the other simulations they're not included.

Essentially every possible combination of short/long EMA between 1-120 are included. Thresholds and tick offset have been fixed to the recommended values.

The results are contained in a number of CSV files with short EMA vertical and long EMA horizontal. The results are BTC optimised given 1BTC was invested at the start of the period. ie. a value of 1.9 means you invested 1BTC and had 1.9BTC (or USD equivalent) at the end. This is also what your return would have seen versus buy-hold.

Check out the results and enjoy

You can download the simulation results here:
http://www.sudosystems.net.au/wp-content/uploads/2013/05/ema-profit-2014-01-09.zip

Exchanges: btc-e
EMAs: 1-120 (short and long)
Test durations: 1 week, 1 month, 3 months, 6 months, 1 year
Trade frequencies: 5 min, 15 min, 30 min, 1 hour, 2 hours
Buy threshold: 0.15
Sell threshold: 0.1
Tick offset: 0
Last trade data: Wed, 08 Jan 2014 23:01:18 GMT

If you find any of this info helpful or are looking for some particular simulations feel free to donate to my trading fund:
1BzDv1bep679dtXchHaWRu9XR5jrTUfFCF

Crunching numbers is one thing but putting it a better format like graphs is another. If you can produce some nice graphics with some critical thoughts that's be better.