How does your system compare with 9 trading signals created by randomly assigning them over that same time period? If you are using data pre-March 2011 it would be hard not to lose buying bitcoins.
It's a very conservative method (I like my sleep). I've compared against a buy & hold position of the same size and over the same time frame. Participating in all 9 trades up to the current USD/BTC rate of ~$4.40 results in a 13% increase over B&H.
For the 4 trades that I have made using the system, I'm up 15% compared to a buy & hold strategy. That's after accounting for the additional funds required above the initial 100% position (funds were added 3x before the first sell signal).
This is using data from bitcoincharts.com and assessing weekly activity, so if a trade is indicated, it will only be executed at (or ASAP after) the week's closing number is in. Here are the trades:
| Date | | Trade | | USD/BTC |
| 2010-10-11 | | -10.00% | | 0.10 |
| 2011-02-07 | | -10.00% | | 0.92 |
| 2011-05-16 | | -10.00% | | 7.20 |
| 2011-06-13 | | -10.00% | | 14.65 |
| 2011-08-08 | | 10.00% | | 6.55 |
| 2011-09-12 | | 10.00% | | 4.77 |
| 2011-10-24 | | 10.00% | | 3.16 |
| 2011-11-21 | | 10.00% | | 2.20 |
| 2012-01-09 | | -10.00% | | 6.75 |
The date is for the start of the new week - opening is the same as the prior week's closing; positive trade percentages indicate buys and negatives are sells; USD/BTC rates are the prior week's closing and current week's opening value which is the specified trade price.
As Bitcoin continues to grow, I'll eventually shift to a monthly basis.
