Need help evaluating the efficiency of a trading system

Hey fellow traders, I need your help in here to evaluate a trading system which I backtested this last weekend and started implementation this week in a demo account.

The backtest period started on the 1st of April and ended on the 29th of April 2005. I know this is not a large period, but the system uses the 1 hour chart mainly and I couldn't get 1 hour historic data.

Anyway, here are the results of the backtest as well as the results of the first 2 days of the demo implementation:


Back Test Period:


Total number of days: 21
Total number of trades: 12

Average trades/day: 0.57


Total number of profitable trades: 7
Total number of losing trades: 5
Total number of breakeven trades: 0

Percentage of Profitable trades: 58.333%
Percentage of Losing trades: 41.666%

PS: This percentage is the only thing that I am worried about, as the number of losing trades is very near to the number of profitable trades.


Total profits: 308 pips
Total Losses: 32 pips
Net P/L: 276 pips

Average profit/trade: 44 pips
Average loss/trade: 6.4 pips



Implementation Period:


Total number of days: 2
Total number of trades: 2

Average trades/day: 1


Total number of profitable trades: 1
Total number of losing trades: 1
Total number of breakeven trades: 0

Percentage of Profitable trades: 50%
Percentage of Losing trades: 50%


Total profits: 29 pips
Total Losses: 12 pips
Net P/L: 17 pips

Average profit/trade: 29 pips
Average loss/trade: 12 pips


I want to hear opinions from you guys, I am worried about the win/loss ratio, but if we are talking money wise, the profit/loss ratio is great provided that 276 pips are made per 21 days averaging 13.1428 pips per day, this is great.

At this point the only thing you need to be concerned about is getting enough history to validate your system. Some very successful systems have more losers than winners. The real problem is finding a system that handles changing market conditions and most significant market conditions don't change much in only a month or two.

FT

12 trades is not enough to know if a system is good or not. you need more like 1000 trades to validate your results.

as a side note, make sure the system performs similar when traded on different currency pairs. for instance, if you have something that works on the GPBUSD, and also works on the EURJPY, that would be a good sign. and dont believe your results unless you have gone through the proper development process. otherwise you will have hindsight all over the place and you wont even know it

I suggest you refer RR's thread:

http://www.forexfactory.com/showthread.php?t=152651

i just wonder why there is no member from stats in finance major?
just use exact z-test or t-test to evaluate the system. normally 30 trades is the minimum requirement.