not so turbulent thursday, page-8

Ok one for all the numbers people.
I have been looking at a mechanical system trade on the SPI.
What I have come up with is as follows.
Data Period 2000-2010
Signals Given 512
Losing Trades 228
Winning Trades 284
R:R $1.27 i.e. $1.27 made for each $1 lost
2 NETT losing years over the data set 2001 and 2009

I am not so much worried about the system at this stage, the numbers above are raw numbers at the moment i.e. no stops or money management employed.

The first question I have is what is the optimum data set to use if any? Is it a case of it being impossible to use too much data or does too much data smooth out the results too much?

The second question is in regard to money/ account management, over the data set I have run a compound trading strategy, basically this entails trading a $1 contract (Cityindex) for each $1000 of account value e.g. $36,000 account would trade a $36 contract. Now I have read in a book somewhere previously about the optimum lot size being 3 contracts ($75 Cityindex) so when I ran the compound strategy but capped the Max lot size to $75 the system actual made $32,00 more over the data set compared to commiting to full compound i.e. a $177,000 account balance trading $75 contract as opposed to a $177 contract.
Why is it so?
I know that loses as a percentage of account would be less for a $177,000 account taking the same loss in points on $75 compared to $177 but conversely the wins are limited in the opposite way.
Obviously the time line when the loses come in the data set would have a bearing but any other tips?