Back Fitting or Back testing?

Im familiar with the term back fitting and have been guilty of it in the past and possibly even right now. The way i understand it is that its when one fiddles with data to make a system work.

One thing I have been doing is using historical records on the local (NZ and Aussie) racing websites to test if my systems will work. So basically I have my win/fav system with the bet doubling and stop protection mechanism and I put all the results from the last year or sometimes 2 years into a spreadsheet and see how it works. To my mind thats not back fitting, thats back testing.

With respect to my fav/place system I did the same thing with respect to inputting all that data into a spreadsheet and applying rules but in this case I then fiddled with the staking process till I got positive outcomes. So that to me would be back fitting?

Except for one thing, I was trying to find out an appropriate staking plan to make it viable (well not viable in terms of making a living). So am I back fitting here or working out an appropriate staking plan based around the fact that I know the place dividends are so low but the frequency of positive returns is reasonably frequent.

Im 50/50 on this. I think its back fitting but not done to delude myself if that makes sense. However thats debatable....

I have started to write a posting about how my fav/win system can work but also how in real world terms it wont work. The trouble Im having is putting it down in a coherent manner. I have been researching this probability based process for months and months, and have had plenty of great advice from people via forums and emails but never anything that truely showed me where it falls down. I think I have a handle now on  a particular weakness and as soon as I come up with a coherent explanation I will put it down here.