Strategies Must Have Merit

Strategies need to start off with a hypothesis. X team will do this because of Y. Back testing can then test whether these results are true (according to your hypothesis), or not. 

Failing to do that could (and often does) lead to data mining problems where something only works because you've got it into a small situation. These sorts of strategies won't last. The hypothesis method can also lead to mining problems but it's less likely.

In my opinion a lot (not all) of the strategies on here aren't robust enough in their testing or their hypothesis, and shouldn't be followed.

I understand that money management can make a bad strategy manageable, but why not start with a good strategy and then add money management?

Just my 2 cents, interesting to see what others think.