well, that's just an awful idea.
heh
Yeah, it's definitely going to depend on the quality of the model. As far as I'm concerned, the concept is validated by these guys:
http://imarketsignals.com/2015/spy-ief-market-timer/
They use more than just watching the prices over time, and include inputs like VIX, S&P500 risk premium, EPS, unemployment and other generic economic data, etc.
About as far as I've gotten is pulling historic prices and dividends from Yahoo and adding up total return including dividends over any date range for a given model (my only model so far is "hold stock XXX" but others will plug right in). Next I want to learn how to use matplotlib for aiding visualization, since at work I make extensive use of matlab's plotting functionality. Eventually I'd like to attach this all to a nice GUI since if it works, it will need to be easily usable on a ~weekly basis.
I am definitely liking python so far over matlab. There are some things that suck in comparison, like no command history and reimporting modules after making changes is a pain, but there are definitely solutions for that -- write a history(n) function that displays last n commands and using test functions instead of the interactive shell that automatically reimport, etc. So really just a change in working style.