I don't know about "predicting", but if you seek to achieve some level of robustness for your system results, the larger the number of "trials" the better your chances of avoiding curve fitting (all other things being equal).
The better a predictor you become, the more opportunity you'll find in shorter term trades (following the coast argument).
2- Longer term data has been researched to death. Shorter time frames still offer amazingly simple ways of making money.
3 - Fundamental data, is (generally) slow moving (subject to all of the pitfalls described above).
Sorry for the stream of consciousness, but these are some of the considerations that have pushed me to switch to shortening my holding periods.
"The shorter your time frame, the sooner you get to the Long Run"
Happy trading
This you'll only find out through personal experience - give all a try and see what you enjoy the most, whatever you find most agreeable should be what becomes the most profitable.