We make predictions all the time, so why not in your trading?

Professional traders will backtest and then add new elements or parameters to their existing system(s). 

Markets will evolve also, so you need to keep pace with evolving market environments. That means experimentation with something new. 

You can also test your hunches within the discretionary percentage of your trading. For example, some traders are 90% systematic and 10% discretionary.

Test your predictions and hunches in the 10% discretionary allocation. Just make sure to follow you risk controls, ie, max risk per trade and correlation studies before you put on the trade. 

I think prediction gets a bad rap because anyone who doesn't have a system is effectively guessing at the market. That typically doesn't work out that well for too long until your rules get systematized. For one, you need to have rigid risk management techniques in place. 

It's likely that you will blow up if you put a large percentage of your capital on any one idea based upon a prediction or hunch. Risk 0.50% instead of 50%.

Your feelings aren't facts and it's better to gauge your reasoning with proper risk management. There will always be new ideas to trade, but if you roll the dice on one name based upon a prediction, with no training, you're likely to get the worst of it.