People make things more complicated than they need to. Just about any style of model can be made to fit a set of data, provided you give it enough variables to tune. That doesn't mean that particular model is of any use whatsoever for modeling the future behavior of that system. You need to have some rational basis that can explain WHY you think the market does and will continue to behave in a certain way.
Like any indicator you need price to move the indicator. The indicator doesn't move the price!! Take a look at the EW Rules. These rules are requirements and if they are violated you are expected to change the wave count. That is if price doesn't conform to what you expect it to do, you change the wave count so that it fits.
I've tried using it and found that it is not determinate, in most cases. There are at least two competing wave counts at all times. Waves can extend and extend with subdivisions of internal waves, and the only real way to find a pivot is the old fashioned system of lower highs and lows, HSR failures and trendline breakout retracements. Momo oscilators also make it look cool on the chart when it all works out. The only real tradable situation I ever found was to enter at the end of wave 2, especially if the next smaller order is also at the end of its wave 2. A third wave breakout usually satisfies all the conditions above, but why go to the trouble counting the waves if you can find pivots the old way. In the end, I found it to be a useful puzzle to make me focus on the chart for more than 5 minutes before betting any money on an idea. The waves are very subjective. A sloppy 5 wave up can also be the start of a B wave down. It all depends on the overlap, and you don't need to count the waves to see that!