That's rather oversimplifying it...most things are. Saying that trading is 'all about probability' is unhelpful, misleading and not to mention untrue. There are such a lot of factors involved in moving the values of currencies that to try to pin it down with stats alone is a sure way to wiping out. I speak from experience and nowadays don't have that risk...and I promise you that it's way more than maths alone that gets my results.
It's untrue if you have some form of fundamental analysis, but if you are purely technical then it's 100% true. Technical analysis is purely probabilities.
No matter what sort of technical analysis you use. Whether it is patterns, advanced patterns, support/resistance, trendlines, candle formations, RSI, MAs, Stochastics, whatever, you are looking at the history and saying, when this has occurred in the past this has happened and the probability of it happening again is higher than it not happening again. Sure after that other factors such as money management, risk, lot sizing etc may affect the result, but the bottom line is that it is probabilities when using only technical analysis.
Is it gambling? Depends on the connotation of gambling. I personally don't think it's gambling if you trade a system that proves itself to have a mathematical edge.
Do casinos offering all their games gamble? No they don't. They know their games have a mathematical edge and in the long run, very important, in the long run they will make money and plenty of it. Sure they can have some variance (Drawdown) and sometimes large variances (must have a bankroll large enough to sustain the drawdown) but eventually the probabilities of their mathematical edge will cut in and they will make their expectation every single time. Nothing more, nothing less.
At any given month one particular table could be losing money due to variance, but they don't offer one table, they have many tables to overcome the variance, to overcome the drawdown.
Likewise, a trading system with a few weeks, a few months, low volume doesn't really tell you anything. The sample size needs to be reasonable, yes that is subjective, but to make a judgement of a system, you need some reasonable time, at least a year with a reasonable amount of volume, before the stats can tell you anything about the system.