The only way to know about trailing stops is to program them and then backtest. I use Amibroker and it's an excellent piece of software. I don't think there's any way around this.. you need to code stuff and then see how changing parameters affects the performance.
For example, I hear a lot of comments along the lines of "move your stop to breakeven as soon as possible". This doesn't hold up in backtesting, it deteriorates the performance markedly in cases.
A popular type of trailing stop is the Chandelier, where the stop (e.g.) for a long position is x * ATR below the highest high for the last 20 days. 'x' is usually set to 3, but you can of course vary it. It's a pretty effective method and can keep you in trends for a long time, but of course you will always give back some equity at the end when the trend reverses.
There is no way to avoid this "bend at the end". The trailing stop is the thing that keeps you in the move for a long time, but it will always be at the expense of handing some money back.
It's impossible to pick turning points, so don't even try. But think about this.. if a FX rate has a large move of 15 pct and you capture 8 pct of it through a trailing stop method, isn't that pretty good anyway? The guy who tells you he got all 15 pct is a liar anyway.
