If you are going to risk 1% of the $5,000 per trade that would be $50 per trade to start.
What you really need to know is your expectancy. This depends on your implementation. How do you get in and out of trades? Using a backtest, you should have an expected value for the number of trades you will do in a given period of time. Will you do 100 trades in a year? 500? 1000? And what percentage of those trades will lose? 40%? 50%? 60%? If you do 100 trades and lose 60 of them with a $50 loss each time, you need to win $75 on the winning 40 just to break even.
If you give some more details about your setups, there are some savvy peeps in this forum who can help you get an idea about your expectancy and how different position sizing algorithms and risk percentages will affect the long term outcome of your effort.
Without knowing those details, I will start you with my big lesson from last January with the SNB shock: keep most of your trading capital in a local bank, give your fx dealer just enough to do your trades. I currently have 60% of my capital in a local bank.
Another thing to consider: the percentage of your trading capital you risk each trade does not only effect your drawdown size, it can also make your total expectancy negative! An example is a coin flip game: heads you win 6:5 odds tails you lose whatever you risk. We look at that and think: "As long as I risk not enough to go bust with a string of tails, this game is an easy winner!" Wrong. If you risk the same percentage each flip and that percentage is greater than 19, you will lose more the more you flip. It is a counterintuitive mathematical reality.
So you really need to know about your system's expectancy or have a method that will make adjustments to prevent the turning of a positive expectancy to negative.
May the forx be with you.