Commencement speeches are rarely inspiring or thoughtful, but when I was senior in high school one of my teachers whose name I long ago forgot said something that has stayed with me for life. “Success,” he said, “is fleeting. But failure. Failure is what we live for.” We looked utterly befuddled but he went on. “Think about success. It is certainly nice, but once we achieve it, we inevitably take it for granted and after savoring it for a few hours or a few days we move on. Failure on the other hand is what drives us. Failing is what forces us to strive. Failure is the key to all progress.”

That ‘s why I am such a firm believer in experimentation. Experimentation is basically failure that leads to progress. Unhampered by the stigma of failure children are the purest experimenters of all. Watch a toddler for a while trying to learn how to walk. Does he quit just because he falls over, once, twice or two hundred times? Of course not.

That's why I am such a firm believer in experimentation. Experimentation is basically failure that leads to progress. Unhampered by the stigma of failure children are the purest experimenters of all. Watch a toddler for a while trying to learn how to walk. Does he quit just because he falls over, once, twice or two hundred times? Of course not.

On the other hand, if children acted like traders, the human race would never walk. Traders hate to fail. That's why they hold on to losing trades, revenge trade with abandon and ultimately quit out of sheer frustration. However, experimentation allows traders to fail safely. It gives the trader the opportunity to try new setups, tinker with ideas and generally be free to engage the market without the stigma of loss.
Don’t get me wrong you will lose money when you experiment, but as long as you set aside a small but meaningful amount of capital (the loss must hurt enough to make you care, but not enough endanger your net worth) you can experiment all you want.

I am often asked why not just demo trade, or simply run some backtest simulations? Those who know me well, know that I despise backtests. They are worse than nothing because they give you false hope. If I had a dollar for every brilliant system that backtested beautifully only to fail in real time, I would be richer than George Soros.

No matter how many elegant statistical tricks you play backtests are simply dogmatic structures overlaid on historical data. One of my first jobs in the business was to oversee the execution of hundreds of algorithmic programs during the overnight shift. After 12 months, not one out of the hundreds programs that I monitored (all of them designed by the most robust of backtests) – ever made money.

As to demos, they always remind me of the simulators I saw at the New York auto show last week, Nice but utterly useless for real driving. Why? Because just as in a simulation, you never actually feel the fear of coming within inches of the racetrack wall, or experience the overwhelming smell and noise of high performance race car, so too on demo you never actually feel the pressure of holding money on the line while price flirts within a pip of your profit target or stop.

How will your ideas perform in real life? You will only know if you experiment in real markets with real money. Experimentation lets you know very quickly if your idea are worthy or dumb. More importantly the process of experimentation often challenges your basic assumption behind your setup. Failure is feedback and experimentation allows to study that feedback and become a better trader.

Boris Schlossberg serves as director of currency research at GFT, and runs bktraderfx.com.