When a trader finds a new strategy using a new indicator, what does he do with it? Does he just observe it for a few weeks and see if he can get a feel for it? Or just straight out trade it in real time and see how verify its profitability by seeing if the equity is better than it was before trading it?



In automated trading, a programmer can code and give straight forward signals based on the indicator where to enter and exit and have it tested against historical data. Unlike auto trading, discretionary traders have a harder task of having to be as objective in observing and identifying where the entry or exit signals take place, then writing it down and later calculating the overall performance of the testing.

The testing of the strategy in a conventional and even unscientific way can be extremely arduous and time-consuming task. It is especially frustrating when there are so many indicators and strategies available to go and verify each for its potential in making profits that it's disheartening to go through the testing one by one. Just that thought would entice the trader to take the shortcut.

One way to do this is testing it by demo/paper trading. For a couple of weeks with a twenty-something or thirty-something trades, he would have a good idea how it works, how much it makes or loss. If he thinks it's ready for real trading, he can start. The only problem is that the strategy has only seen a current market condition, it may work in the next few months, but when a new market condition emerges, such as from an uptrend to a consolidation stage, this may cause a period of losses, which may lead to big drawdowns. Only using the demo stock trading may not be enough. Plus the sampling size is so small to have adequate judgment on the validity of the performance results.

Going the opposite direction, a trader may decide to backtest with historical data and review the possible profitability of the strategy. This might be a great way since the historical data provide more than adequate periods of time, normally in years. The sampling size may be in the hundreds of trades, with it in many different types of market conditions. With this type and enormity of research, this would be more than sufficient to decide that there can be no doubt profitable once it's put into real trading right? Wrong, historical data in charts do not completely represent what and how it looks like in real time trading, no matter how accurate the data is. There are many elements where the data was "cleaned" or "corrected." It only happens after the fact. But the biggest hidden factor are the indicator: indicators are in constant real time motion. In backtesting, the indicators are already resting in place so this eyeballing old data with that indicator with historical data is like "shooting fish in a barrel," and not shooting fish in the ocean. It is well known that indicators lag so when the data is already shown and known, the indicators would show what it would have been like look like in real time. But that is not the case when the real data is still unfinished. This is one reason why many trading systems tested against historical data fail in real time, automated or discretionary.

In order to make sure that the strategy works well with real money, it needs to be tested with historical data as well as in paper trading. These two tests then need to be compared to carefully find if there are discrepancies not only in performance results, but also the signals for entries and exits. If there differences do exists, historical data testing need to be revised, retested and corrected.

Taking shortcuts may prove a doer to be a clever person in the real world, but in the trading world, it can be a costly lesson. In fact, there are no shortcuts in trading; it's a long, winding hard road filled with potholes, in conditions with little visibility, with many toll booths along the way. Do it right and the road will be more enjoyable and profitable.