The goal of evaluating any aspect of a trading plan is to minimize risk, create consistency and generate greater profitability. Even after a system “checks out” during testing, it is important to continue evaluating a system during live trading. Measuring live trading performance allows you to determine how good your trading plan really is.
One method to determine if live trading results correlate to testing results is to compare equity curves. An equity curve shows profits and losses over a specified period of time. Ideally, an equity curve is generally upward-sloping; it is normal for the curve to fluctuate, but the overall trend should be up. If your testing results show this ideal equity curve but your actual trading results show a downward sloping curve, the results are not correlated. If, on the other hand, both equity curves show steady growth, you are on the right track. If live trading results do not correlate to historical and forward performance testing, it may be necessary to revise the trading plan. This may entail making changes to the plan that decrease historical performance while allowing more realistic expectations for live performance.
Even systems that have been appropriately tested and show a high level of correlation between in-sample, out-of-sample and forward performance testing can perform not as expected in a live market. With experience, it becomes easier to figure out why. Common explanations include:
Trader Error (Pilot Error)
Even the best trading plan can fail if the trader is unable to effectively trade it in a live market. This may be the result of a trader not sticking to the trading plan or actual order entry errors. Many traders can recall the time they accidentally went short instead of long, or added to a position instead of closing it out. Practice, experience and using some type of automated trading can help mitigate these types of problems.
When you test a trading system – whether on in-sample historical data or during forward performance testing – you count on the computer to “fill” your hypothetical orders. In real trading, trades will experience slippage and may not even get filled at all. For example, a trade entry in historical testing might get filled at the absolute high of a bar, resulting in a profitable trade. In real life, however, this trade may not have been filled at all since only a few transactions occurred at that price. Reviewing your historical trades may help you find these instances, and it is sometimes best to assume they would not have gotten filled in real trading.
Technology is a powerful force in trading and it is what allows independent traders to work from the comfort of their own homes. That said, technology does fail and it can lead to unexpected losses. You could be in a stellar trade, for example, only to be disconnected from the trade server. Or you could lose your Internet connection or an exchange could close. While these occurrences are thankfully rare, they do happen and they can make your live trading results less attractive than historical testing.
Unique Trading Conditions
Economic and political events can have an immediate and profound effect on the markets. While regular announcements such as the weekly Petroleum Status Report can cause a short-term drop or spike in price, unexpected news – such as a natural disaster or war – can have more significant consequences on market activity. No trading plan can account for these situations and, as such, your live trading results may fail to match your testing results.
TradingCorrelations between backtesting and forward performance testing results can help you optimize your trading system.
TradingWhether you're a novice or an expert, these 10 rules should be the backbone of your trading career.
Personal FinanceSystems traders divide their time between trading, developing, backtesting, optimizing and forward testing, to create viable and high-probability trading systems.