TRADING SYSTEMS
A Solution To Backtesting
Forward Testing
By Martha Stokes, CMT
A simple technique, it can reveal the true profit or loss
capabilities of a trading system.
Backtesting has been popular among traders
for many years. The concept is that you take a set number of months or
years and test a trading strategy over that period's data. You calculate
the average results and performance of that trading strategy as it performed
during that time. In theory this sounds like a fantastic tool for traders
— backtest and know if the strategy works. But that is not what backtesting
this way actually gives a trader.
WHAT EXACTLY IS IT?
Backtesting can be defined in two popular ways:
1. The process of optimizing a trading strategy using historical data
and then seeing whether it has predictive validity on current data.
2. The process of testing a trading strategy on prior time periods.
Instead of applying a strategy for the time period forward, which could
take years, a trader can do a simulation of his or her trading strategy
on relevant past data in order to gauge its effectiveness.
Most technical analysis strategies are tested with this approach. At
first glance, it makes sense. Go backward, pretend you're using your strategy
or system on a historical chart and see what would happen. Sounds
good, right? Unfortunately, there are problems — inherent and significant
— that cause backtesting to fail.
The problem:
Backtesting a trading strategy or system does not take into account
whether a market is in an uptrend, downtrend, or sideways trend. It does
not determine what market condition is prevailing during current trading.
Therefore, a trader has no way to determine and apply the correct market
condition to the backtested time period. The results of the backtested
period are averaged to give you a profit to loss ratio for that trading
strategy. An average of historical data over several years doesn't give
you the facts you need about a system's viability and true potential profitability.
Let's see why:
First of all, averaging statistics does not tell you much. An average
can't tell you that for 20 weeks, the strategy had continuous losses or
that if you had simply traded more aggressively during the first 10 weeks
then you would have made higher profits. An average simply gives you a
figure based on the total set of data divided by the number of data set
within the test range.
All trading systems are designed under specific market conditions within
each trend of the market. There are five primary types of trading systems
for stocks under which all trading systems fall. By averaging the results
into a ratio of profit versus loss, the results that backtesting gives
you are smoothed to reflect a long-term perspective of your profits or
losses using this system. You are essentially testing based on long-term
hold situations. This is fine if you are a long-term investor.
The fact is that most people who use trading systems are short-term
traders. If your system has a period where it gives you continual losing
trades for weeks or months, your capital base and confidence will erode
to the point where you will no longer be able to trade when the market
conditions finally shift and become conducive to that trading system.
Second, it is important to identify what market trend and market condition
suits that trading system rather than merely backtesting for several years.
By selecting the market trend suitable for that trading system, up, down
or sideways, and by knowing the condition of that trend, whether the strength
of the trend is weak, moderate, or strong, a trader eliminates large chunks
of data from past years that is not functional for backtesting that system.
...Continued in the January issue of Technical Analysis of STOCKS
& COMMODITIES
Excerpted from an article originally published in the January 2008
issue of Technical Analysis of STOCKS & COMMODITIES magazine. All rights reserved. © Copyright
2007, Technical Analysis, Inc.
Return to January 2008 Contents
