Earlier I’ve described the statistical Forex systems and introduced the problem of choosing the right timeframe to gather the statistics for such systems. Today I will try to describe the problem of choosing the right information that is collected for the statistical trading system.
Gathering the statistics over a chosen period of time for the given market instrument is the next step to create a successful statistical Forex strategy. But what data should be used for this statistics? Is it a good idea to record bare chart data? Should you gather any additional information? Here is my view on all possible statistics type that can be used in the process:
Pure market quotes. This includes high, low, close and open rates for bars and bid or ask rates for ticks (if you think that tick-based statistics is a good idea). This method of statistics gathering is the most obvious. You gather the market quotes then compare them with the current situation and decide whether to buy, sell or hold. But there is a problem with the changing of quotes range. For instance, 1 year ago EUR/USD was in 1.4000-1.5500 range, a month ago it was far above 1.5500 level, so the data gathered in another price range would be completely useless. Alternatively a normalization of some sort can be used to store such statistics — e.g. store not a quote like 1.5404 but its relation to the next bar’s open price — 1.5404/1.5423 = 0.998768073. This way you’ll have data that is informative in any price range, but still uses no indicators or other complex calculations.
Indicators. These are probably the best data to be recorded as the statistics. Even standard MetaTrader indicators allow recording a lot of information and then using it to compare with real-time current market situation. With a large part of the indicators the normalization similar to the one used with the raw market data will be necessary. It’s probably a good idea to use indicators that change in the certain range — like RSI, DeMarker, Stochastics, Larry William’s Percentage Range, Money Flow Index, etc. The length of the arrays of the indicator values recorded for each tick or bar is also an important parameter of the statistics gathering. Remember that the longer this length is the more uninformative this statistics becomes. Ideally, it’s better to use single value of each indicator that is unique for the current bar or tick.
Additional information. It can include the time of the day to capture the trends and patterns that are specific for some trading sessions only. Another parameter that fall into this category is the day of the week — trading usually differs depending on the busyness of the day (often with less price action on Fridays). The statistics can also note if the day is some major holiday, current daylight saving time mode for the major countries and the volumes of the trades (although in Forex they are not very informative).
Complex calculations. This can include not only calculations based on the market data and indicators, but incorporate the additional information such as time and the day of the week into the calculations. In this case the produced number-formatted statistics would be easy to compare to the real market data. Considering the current development level of the PC industry it wouldn’t be a hard task to incorporate even the most complex calculations in the MetaTrader expert advisor that utilizes a statistical Forex strategy. Additionally these calculations can be accompanied by the various pivot points and resistance/support levels to help with choosing the position’s parameters.
No comments:
Post a Comment