Studying Your Data Across All Bar Sizes
Have you ever wonder why you like to use 5-minute bars to trade something, and then, all of a sudden, switch into another time frame like 15-minute bar to monitor another instrument. Why we somehow have different level of comfortability with different time frames? Here is an interesting way to study an instrument to find out what time frame may be best for you.
Producing the Table
The table above can be produced pretty easily. Just load up your chart with the highest possible N-minute bar. Then throw in a short formula (in NeoTicker’s case it is fml
cum (h - l)) to collect the total accumulated range. Repeat the process for the next highest possible N-minute bar, until you reach 1-min.
The Ratio column is the ratio between the accumulated range for a particular row, comparing that to the first one, which is the accumulated range for a daily bar. Since our example is done on S&P Emini, there are 390 minutes from 9:30 am to 4:00 pm Eastern Time for regular stock market trading, thus I chose 390-min bar as the daily bar. Some people may want to include the extra 15 minutes after stock market close, I will leave that as an exercise to the readers.
Visualize Our Data
What is so important with this table of numbers? That can only be explained through a chart. Here we go.
This is a chart comparing the number of bars per day against the relative size of the accumulated range. In layman’s term, I wanted to see what effects we have when we look into higher resolution on the data we are given.
What we can see from the chart is that even though we increase the number of bars significantly, the effect of increasing accumulated range diminished. i.e. we do not get a lot more information from having more bars a day.
That tells us something important here. If you are looking at pure volatility in any particular time frame, then, it may not worth your while to look into higher resolution time frame afterall. Using our example, if you are tracking 5-min bar data for your trading setup, then 3-min bar will not provide you much more information.
So, are higher resolution data useless then?
That depends. Using our example again, if you are looking for moves that can easily produce 5 points or more, then looking at 30-min bars is the perfect time frame. However, once your position is becoming profitable, then you may want to drop down to 15-min, or even 5-min time frame to protect your profitable position.
Thus, constructing the table we have here can serve several different purposes.
First, it provides you with summarized information of the expected volatility in the market you wanted to trade.
Second, you can use it to pinpoint a time frame for you to seek for trading opportunities that fit your profitability goal.
Third, money management is almost always a job for a higher resolution time frame. You can choose one that give you minimal amount of work, while maximize the profit potentials.
Here is a simple example.
If you are seeking for 3 to 1 risk reward ratio opportunities, where the reward is in the range of 15 points, then it is much more likely you can get setups of that kind in time frames with 60 or more minutes per bars. To manage your risk, you need to monitor a higher resolution time frame like 15-min, in order to be effectively monitoring your risk. On the other hand, dropping down to 1-min bar may not be helpful if you do not want to deal with market noise within a 2 point range.
More Hidden Information from the Chart
The chart above demonstrates what is called a correlation in log scale.
Here is the same set of data plotted in log scale for illustration purpose.
We already know that dropping from 5-min bar down to 1-min bar does not give us much more information. However, it just means that using time based sampling of the data has its limitation. What happened is that some information in the raw data is inheritedly lost during the process of summarization that gives you those bars.
Thus if you are looking for finer control of your trading, you need to use a completely different sampling method in order to obtain information outside of the time based resolutions. For example, using N-tick bars, or fixed range bars, may help you see the data from a completely different point of view as the data sampling process is fundamentally different.