Turn Down The Day Trading Noise
In the last couple of articles we discussed volatility filters. The first article surveyed a number of indicators typically used to gauge volatility, and that was followed by a more detailed look at the volatility squeeze indicator. In this article we’ll look at a filter that is the logical counterpoint to the squeeze. The volatility squeeze identifies periods where price action constricts and movement is limited. These tend to be difficult periods to trade as there is no follow through on trade setups. At the other extreme we have periods where price moves wildly and unpredictably. These can also be extremely difficult to trade as seemingly random price spikes can easily stop us out of trades. We refer to this explosion in volatility as market noise and it is usually seen at the market open and as a result of news events. Figure 1 shows one such period on a 377 tick chart of the Russell 2000 e-mini futures (TF). Figure 1 – Noisy Price Action Notice how the size of the price bars varies wildly, the relatively large size of the candlestick wicks, and the frequent changes in direction. Below the price bars we plot the range or size of each price bar and a horizontal line that shows a visual approximation of an average range. This will be a difficult chart to trade. Contrast the previous chart with Figure 2. This shows the same market, but notice that the price action is much smoother, bars are more uniform in size, and the range of the price bars for the most part falls comfortably below the horizontal line we established in Figure 1. This chart will be easier for us to trade. Figure 2 – Low Noise Building a Noise Filter It’s possible to build a simple noise filter using what we’ve seen above. One of the first things we need to do is smooth out the range data. We’ll use a moving average to do that, and we have several options, from a simple moving average, to an exponential, weighted, adaptive, etc. average. For this article we’ll use the simple. As it happens most charting platforms already have a handy indicator that we can use, and that’s the Average True Range (ATR) indicator. The true range is almost identical to the range of the price bar except that it also considers the close of the prior bar in relation to the current bar’s high and low. And the ATR, as its name implies, measures the average of the true range over a number of bars, 14 being a common default value. Figure 3 shows the ATR applied to our Russell chart. Figure 3 – Average True Range The one disadvantage of smoothing the true range value is that it introduces lag into the indicator, a common shortcoming of most indicators. You can reduce this lag by shortening the ATR period from 14 to something lower. A simple noise filter can now be built by comparing the value of the ATR to some predefined average noise level. From Figure 3 we can see that 1.0 might be a good value for the Russell, perhaps something a bit larger like 1.1 or 1.2. As long as ATR is below this value we trade our setups, if it’s above then we filter them out and skip them. The problem with this approach is that the average noise level will vary with the instrument and time frame that we are looking at. For example, the average noise level for crude oil futures on the 377 tick chart is about 0.15, and for the Russell on a daily chart it’s approximately 15 points. And it’s also a subjective level determined by eyeballing the chart. A better approach to developing a noise filter is to make it more dynamic. We can do this by comparing a fast ATR to a slow ATR. The slower ATR will give us a proxy for average noise level, and the faster ATR will tell us if price action is noisier than average or not. We could simply take the difference between the two, or better yet use the ratio. By taking the ratio of a fast ATR to a slow ATR we know that a ratio greater than 1.0 signifies noisy price action. Since there will always be minor fluctuations in this ratio, it’s prudent to set the noise threshold slightly above parity, and I’ve found 1.1 to be a good number. Our noise filter can thus be turned into a simple indicator: Noise Factor = ATR(14)/ATR(50) If Noise Factor >= 1.1 then filter setups. Figure 4 is an example of the noise filter on our Russell chart. Figure 4 – Noise Filter Notice how the noise filter keeps us out of the early volatility and chop but allows us to take setups throughout the subsequent move. We plotted it as a histogram instead of a line because it’s easier to detect breaks of the 1.1 threshold level this way. This simple filter can now be applied to any chart in the same way. We don’t need to customize the threshold level any more. Going Further This article described a very basic noise filter. I encourage you to take it further and experiment with it. One of the first things you can try is altering the fast and slow ATR periods. I’ve found 14 and 50 to be good all around values, but you can get a much more responsive indicator by shortening both. Values of 8 and 21 for example will significantly reduce the lag in the indicator. You can also vary the threshold level. Consider slightly higher thresholds on the faster filter, 1.2 may work better with the 8/21 ratio for example. As mentioned earlier you can also apply different types of moving averages to the true range. Exponential moving averages react faster than simple averages, and other types like smoothed, weighted, etc can help further reduce the indicator lag. Experiment; find something that works for you.
About the Author:
Will Feibel is a forex expert at NetPicks.com. You can read more of his articles as well as other free day trading articles, videos, webinars, and more at http://netpicks.com/trading-tips.
This article has been viewed 2 time(s).
Share this Article
