A few weeks ago, we looked at how a volume surge a breakout trading system affects its expectancy with various sets of exits. That study can be found here.
This week, I wanted to take a look at how price action associated with a volume surge changes things. There are a number of ways to look at price movement- as a percentage change, relative to average true range, relative to standard deviation, etc. This week we will look at percentage changes and next week, we will take a look at average true range.
Unlike the last few tests, primarily due to the amount of data involved, we will only look at one exit, a fixed 8% protective stop and a fixed 20% profit target. Neither of these exits are trailing and we allow either exit to be hit before exiting. While we concluded in studies, here and here, that there are other exits with much higher expectancy, the holding times increase substantially. The 25% trailing stop, for example, essentially becomes a more risk-managed alternative to buy-and-hold or a longer-term trading/investing strategy. We can assume that the longer a stock is held, the less entry criteria matter and the more important stock selection and long-term growth become. Therefore we will use our shorter term exit stops for these tests as they place more weight on entries for an increase in expectancy.
As usual, we will run the test on three separate portfolios to increase our confidence that the results are systemic and all tests will be run from 01/01/1998-12/31/2007.
Our entry criteria will a liquidity condition requiring 10-day average daily volume to be greater than 100,000 shares. In addition, we will look at both volume as a percentage surge over 10-day average daily volume using a >= condition, and a percentage price breakout over the previous day’s close in a >= condition. Trades will be entered at open with a market order the day following the entry signal.
I am getting fancy this week with some excel surface charts which allow us to look at two variables tested together simultaneously. Previously, we have tested only one variable at a time and the 2D charts were adequate to represent the results, when testing two-variables, the 3D nature of the surface charts makes it much easier to visualize the results. We will begin by looking at the EDR results for the three portfolios.
EDR=Expectancy per dollar risked. This is my primary evaluation ratio and has been covered extensively in previous posts. It is calculated by Average Trade/Average Loss. The results are in dollars with .28 representing 28 cents per dollar risk.The Price% axis of the charts represents the percentage change in price, while the Volume% axis of the chart represents the volume surge. This is consistent in all tests. The z-axis represents our various evaluation ratios.
Let’s begin by looking at the EDR ratio. The first thing that is immediately evident is that our IBD portfolio is substantially different than our other two portfolios. We will discuss this shortly but let’s first discuss the similarities. In all three portfolios, unlike price action, stronger volume surges generally increase expectancy, irregardless of price action. There is a drop along the 400% volume axis in the Nasdaq and IBD portfolios but the general trend is clearly more favorable as volume increases, particularly up to about a 300% volume increase. This is consistent with our previous volume study.
The price action however is quite a different story. Compared to volume, price action when evaluated as a percentage increase over the previous day has a much smaller affect on expectancy. Remember that the filter operators for these test was a >= operator so by selecting only 2% price increases (for example), we are still taking large price spikes, we just aren’t limiting our selection to those. Furthermore, the IBD portfolio experiences a severe expectancy decline with higher price thresholds. There are several possible ways of explaining this but there are two key points I want to make.
First, if we assume that there is a fair amount of random noise in day-to-day price movement, then a percentage change not only includes a certain unquantified amount of noise, but also derives the percentage change against the previous day which also contains a certain, unquantified amount of noise. It does not acknowledge typical volatility for a given security or how either day’s price movement compares against the average. In some instances a 3% move may be substantial, in other situations it is not. Some of these issues are addressed when using average true range which we will test next week.
The second point relates to selectivity. In reviewing these results, I noticed that the 0%/0% figure represents an interesting condition (indicated by the arrow). This is obviously the least selective but if we think about it for a moment, it is basically taking every trade that has a positive close on average volume, remarkably similar in fact to buy-and-hold but traded through a method that allows us to compare results. If we use this 0/0 point as our baseline, it allows us to evaluate whether or not increased selectivity with a given criteria improves things. Through these results, we can conclude that increased volume selectivity is effective for filtering trades, while increased price movement alone (as a percentage) is not, in fact it can hurt.
One other thing that occurred to me is an idea for introducing an adaptive component into a system that allows it to modify selectivity based on the native performance of the underlying basket of securities being traded. By looking at the 0/0 points for the three baskets, we can see that expectancy on the IBD is the highest natively, and that portfolio suffers the most from over-selectivity. I have also seen the same thing happen with using relative strength to filter NASDAQ stocks during the dotcom boom, nearly anything you did to “slow-down” resulted in reduced expectancy. This is only an idea but putting it here will make sure I remember it. This blog is increasingly going to serve as a development journal of sorts – hope everyone is ok with that.
Below, I will post the EER and DDR results for the tests. I will not comment on them but am making them available for those motivated to dig a little deeper. If anyone has any questions, feel free to ask.
EER=Efficient Expectancy Ratio. This is our EDR/Avg. Days in Winning trades. This number is meant to reveal the most efficient use if capital as it calculates expectancy per dollar risk per day in trade. The results are in dollars with .01 representing 1 cent per dollar risk per day in trade
DDR= Drawdown ratio. This number is calculated by Net Profit/Maximum Drawdown. A higher number is better.