Throughout MarketHistory we use a special type of chart to perform Event Analysis - we take any event, and measure how a market has responded in the past following that event's occurrence. We developed what we call our Event Chart to provide a visual way to understand the direction, strength, time horizon, and quality of a bullish or bearish edge.
To make this possible, we take the close on each of the historical event dates, and measure the percentage move of the instrument from that level, to the close the next day, the day after that, etc. all the way out to some number of trading days in the future. Our usual number is 30 trading days (t+30.) Here is an example:
The chart plots each of these individual occurrences as a grey line in the chart, starting with the event date 't' at the left hand side of the chart. Time proceeds to the right, and the vertical scale of the chart represents the percentage return relative to the close on each event date at 0.0%. If the grey line proceeds above the 0% line it means it has rallied; a line below the 0% line indicates a decline.
We aren't as much interested in the individual occurrences as we are the overall tendency of all the occurrences together. The tool we use to represent the overall tendency is an average. For each trading day after the event date we average all the percent returns for the instrument as of that date. This average is plotted on the chart in a thick blue line.
In order to measure how well the average represents the underlying data we calculate the statistical standard deviation of the returns. This statistic tells how widely varied the returns are relative to the average, and we make this relationship apparent on the chart by plotting red Hi-Lo bands by adding the standard deviation to the average to get the upper red band, and subtracting the standard deviation from the average to get the lower one.
We take the ratio of the average return to the standard deviation of returns as our basic measure of trading edge goodness. This ratio shows up in our [Data] view in the EarningsEdge and SeasonalEdge tools as the ZStat. It is also very similar to the Sharpe Ratio. This ratio is displayed using the stacks of triangles which run along the bottom of the chart. If the ratio is bullish (positive) the stack of triangles is green; if bearish (negative) they will be drawn in red.
The first triangle appears if the ratio (bullish or bearish) is 0.50 or more. For every 0.25 increment in the ZStat we add another triangle to the stack for that date. The peak of the trading edge of any event is selected based on the point along the time horizon where the stack of triangles (or the ZStat) is largest. The peak of the edge, bullish or bearish is where the reward (the return) is strongest relative to the risk (the noise, the standard deviation).
So, the way to understand the charts is to consider that this indicates the peak holding period from the event date to whatever date in the future for this particular edge, based on all the individual historical occurrences taken together. Of course, as you can see from the individual grey lines, there are cases that will be higher or lower than the average on the peak event date. There are cases when getting out of the position on the peak edge date would have been the best thing to do, but also cases where letting the position ride after that date would have provided more return than average. You should also keep in mind that there are many cases where the returns are much lower than average - to the point where you may get stopped out of the position without ever getting to the peak edge date.
I like to see a consistent edge leading up to the peak edge date, and I like the peak to be at least three triangles - which corresponds to a ZStat of 1.0 or higher. This can be relaxed a bit if the number of occurrences is more than 25-30 because of the additional confidence provided by the larger sample.
The great advantage of these charts is to give the Event Analyst an immediate appreciation of how the instrument has performed in the past following a given event, no matter what that event is. The Event Chart provides:
the direction of the historical trading edge, bullish or bearish
the magnitude strength of the edge (average return)
the time horizon over which the edge typically unfolds
the quality of the edge (edge index triangles, and the width of the red Hi-Lo bands)
and the plots of the actual returns each individual occurrence lets the analyst see the anomalies as well - those outlier events which either vastly outperformed, or importantly for understanding the downside risk, the drawdowns a position may have had to endure to earn the historical return. This can be very helpful in figuring out where to place protective stops orders to reign in risk.
If you click the Play button [>] in the movie controls beneath the chart shown below MarketHistory Editor Gibbons Burke will explain the features of the Event Chart, pointing out the various features as he goes along.
You must have QuickTime installed to view this movie.
