Thursday, May 8, 2014

Learn about Pattern Projections with ChartDNA

ChartDNA is a market monitoring tool available on Bloomberg Terminals. It can be accessed by typing APPS DNA <Go> on the terminal.  A 30-day commitment-free Preview is available. 

How to start:

First decide what asset or market to study. In this example the outlook for the S&P ETF, SPY , is investigated. 

Step 1 is to load a chart for the SPY .

Once  loaded, select the pattern  to analyse.  The simplest way to do this is to  load the last 4 bars .  However, ChartDNA also supports pattern selection directly from the chart, or custom patterns can be built using the pattern builder. 

The selected sequence will appear in the image box in the Analysis tab.  The Pattern sequence will also show underneath the asset name.  In this case example, the pattern sequence is   JBFC :

Patterns are analysed against a group of assets. The bigger the group of assets, the more robust the analysis.  A default set of 30 assets, based off the Dow Jones Industrial Average, is offered to new users.

If ChartDNA is used as a screening tool for patterns, then the watchlist should include all assets of interest.  

In this example, the 481 components of the S&P are used as the core set for analysis.  

Step 2: Creating an asset list? 

Click Configure in the Analysis tab.

This opens a pop-up, Watchlist Manager dialog.  From there, select one of the major Indices (the easiest way to start). In this case the S&P 500 is selected.

And from there, select all components to save into a new Watchlist, here called "SPY"

Step 3: How is Analysis done?

With the Watchlist created it has to be selected for analysis.  This is done from the dropdown menu.

Next, the risk:reward, match strength and trade direction are determined.  

Match strength is used to filter the number of matches.  Short patterns can use stronger match strengths, while longer patterns (6+) may get more matches at a weaker match.  For starters, stick with the 'Neutral' match strength.

For risk:reward, 3:1 ratio is a good starting point.  In this example, using a 5% stop and a 15% limit.  

Finally decide if Long or Short trades are preferred.  In this example, 'Long' trades are selected.  

Once these steps are complete start the analysis process, hit [Recalc].

What happens during the Analysis process?

The selected pattern, #JBFC, is compared against 1 year of daily data for each of the assets in the Watchlist. 

Historic matches will appear as red and green highlight regions in the chart.  Along the bottom of the chart is a sequence of letters from which patterns are matched; each individual letter corresponds to the characteristics of price action for that day.

It's this price letter sequence on which the pattern is compared against. The Match Strength determines how strongly the pattern is matched against the price sequence; the weaker this setting, the more gaps and replacement letters tolerated in a returned match.

In ChartDNA, returned matches are highlighted on the chart: green if profitable, red if losing.  The table on the right lists all matches within the Watchlist, including the % Winners, Total Return, and Average Return, based on the % Target and % Stop selected.

All matches for a given asset can be viewed by clicking the eye icon in the results Table Action column.

An overall summary of the pattern is also given, showing total number of matches: winners and losers, and an average return per trade, under Loss vs Profit .  The closed trades only looks at patterns which have completed. 

How does analysis on Watchlist stocks help forecast for the asset I am interested in?

Ideally, one would only be interested in how the pattern performed on the selected asset, i.e the SPY. Unfortunately, there will be insufficient historic matches in the chosen asset to draw a conclusion from. By using a watchlist of assets a much broader data set is available to search for matches and test historic pattern performance.  

How does the forecast work?

In addition to trade metrics, a histogram is provided, showing how the pattern performed in the near term.  This is a simple probability measure, measuring from the open price from the bar immediately after pattern completion - the earliest time you could act on a pattern - to the close price that day, and for the next four days afterwards (five in all).

If the green bar is higher than the red bar, it means there is a greater probability for a higher close that day (relative to the Day 1 open price).  If the red bar is higher than the green bar, it means there is a greater probability for a lower close that day (relative to the open price on Day 1).

The forecast histogram is a good way to determine if you should trade a pattern Long or Short, depending if green or red bars are higher.

To get the actual value for a bar, roll your mouse over the the histogram

Building a forecast map .

By tailoring the length of pattern, to the match strength; longer pattern sequences using weaker match strengths, a robust forecast can be made.  The greater the number of matches, the more robust the forecast.

A probability table can map the ChartDNA 5-day forecast for the initial 4 character sequence, then repeat the process for the 5, 6, 7 and 8 character sequence (matches become fewer beyond, and the results less useful).  

For example, the table below shows the forecast for May 8th 2014 at Neutral match strength.  The forecast is starts with the 4 character sequence, JBFC, to the 8 character sequence, YEBZJBFC.  The end character 'C' is the May 7th candle - the most recent daily candle from the day before, 'F' the May 6th candle etc.    

Trade statistics are also included for SPY components (i.e. individual stocks) using a 15% limit and 5% stop if traded on the open price the day after the pattern completes: 

In the above forecast, as the pattern is lengthened (effectively tightening the match criteria), the bullish outlook improves.  With the SPY just shy of a new all-time high, the chance of it making a break look very good.

No comments:

Post a Comment