However, this was not the case here. Given that, the table is a repeat of yesterday's projections, with the confirmed end-of-day close bar. The table is an aggregated output for each of five separate pattern analyses as determined by ChartDNA using Bloomberg Data. What follows is an explanation of the table output.

**What is ChartDNA?**

**With ChartDNA, you can look at how well given patterns - candlestick sequences - performed historically in your test data. In the above example, we are looking at how the most recent end-of-day pattern in the SPY performed in the past.**

**What's the "test data"?**

Ideally, we would only look to prior matches in the SPY to get an idea as to what might be expected to happen in the future, but this would return very few, if any matches (other than the current match). To get around this, we create a test data set ("Watchlist"), which can assets from any price series you choose. In this case, the price data is taken for each component stock of the SPY ETF. This creates a very robust data set on which to run the analysis.

**What is the list of Patterns on the left?**

ChartDNA can look at patterns from 4 characters in length to 10 characters in length. While the above example looks at the SPY, this analysis can be done for any price series. Simply select the asset you wish to analysis, then load the last 4 (to 10) bars (remember to press "Go"). The analysis will be done using the selected watchlist of default stocks as the base, but this watchlist can be changed.

*Remember, the components of watchlist generate the results output, but it's the pattern in the selected asset which is been measured*.

In the SPY example, the first pattern is the shortest 4 character sequence, and each row looks to add another character to the pattern, before the analysis is run again (by selecting "Go").

Each additional character increases the length of the pattern, meaning fewer matches will be returned. In the "Matches" column above, the number of returns from our 481 test equities drops dramatically as patterns extend from 4 to 6 characters in length.

*Note: There is also a Match Strength variable, which can be adjusted. This example uses the default 'Neutral' setting, but if your interest is in performance of longer patterns, then the Match Strength can be weakened so more matches are returned.*

**What is the probability output?**

**In ChartDNA, there is a histogram output showing the probability of a higher or lower close price (per bar - in this case, a daily bar) based**

**. To see the actual % value roll over the histogram with your mouse, although the relative height of the two bars will give you a visual clue as to whether the outlook is more bullish or bearish.**

*from the open price after the pattern completes*Because this is an end-of-day analysis, and the market is closed,

__the open price used is equivalent to the next day's open__. This allows the User plenty of time to run analysis and be prepared for the next trading session with either favored assets (or favorite patterns).

*Note: If using intraday time intervals, the open price is determined from the start of the next bar.*

*The aforementioned Excel table lists the rollover information for each probability of a higher close from Day 1 (same day) through to Day 5, based from the open price of Day 1. Each row of the table gives the probability for each extended pattern sequence, as a new day is added to the prior pattern.*

**What is the "Average" probability?**

This the probability of each pattern match, weighted by the number of matches. In the above instance, as the pattern lengthens, it becomes more bearish in outlook. The disadvantage is that fewer matches are returned. In the case of the 7-character SPY sequence, there was only 16 matches from 481 equities, dropping to just 4 matches for the 8-character sequence. As the results are quite stark, it's often a good idea to re-run the 7-character and 8-character sequence using a weaker match to see if this is a more probable outcome. This can be done from the Dropdown menu, and then doing a [Recalc].

*Note: This will done in a different article*.

**What is the "% Win" and "% Return/Trade"?**

**In addition to probability outcomes going forward, ChartDNA will also look at the trade opportunity for the selected pattern.**

The Trade potential is determined using a

**.**

*% Target and % Stop, calculated from the open price of the next bar after pattern completion*The analysis looks at past pattern matches and determines what percentage are trading above the open/entry price, and what is the percentage return per trade. The "% Win" and "% Return/Trade" can include open positions; i.e. "All" positions:

Or just data from closed positions, by simply switching the tab:

The aforementioned table was updated just to show Closed historic matches. This gives a very different outlook than the one where Open positions are included too.

*Note: The Trade analysis was done at a Neutral match strength.*

**Okay, so put it all together?**

The shorter pattern sequences are not suggesting any sharp swing higher. Today and tomorrow are more likely to have a weaker close than next week, which is slightly positive.

The longer pattern sequences are more bearish. Today ("Day 1") is more likely to finish with a closing price lower than today's open (which fits with the modest weakness from the shorter pattern), and if this continues into Friday ("Day 3") then it's likely to nullify any modest bullishness into early next week ("Day 4" and Day "5") as determined by the shorter pattern projections.

Independent of time, there is a 70% chance of a 5% loss from today's open in the SPY compared to a 30% chance for a 15% gain.

**I want more?**

You can test your own equities and indices outlook by Previewing for free ChartDNA on the Bloomberg App store. On the terminal, type APPS DNA<Go> to get started.

## No comments:

## Post a Comment