New Members: Be sure to confirm your email address by clicking on the link that was sent to your email inbox. You will not be able to post messages until you click that link.
Bollinger Bands - W bottoms/ M tops
Any recommendation for scan W bottoms/ M tops using bollinger bands?
0
Comments
The problem with any type of pattern scan is that the patterns are highly variable. The human eye can recognize them, but a computer requires very specific, unvarying requirements, e.g. the length between notable pattern features, heights of legs, and so on, while on the chart these values are far from constant. One instance of a pattern might roll out over three or four weeks, another eight or ten. In one case the height of the pattern might be 10% of a base value, in another case 25%. It's not that it can't be done, it's just very time consuming, and then you will still miss some instances you would recognize by eye.
@selva23 do you have any examples of charts that show your W bottoms and M tops on bollinger bands?
I will mention that Point and Figure patterns are often a great place to look for 'structure' as they all have pretty rigid definitions. The "Spreads" and multiple top and bottoms have reached a certain level a few times and then have broken out. Bear Trap and Bull Trap have headfake breakout reversal characteristics. All of the patterns have definitions. When you look at them on a time chart you may find what you are looking for, but you still have to look. I'm partial to Low Poles myself, but you still have to look at the charts as not all Poles or patterns are created equal.
Point and Figure patterns can be a great place to start.
Predefined Scans Page PnF section
ChartSchool partial listing of patterns. They ignore Poles here. You shouldn't, in my opinion. Very important pattern.
Here's an example I am looking for
https://schrts.co/AzGJIsaS
For bottoms, I am looking previous low ( on closing basis ) closed below the lower band, whereas the current close was above the lower band but the low was below the band.
So you'd need to break that up into a few requirements. Last period is this and current period is that. Your example is on weekly charts so try this.
and [[1 weeks ago weekly close < 1 weeks ago weekly Lower BB(10,2)]
and [weekly close > weekly lower BB(10,2)]
and [weekly low < weekly lower bb(10,2)]]
The 1st line looks at the prior period to find securities that the weekly close was less than the prior periods weekly lower BB.
The next line looks at the current period close to be greater that the current period lower BB.
The next line looks at the current period low to be less than the current period lower BB.
This will find the left side of your Blue Box example. The right would need further definition.
As this is looking at the left side and not the right side, you can run this using prior periods in the scan engine by changing the date at the top of the scan workbench to prior period Fridays (to capture the full week for weekly scans.
As you can see from below, if you were to run this today, the scan would be looking at a full week for the prior period and 4 days in the current period for the week.
The calendar feature is a great way to look at your scan against prior periods to see what it was returning at that time.
You can also 'nest' requirements using the OR instead of the AND. So you can add additional segments of the same code but pushing it back a period. So you would duplicate the line 1 from above and change 1 weeks ago to 2 weeks ago and modify the current period to be 1 weeks ago. Then, when you run the scan you would be capturing 2 weeks of results.
This can be replicated as far as you would like to capture. As you are only getting the left side of the blue box symbols, you can add them into a list to review for right side activity.
or [[2 weeks ago weekly close < 2 weeks ago weekly Lower BB(10,2)] and [1 weeks ago weekly close > 1 weeks ago weekly Lower BB(10,2)] and [1 weeks ago weekly low < 1 weeks ago weekly Lower BB(10,2)]]
If you have any questions on Point and Figure, I love to talk about Point and Figure. You can also use the Search box at the top of the forum to find prior discussions/ questions/ answers on various topics of interest to you.
If the stock is in a longer term up trend, price returning to the 13 week lower channel is often a decent long entry area. If the longer term trend is down, price returning to the 13 week upper channel offers a similar short opportunity.
Note if you fail to get a new high from a lower channel entry, it could be a sign of change in trend. Vice versa with new lows from an upper channel entry.
On this chart, the green channel is 13 weeks.
https://schrts.co/FfeRahDb
So for 13 week lower channel crossovers:
and [weekly Lower Price Chan(13) x weekly low]
"x" is the cross ABOVE indicator. There is no cross BELOW operator. So even though it appears the low crosses below the channel, for a scan you would write the channel crosses above the low, even though on the chart the channel is going straight at the time the low crosses it (usually).
You would also want the lower channel to be generally rising (going up in a stair step fashion) at the time of the crossover, so test whether it is the max value over some number of weeks. A channel can be flat for a while, giving an equal value over the look back period, which might not register as a max value. So you have to experiment. Or you could just do a value over 13 (the length of the channel - it will change value when the look back period (the parameter) expires (almost always). This method might miss some early up trend crossovers, though.
and [weekly Lower Price Chan(13) = weekly max(14, weekly Lower Price Chan(13)] // rising price channel
and [weekly Lower Price Chan(13) x weekly low] // low crosses lower channel
For down trend upper price channel crossovers, it's a little different, but same principle:
and [weekly Upper Price Chan(13) = weekly min(14, weekly Upper Price Chan(13)] // falling price channel
and [weekly high x Upper Price Chan(13)] // high crosses upper channel
This could also miss early down trend crossovers.
Note that the cross over is a set up, NOT necessarily an entry (although it can be). Often price will run beyond the channel some number of bars but without making much actual price progress. Other times it will turn on a dime.
Finally, any scan that looks back over time is going to require trade-offs - too broad and you get a lot of hits that fit your conditions but are not what you want; too tight and you miss hits that are close enough that you would want to know about them. Also, markets change over time. A scan with parameters that work well one year may need some tweaking the next. So don't be afraid to back test once in a while with different parameters.
For those that wonder where the Predefined Scans are, the best way to see them is through the Advanced Scan Workbench.
Accross the top of the workbench screen there are links listed. Predefined Scans link takes you to that page.