Can I Scan seasonal charts?

dswanson2112

I wanted to know how to use seasonal date in my scans? for instance stocks that have a month that they gain year 90% or greater over a x amount of years. maybe 10 or 20 years.

markd

I am not aware of anything canned or built-in that will do that. At least in theory you could write a scan that would do it, but it would be a monster to write.

lmkwin

I've played around with this thought from time to time. Came up with something like this

// add your universe to look at here


and [[9 months ago monthly PctChange(1,close) > 0.1]
and [21 months ago monthly PctChange(1,close) > 0.1]
and [33 month's ago monthly PctChange(1,close) > 0.1]
and [45 month's ago monthly PctChange(1,close) > 0.1]
and [57 month's ago monthly PctChange(1,close) > 0.1]
and [69 month's ago monthly PctChange(1,close) > 0.1]

You would need to select a beginning of a month date to run it, recommend 2 or more months back. You need to add your Universe of securities to look at. You can boost the 0.1 to something higher. And add more years by adding more lines and adding 12 to the month's ago.

I've since expanded on the concept to look for "seasonal periods" but I don't use it as a primary scan. It's something to keep in mind perhaps, but definitely not a primary indication. In other words, I wouldn't bank on it as the performance shifting one month earlier or later in any of the periods throws the whole thing off.

markd

@lmkwin I like your use of PctChange( ). I keep forgetting that's available, even though its in several of my own scans.

A couple of comments:

You need the "monthly" modifier in front of close - so monthly PctChange(1, monthly close). You will get results without the modifier, but they may be unexpected.

Also, I think I would run it from a month's end date, e.g. Feb 28, then look back 12, 24, 36 etc. months. So,

[group is SP500]

and [monthly PctChange(1,monthly close) > 0.1]
and [12 months ago monthly PctChange(1,monthly close) > 0.1]
and [24 months ago monthly PctChange(1,monthly close) > 0.1]
and [36 month's ago monthly PctChange(1,monthly close) > 0.1]
and [48 month's ago monthly PctChange(1,monthly close) > 0.1]
and [60 month's ago monthly PctChange(1,monthly close) > 0.1]
and [72 month's ago monthly PctChange(1,monthly close) > 0.1]

Finally, this scan gets stocks that are up EVERY year for a particular month, which is very useful. I think @dswanson2112 had a lower bar - up maybe 90% of the time. That would require a mass of "or" statements which would be very time-consuming.

But playing with your approach was very enlightening for me - I'm going to look into it some more. I agree it's a secondary indicator, but if you get a hit with another scan, then run this scan against that stock for the current month (i.e., as of the prior year for x years back), it would be a nice piece of confirming evidence. Or, get a list of stocks for the month coming up, and concentrate your set up scans on those stocks.

lmkwin

@markd, Thanks for the formatting help.

If I would have kept notes there must have been a purposeful reason I did the month back setups in the manner I did. Probably based on advanced effort in search of seasonal periods. Like scan for securities that show positive returns for the "next" 3-6 months seasonally. Also probably had a "backtest" reason like ranking the results by return for the subsequent month(s) or something.

I know I started out with the 12,24,36 but found it lacking for some reason. Anyway, yes, it was quite a fun "test" of using the scan filters.

I really like the PctRelative function and try to work that into scans, including the rank by. And don't forget about the new Count Up and Count Down, as well as the Streak Up and Streak Down. Those can probably be of assistance in the scans looking for a movement/ trend in prices and/or indicators.