STOCK'S DIRECTION BEFORE AND AFTER REPORTING DAY


27 Oct
27Oct

In this example I was curious to know if I can effectively guess the direction of an arbitrary stock just before the reporting day by using the 5_LAYER forecast model only.

For that task I went to Yahoo Market Company Earnings Calendar page on Friday before the reporting.

I picked 12 stocks randomly , I then downloaded the price history from Yahoo, making sure there is enough price history for the 5_LAYER algorithm.

I then ran the 5_LAYER model on each of these price instruments, placing the learning border cursor (LBC) on July 2019, allowing almost 4 months of validation. The price on the validation period between July and Oct is not included in the calculation (i.e. it is out-of-sample).

The logic here is that if the 5_LAYER projection lines are correlating well with the actual price on the validation period of 4 months, the chances are that they will also be well correlated to the price in the period of a few months after.

For each price instrument I allowed an additional month of forecast to be visible, until Dec 2019 so you can track the performance later on. 

For each price instrument I added the summary of the cycles for the 5_LAYER MODEL, in this case using the 4 layers out of 5 just to allow a better visibility.

No fundamental analysis was done. No technical analysis was done.

For each price instrument the cycles represent the following indications:

  • ANNUAL CYCLE (Dark Blue)  - indicates the direction for the next few months
  • QUARTERLY CYCLE (Light Blue) - indicates the direction for the next few weeks
  • MONTHLY CYCLE  (Green) - indicates the direction for the next few weeks
  • BI-WEEKLY CYCLE (Orange/Red) - indicates the direction for the next few days

The presented combination of cycles on each of the following slides, a slide for each stock,  can show a very good "guess" for the direction of the stock just after the reporting and also for a longer period.




Comments
* The email will not be published on the website.