Engineering Professor (EngProf6)
Hybrid Engine for Modeling Stocks and Indices
Why Are There Events That Go AGAINST the Model?
This is one of the important questions that many of you would like me to answer. First of all, let’s assume that the model ‘works well’. Under ‘normal’ circumstances, it is able to capture viable Entry and Exit points reasonably well. Let’s also assume that it is capable of formulating time cycles for the stock or index we are looking at. That having been said, we also know that there are events that it does not anticipate. An example could be the release of earnings. With that in mind, let’s consider the events surrounding the release of earnings by Google after the close on Thursday, April 20.
The model had GOOG in a DOWN cycle when the earnings were released. After the close on Thursday, April 20, GOOG announced results that were substantially better than the average consensus. GOOG went on to increase by as much as $35 after hours. The model only uses market data from the regular hours, thus on Thursday evening the model was not aware of what was going on. But I was. Even with that rise, when I queried the model with such a move the model indicated that the down cycle was looking fragile but intact.
Naturally, I had a number of rabble-rousers (i.e. the antagonistic constituency) who were quick, as they are with any event, to seize the opportunity to say that the model is worthless garbage. It is useless; it is a contrary indicator.
Let’s review some of the facts in the GOOG case. Prior to the release of earnings, the consensus of a number of analysts was available to the public. Thus, the collective body was expecting something and the stock was trading as if this something was going to be the subject of the release. GOOG stock was in a DOWN cycle based on this expectation, which one would suspect was reflected in the stock’s price. When the earnings were announced, they were substantially better. They beat most of the predictions of the analysts and they, in effect, beat the model. What it does show is that the company concealed the results fairly well because the model (which is totally objective and mathematical) did not pick up on them. Naturally, this implies that the traders did not pick up on them because the model only reflects what the consensus is doing. This is important to note. The model synthesizes the views of all concerned at points in time and generates an evaluation of what it thinks is the path of least resistance based on the past performance that it is aware of. When the model says UP or when it says DOWN, it is, in effect, making a conclusion based on the available data. Naturally, if there is data that is yet to be disclosed which goes against current thinking, then the disclosure of such data will cause the model and traders to scramble to re-evaluate their positions.
Now that the earnings have been released (today is Friday April 21) and the stock has risen, what does the model say? I can only speculate at this point. The model is looking at GOOG to be a candidate for a double bottom (as QQQQ is currently). Thus, the down direction will be maintained (check the latest results to make sure). The model will flip GOOG to the up mode in due time when it is satisfied that it’s the right thing to do.