Engineering Professor (EngProf6)

Hybrid Engine for Modeling Stocks and Indices

Response to Several Messages on the AKAM Message Board (Yahoo):

I was asked several questions by a few readers (e.g. repelsteentje2001 et al..) on the AKAM board. The overall tone of the questions is paraphrased as a series of the following questions: “How do you compute if a move is right or wrong?” – “Do you take the difference between the closing prices on the dates you list?” – “And how does the threshold price fit in?”. One should note that the 3 questions are connected. A more detailed understanding of the logistics of the model will help us formulate the answers.

Let me start off by stating that the one model is used to analyze all stocks etc… It is a ‘one size fits all’ approach. You know what that means. The model is not going to fit any unit ‘that well’. With one it will be better and with another it will be worse. You need to have experience to gauge this.

You should know that I have a ‘count down’ and that I announce a threshold value before the fact (i.e. the reversal). The idea is that you are supposed to use the opportunity to position yourself, based on the forecast. My forecasts are to the ‘nearest day’, and that is a fairly long time. Sometimes, it is obvious that the stock will flip because the threshold is too far away. For example, on Monday April 17 SPY was to flip the next day. The threshold was the equivalent of 15 S&P500 points below the close on Monday. For the astute investor, it became clear Tuesday morning that there would not be a collapse of the market that day. The investor who has acquired some experience with my model would have made a move shortly after the open. Would I have been happier if SPY had reversed the day before, of course, but the model was certainly pointing at the reversal. Remember, the model tends to hold back. So anticipating a move that will happen is normally to one’s advantage. Waiting for the close is not part of the model. Normally, models issue signals ‘after the fact’. That’s the way it is when modeling time events.

Going back to Tuesday, April 18, the Dow had been flipped to the UP the day before. AKAM’s count down was approaching the reversal and the threshold for Tuesday (as generated Monday night) was within 20 cents of the Monday closing price. On Tuesday morning, someone who was serious about buying should have. Remember my increments are in days. People can think and react in terms of minutes. That’s the way it is. Sometimes I write statements to the effect that you should use tomorrow to position yourself. I give an example in the next paragraph.

If one goes back to the start of the down move for AKAM before it went to the UP on Tuesday the 18^th, one finds that the down started on Thursday April 6. However, if you go back to the archive and look at what I said the day before on Wednesday April 5 you will find the following:

Number of days remaining: 1 (a decrease of1)

AKAM was relatively flat today. The ‘count-down’ has come down to 1 day. AKAM will reverse to the DOWN tomorrow (Thursday). The threshold price for tomorrow (Thurs) is $35.07. Given that AKAM will not go above this price, AKAM should be sold whenever you feel it is appropriate unless, of course, AKAM has a dramatic surge in pre-market.

Thus, one should have sold AKAM shortly after the open that day. That would have been good because AKAM went on to drop something like 70 cents during the day. So if you do the math, you would find that by reacting a bit more quickly – that is selling at the beginning of the day when the down move started and buying back the position on Tuesday morning (around 11 am) when it started to become clear that AKAM would go above the threshold, one could have made a profit of close to a dollar. However, if one had transacted at the closing prices, the move would have lost a dollar or more. Remember Tuesday the 18^th was a stellar day for AKAM and waiting until the close would have created a sizeable loss.

With regards to counting whether the down move for AKAM was right or wrong, the answer is wrong. As I’ve stated, if one uses closing prices only then the math is simple. For purposes of analysis I use closing prices and thus only 75% of the moves are correct. However, for investment purposes, rarely would I actually transact on a closing price. I would use my experience and judgement and couple it with what the model is pointing to. Typically, my decisions incorporate all these factors.

If you get to know a particular stock, you can get more from the model. You will be able to ‘adjust the size’ to make it ‘fit better’. For instance, with AKAM you have to include a bit of anticipation because AKAM tends to move quickly. This is where the individual comes in and makes his/her own decisions. I’m sure you can appreciate that the closing prices are not necessarily the best prices. If someone has zero knowledge, then, I say use the closing prices. However, for the rest of you, my advice is simple. Acquire enough experience and use your judgement to make your decisions. Use the model to guide you; don’t use it as an absolute indicator.

Good Luck.