Stock Predictors ... or Cousins

Once upon a time we talked about Sister Stocks where we got us two stocks that tend to move together.
When there's a dramatic deviation from their historical relationship, then we consider buying or maybe selling and ...

>Assuming they'll soon return to their historical association, right?
You got it. Often, you're comparing a stock to some benchmark.

Anyway, here we want to talk about ... uh, cousins. Stocks that aren't that close, but separated by a few days or weeks.
We'd like to have two stocks, Stock#1 and Stock#2, where Stock#1 tends to make its moves in advance of Stock#2.

Like Figure 1

>You mean #1 predicts the movement of #2?
Hey! You're gettin' schmarter! That's exactly what we're looking for!
Well, I mean .. not exactly the day-to-day movements.
We'd just like to have the maxima and minima for Stock #1 precede the extrema of Stock #2, as in the Fig. 1 example. That way, we'd know in advance that, pretty soon, we should buy or sell Stock #2 and ...

>You're kidding, right? Stock #1 tells you when to buy or sell #2?
That's our hope and ...

>You're dreamin'. Figure 1 is an invention. If I knew how to predict like that I wouldn't be listening to you. I'd be ...
Sitting in your mansion in the Caribbean sipping Piņa Coladas ... yeah, I know, but I'm not asking that this advance warning be 100% accurate. Just consistent enough to make a few bucks.

>I assume you're investing in #2 and looking for #1 to tell you that a buy or sell is coming up ... soon.
Yeah, exactly. In most pairs of stocks, there'll be no such predictive qualities. Indeed, it's easy to find pairs of stocks that behave almost exactly the same, day-by-day.

Figure 1
Clearly, simultaneous movements up and down ain't useful. For example, the DOW and the S&P500:

Then there are pairs that have little or no relationship:

>And then there's your invention ... in Figure 1, eh?
Yeah, Figure 1.

>It is an invention, right?
Uh ... actually it's the past few months of GM and DRYS.

>And you've actually made a bundle on the predictive characteristics of GM, vis-a-vis DRYS??
If I'd made a bundle I wouldn't be talking to you. I'd be ...

>Sitting in your mansion in the Caribbean sipping Piņa Coladas ... yeah, I know. So where's the spreadsheet?
Patience ...

Searching for Cousins

Okay, here's a spreadsheet:


  • Pick a stock that we're interested in (Example: CBQ.TO in cell Q6)
  • Pick six prospective "cousins" (In cells M1 to M6)
  • Move a slider to pick a range of days over which we identify Maximum and Minimum (Example: we're look at Max & Min over 20 days)

>Wait! What does that mean ... looking at Max & Min over 20 days?
We run through all the days and identify those days that have the Maximum price over a 40-day range, from -20 days to +20 days.
For example, on July 23, 2007, GE closed at $40.82 which was the largest closing price from (July 23 - 20 days) to (July 23 + 20 days).
Then we stick the date on the charts.
Finally, we:

  • Move another slider to pick one of the six stocks (Example: Stock #4 = GE)
Each time we move that stock-picker slider, a year's worth of prices are downloaded for the "stock of interest" and the prospective cousin ... and the charts updated.

>That's it?
That's it. Click on the picture to download the spreadsheet ... and play. It's great fun!