Playing the odds: Part 2

This week I'm reworking my analytical methods to make them more nimble in picking up changes in trend.

On Tuesday I posted the details of what I was doing and what I'm doing now. It's available here.

Today I continue with the shake-down cruise by dealing with a problem raised by the new, short-term method.

Change is good, but all change needs improving.

The problem arose from my pre-screening once a week of the odds on all 9,000-plus symbols traded on the major U.S. stock exchanges. I ran a rather long-running routine that picked out those that had better than even odds of a successful trade in one direction or the other.

That worked fine when I was scanning bull and bear signals dating back to January 2009. But the system broke down when I shrank the study period to a year.

The number of signals since early 2009 generally runs around 20 or 30 for each symbol. An additional completed signal doesn't change the odds much.

However, the number of signals in the past year is generally five or less. An addition to the mix can make a huge difference in the odds.

My Monday analysis ran just fine in selecting new signals from my weekly list. All of the symbols had greater than 50% success rates to the bull or bear sides.

Not so on Tuesday. Most of the symbols had odds lower than 50% in both directions.

Clearly, weekly pre-screening wasn't going to work. And the screening routine took too long to run for me to generate a new pre-selected list of high-probability symbols every day.

So I spent much of last night testing ways around the problem, and here how I've decided to handle it.

My list will be all of the stocks in the analytical universe of the stock rating house Zacks. Their list changes, but slowly enough so that a weekly update with be sufficient. I don't do anything with the Zacks ratings. I just capture the symbols.

Today the Zacks list has 2,319 symbols, and that was my analytical universe.

I could have created a list some other way, such as on the basis of average volume or market capitalization. But I want to cast a wider net, and the Zacks list runs the gamut from penny stocks to transnational giants, and from less than a hundred shares a day up to eight-figure volumes.

The Zacks list also includes some over-the-counter (OTCBB) stocks, a market that was missing from my pre-screening lists.

Also, a spot on the Zacks list ensure that the symbols are on someone's radar. Many of the 9,000 symbols traded are only on the radar of their CEO.

Monday's pre-screened analysis of the one-year data produced nothing I could trade.

The same routine run against the Zacks list produced three possibilities after I had weeded out the low probabilities and earnings exclusions. They are all bull signals: IBN, HDB and MEAS.

That leaves three symbols for further analysis. And that's the next task on my agenda.