Building a Better Universe

It is a fact of trading that what you see limits what you get.

Analyze optionable stocks with volumes of 5 million or more shares a day, and you'll have a universe of about 50 symbols, and trading signals with be few and far between.

Look at all stocks traded on the major exchanges, and you'll have a universe of 9,199 symbols, with many more trading signals.

My method in recent months has been to select all stocks tracked by the rating company Zacks that are priced at $15 and up with volume of 1,000 shares or more per day. That produces a universe of about 2,400 symbols.

A universe is just a starting point, of course. Once the universe is in hand (an interesting image), then the screening begins, with the goal of having a manageable list of stocks and exchange-traded funds from which to pick a trade or two.

My initial screening method is simple: I want stocks and exchange-traded funds that, historically, have produced profitable trades in the direction of the most recent breakout. Other ways of putting it is that they have better than 50:50 odds of success, or a success rate of better than 50%.

It's a brutal barrier. When I analyzed the Zacks universe Thursday night, only eight stocks among the 57 that produced a trading signal had better than even odds in the direction of breakout. (See my posting from last night here.)

Further tests for earnings within 30 days and a sufficiently high yield for winning trades after adjustment for the success rate reduced the survivor list to two symbols out of 2,399.

That's not many, and as we dive into the 1st quarter earnings announcement season, my rule excluding new positions on stocks that are within 30 days of earnings means fewer opportunities to trade.

So, when the old universe goes bad, the obvious answer is: Build a better universe.

My goals are to increase the number of symbols in my analytical universe, but to limit them as far as possible to those that have a better chance of surviving the initial barriers.

I began with the 9,199 symbols traded on the major exchanges. From those, I selected the 2,220 symbols that had a better than 50% success rate in the case of bull or bear signals (or both).

This universe has no limit on price or volume.

I then did my usual screen for new trading signals, and found 18 bull signals and 12 bear signals. Of course, my pre-screening method is no guarantee that a symbol's new signal will meet my test for better than even odds in the direction of the signal.

GOOG, for example, gave a bear signal, but it only has a 35% success rate in that direction. It has a place in my universe because of its 57% success rate to the bull side. Nineteen stocks failed the odds test, reducing my universe to 11 symbols.

Five symbols had sufficiently good odds, but the returns were too low to produce a high enough score when adjusted for the success rate. I require an adjusted yield score of at least 5%. The five failed that test, reducing the universe to six symbols

In addition, two have earnings announcements within 30 days of Monday, when trading resumes.

That exclusion shrinks my universe to four symbols, one of them an exchange-traded fund. The stocks, in descending volume order, are FLR, MENT and FFBH. The fund, with extremely low volume, is UMX.

My redesigned universe is a super-set of the Zacks-based universe. It includes the two survivors from my Zacks analysis, but doubles it to four.

The new universe delivers more symbols, but at the cost of liquidity. The universe of 50 stocks having very high volume would produce few trading signals, but the ones that appeared would allow me to trade options structured a vertical credit spreads, providing both liquidity and a hedge.

Many more of the signals drawn from my redesigned universe will be rejected because of the logistics of trading.

On Monday I'll be looking at the four survivors from my new analytical universe and will use that universe all week, to get a sense of whether it really is better.