The model is based on 9 paired trades of extremely liquid stocks. In most cases they are some of the largest market-capitalized investments in their sector. These are the mega-caps typically. The idea here is to embrace liquidity in size.
Let’s dive right in and take a look at what stocks we have in the model, why they are in there, and what is driving the thoughts behind the process of selecting them.
I have set up this test model with a number of intentional failures, to see how it would treat these events on its own. Let’s review those trades, and see how the model is coping with them.
The model is reporting three false readings on its stable trade checkpoint. These would trigger the removal of the trade from the basket, and a new replacement trade being plugged in to redeploy the capital.
I am using .75 as the hard cut-off for a trade that is flirting with an equal ratio. In each of the test cases, the model would reject the trades, while two of the three are profitable trades.
As is obvious, the model caught the Coke versus Pepsi trade and would have put it on in the reverse direction. This was a planned test, and I was glad to see it. Coke has overperformed Pepsi YTD, and while the model is exposed to it from a test point of view, it would never have been put on or kept in.
The test model has a false positive trade embedded with its love of Ford over GM, while the equity markets have sold Ford and bought GM in comparison to each other YTD. This trade is listed as a true trade, but is in fact underwater as will be seen below. The trade would be allowed to stay in, due to its positive reading. It would get rejected if its losses reach 1.5% of the portfolio AUM.
The following model snapshot was taken end of yesterday during the writing of this article.
The only real failure in the model selection process is JPM versus Wells Fargo in the battle of the TBTF.
JPM is up YTD more than WFC is, and even with the model saying to reverse this trade I would instead pull it and watch. The model is trying to put on a losing trade YTD here.
While it may turn out to be correct, the vagaries of bank balance sheet reporting are such that I would suggest the model is only as good as the data being put into it. I believe, in all cases, a model should still have a human reviewing the individual trades.
The third false trade reading in the test model is based around XOM versus Chevron. These two companies are almost identical, and while the market prefers XOM over CVX, the model would reject this trade and book the profits. The Ratio strength level has hit a point that would generate a sell.
At this time, the model should have the PEP versus KO trade kicked out due to false ratio before .75, along with the XOM versus CVX trade and the JPM versus WFC. This would leave in the F versus GM trade due to it only being underwater, but is expected to turn around due to its current model score.
In reviewing the rest of the Model Portfolio, the additional trades, specifically CLR versus PBR, SRE versus EQT, AAPL versus GOOG, DVN versus CHK, and MSFT versus ORCL are positive trades, with a positive score in the selection process.
I will return to discuss the successful trades and the sub-sector scoring engine in future articles in this series.
Jack H Barnes Jr.
CEO & Founder