Predictive Model Output - Nov 25, 2005

[nodoodahs]

From the data as of November 25 2005 closing, with the exception of the M2-based variables, since the aggregates haven’t been updated by those lazy folks at the St. Louis Fed yet. I’m leaving those inputs at last week’s levels. Keep in mind that I am doing the models now with M2 instead of the more predictive M3 now, because Bubbles Bernanke prefers the “stealth” helicopter to the regular version.

Last week, the EPCR had been progressively more bullish in terms of consensus, with the 20 day and the 100 day EPCR both topping out on Tuesday and declining through the rest of the week. This week, the 20 day EPCR carried on that trend while the 100 day EPCR was fairly stable. Our 21 day return on the S&P 500 has been tremendous, a 6.5% increase. Normally the EPCR model has a good bit of mean regression component, but more extreme returns tend to soften that effect somewhat. The increasing bullishness of sentiment in the EPCR tends to lead to a lower prediction as well. Output for this model remains in the 2nd decile, which is bearish for the next 21 sessions, suggesting average to flat returns. This model suggests actual negative returns only when output falls into the 1st decile.

The 13-week model for the S&P 500 is has moved from reading near the bottom of the 9th decile, bullish, to a 7th decile reading, which is average. A 7th decile reading would suggest median and average returns of 2.9% for the next quarter, with greater than 65% chance of gains, 50% chance of above-average gains, and 2% chance of ending the quarter down more than 10% from where we are today. For reference, the data period used to build the model has an average and median quarterly gain of about 2.8%.

The 52-week model for the S&P 500 has moved from the 7th decile, bullish, to the 6th decile, neutral in the sense that it suggests an average 52-week return. Median and average returns for this reading exceed 11%, with 80% chance of gains, greater than 50% chance of above-average gains, and a relatively hefty 13% chance of ending the year down more than 10% from where we are today.

Overall the models suggest not exactly a screaming buy signal, but nothing to be scared of, either. I would still think a moderate correction in the S&P 500 might come in the next few weeks. Personally, I tend to buy individual stocks as soon as I’m convinced they are a bargain, regardless of what the overall market is doing, unless and until I get some serious 1st decile readings from these models.

The 13-week model for the 10YT is way down in the 1st decile, and has been for four of the last five weeks including last week. From the current yield, this is suggesting that yield could drop below 4% by the end of the quarter. While the standard deviation is high for results when output is this low, the model suggests a 95% chance that yield will be lower in a quarter from now.

The 52-week model for the 10YT is down in the 3rd decile, suggesting that yield should be below 4% in a year. At any rate, the model suggests 93% odds the yield will be lower in a year. For seven of the last eight weeks, this model has been in the 3rd decile, with one output in the 2nd decile. I don’t know if I believe that the yields will get below 4%, but that is the model output, and I can believe that yields will stay low.

The overall indication is for lower yields from today, or at least not very much higher, if at all.

The 13-week model for the USDX has moved up from the 4th decile to the 5th decile, still suggesting 60-65% odds that the USD will decline short-term by a point or so. Standard deviations are very high compared to median and average output for this model, so I think of this model more as a general directional indicator.

The 52-week model for the USDX has also risen, from the 6th decile, which is neutral for the USD, to the 7th decile, which is somewhat more bullish in amount of rise if not odds of rise. The chance of a rise in the USDX is good, but one funny thing about the USDX is that generally, it rises. The declines in the USDX are rarer but larger than the increases. Example, from Nov 1980 the USDX has risen 52.5% of the time in each year period, but the average increase over each year period is negative 1 point, meaning that large declines are skewing the average.

Both of these new models, the 13- and 52-week models for the USDX, have rolled over in the last 2-3 months, and the slight increases of the last week haven’t changed that – the readings for this week are still WAY lower than anything seen from July through September. Three months ago the 13-week model was 10th decile for increasing USDX, and less than two months ago the 52-week model was 9th decile.

Overall indication is the USD will top out sometime next year.
_________________
I haven’t seen a beatin’ like that since somebody stuck a banana in my pants and turned a monkey loose.