Stock Market Forecast: July through December,2017: Slightly above average

The forecasting models continue to predict a typical weak/flat stock market during the summer months followed by a good upturn in the fall.  The one month forecast is for a gain of just 0.1% for July and nearly a 6% gain following through the end of December.  The model gives the probability of at least breaking even over the period as slightly over 90% -- decidedly better than average.

The stock market started off this year strong -- much better than my models had predicted. More recently the actual and predicted market performance have been close.  The typical causes of a stock market crash seem to be taking the summer off.  Be grateful for small favors.

(Click on image to enlarge)

Stock Market Forecast June through November, 2017: Looking Better

My 4 semi-independent six month stock market forecasting models are in general agreement.   The one-month forecast is for a negligible gain of just 0.1%.  Once the weather starts cooling, the outlook improves. The composite prediction is a gain of roughly 6% from June, 2017 through November, 2017.  That is better than the average gain of 4.8%.

The likelihood of at least breaking even over the next six months is around 90%.  Historically, stocks break even or rise about 70% of the time over all 6 month stretches since 1984.

My models still do not know how to read newspapers so they can't account for any of the political foibles are are now playing out in Washington. Wish I was so lucky.

Have a nice summer.  It doesn't look like anything really bad is on the horizon.

Stock Market Forecast May through October, 2017

The current six month stock market forecast for May through October, 2017 is nearly flat  -- a sub-normal 1% gain over the coming half-year, and no gains during May.  The likelihood of the market at least breaking even over the next half year is 0.71, slightly less than average. Meh.

According to MarketWatch.com the Dow Jones Industrial Average has had record gains (14%) from the election through President Trump's first 100 days in office -- best returns for the start of a president's first term since WWII.  My prediction models didn't see that coming.

The models anticipated flat market performance for most of the past half-year.  So, either my models were just plain wrong, or more probably the stock market was propelled not by economic data, but by hope of wonderous Trump-promised gains in corporate profits.  The 'Trump Bump, however, may have stalled -- for the past two months the market has been flat.  I would not be surprised if the stock market stumbles if the hopes of huge profit gains vanish as Trump's version of tax reform flounders in Congress.  Either way, the basic economic data does not presage a huge market move up or down.

Stock Market Forecast April through September, 2017: Start to worry

My U.S. stock market forecasting models expect a flat or weak stock market for the coming month and the next six months -- not too surprising as the second half of the year tends to be weaker than the first half.  For April, the model expects a minuscule gain of 0.2%.  The next 6 months are projected to yield a loss of 1%.  The models are just a rough estimate of future results, so the current forecast is an expectation of little market change.

Forecasts from all of my semi-autonomous models have been declining for  several months.  If that continues, my forecasts and the market will get even weaker over the summer.  (Read below chart.)

The next graph shows, beginning in 2007, the actual 6 month return of the market (black) along with the projected return rates forecast by my models.  (I use the Value Line Arithmetic Average as my market index.) The red line is the composite projected by combining the forecasts of several of my models.  For the past several months the composite forecast and all the semi-independent forecasts have trended down.So far, there is no reason for alarm.

Stock Market Forecast March through August, 2017: Weaker

The six month stock market forecasts from my models range from a possible gain of 6% to a loss of -2%. The composite prediction is for a loss of roughly -1% with a probability of at least breaking even of 0.70, which is slightly below average.  Generally, the models do not see a basis for the sharp stock market gains since the election. The forecast for just March, however, is positive.

About the chart:  Each data point was an actual U.S. stock market forecast made in real time starting in 2007.  Red points were predictions that the market will fall -5% or more over the six months following the forecast.  Green points were positive or neutral market expectations for the coming six months.

Stock Market Forecast February through July, 2017 -About Average

Through July, 2017 my forecasting models are expecting the U.S. stock market to rise about 6% to 8% . That is better than average.  The models see the probability of the market at least breaking even as somewhere between 66% and 85% -- roughly about average.

Besides my tested original model,  monthly market forecasts going forward will be derived from several additional, largely independent, econometric models. All of the models stem from business/economic fundamentals rather than Technical Analysis or other forms of trend projection.

What is interesting is that a new model largely based on corporate profits and a different model based mainly on various interest rates  'tell' basically the same story over several decades.  My hope is that with this broader base of economic variables, the net effect will be that the outputs will be less susceptible to bits of data that are unusual, and probably misleading, outliers.

As always, don't bet the farm on these models.  They now have a significant experience base, but reality will often be different from expectations.

For example - my models have no direct knowledge of the administration of Donald Trump.

On a personal side, I don't see that ending well.

There is far more downside potential than upside potential in the market at present. The Market Fair Value chart at morningstar.com reports that the market is currently estimated as 3% overvalued.  Go to the 'Max' time period view.  The market seldom rises much above the 3% value without a significant correction fairly soon thereafter.

Evaluating a Decade of Results from My Stock Market Forecasting Model

Each month since 2007 I have posted the 6-month predictions for the U.S. stock market generated by my econometric forecasting model. That amounts to 116 monthly forecasts that I can now evaluate. How well did my model perform?

There are several ways to judge performance. Cumulative return on investment, however, ends up being the best measure of success or failure. Based on cumulative returns, following my models would have produced about three times the return of a Buy-and-Hold strategy.

Not that Buy-and-Hold is a bad idea. If you followed a Buy-and-Hold strategy from 2007 through July, 2016, simply holding an S&P 500 index fund, your holdings would have grown 51% plus dividends. Pretty good for a portfolio set on autopilot.  Especially considering that shortly after the test period started the stock market crashed horribly and took years to fully recover. You still would have come out OK.

Acting upon the 6-month forecasts from my model would have been somewhat better than just following a Buy-and-Hold strategy. (i.e. Buy when the six-month forecast was positive and sell when the 6 month forecast turned negative.)  But, while the 6-month forecasts were surprisingly accurate, they really didn't say much about what the market was likely to do in the month immediately following a forecast.   In the end, they didn't do very well at picking the best buy and sell points. For example, my model was generating fantastically positive  6-month performance forecasts while the stock market was still crashing down in 2008-2009.  The market did, indeed, climb over the following 6 months, very nearly as expected. But, in the meantime the stock market was still falling like a rock. Buying when the 6-month forecasts first turn positive or first turn down ends up not being such a good idea.

 A while ago I learned that I could apply different weights to several of my 6-month forecasts from previous months to give a better decision on buy and sell points.

An investment in the SP 500 index that followed the forecasts generated by the weighted predictions from my model would have largely missed the market crash of 2007-2009 and would have gained 169% -- over three times the return from a Buy-and-Hold strategy.

An important caveat is on order.  The calculation above does not consider dividends. Dividends would have been somewhat less for the trading strategy since the strategy would have taken you out of the market for over a year.  Also, the net gain would be significantly less for the trading approach for stocks held in a taxable brokerage account due to taxes levied on profits from sales of stocks.

That said, being able to dodge a major market crash can significantly beat a Buy-and-Hold strategy -- especially in a tax-favored account such as an IRA. It looks like my forecasting model is doing what it is supposed to do.