The best-known calendar anomaly, the "January effect," suggests stocks, especially smaller ones, do best then, clearly the case this past January, the best since 1987. Other stock market anomalies identify September as the worst month for investing, Monday the worst day and Friday the best.

My research concluded that calendar anomalies also exist with business cycles. I coined the term "Turn Of Decade" or "TOD" effect in November 1990 to refer to the unusually high likelihood of a recession beginning in years ending in 9 or 0.

I stated then that a mild recession had begun in August or early September of 1990 and would end the following year. The National Bureau of Economic Research later declared a recession began in July 1990 and ended in March 1991.

In January 1999 I predicted a Y2K Recession, noting increased consumer and business uncertainty entering the new decade. The TOD effect was again accurate, but there was a rare controversy over its start date as either the end of 2000 or beginning of 2001: the former would be politically billed as the Bill Clinton (Democrat) Recession vs. the latter as the George W. Bush (Republican) Recession.

The NBER reported a March 2001 start date based on only one (employment) of its three indicators, with the other two, industrial production and real sales, suggesting September and August 2000, respectively.  Harvard professor Greg Mankiw and the Council of Economic Advisors later found that the data on which the NBER relied was substantially revised, with the new data documenting a late 2000 start date.

Two well-known time-series econometricians subsequently analyzed all 14 NBER start and end dates for the seven recessions between 1960 and 2001 and verified all of them, except the Y2K recession start date, which they concluded was September 2000 rather than March 2001. So, it is reasonable to refer to a Y2K recession, but there is disagreement on its precise start date.

The TOD effect did not predict the Great Recession, which began in December 2007 and ended in June 2009, although the economy was still in a deep funk in 2010 when bank failures peaked and the housing market bottomed.

Using official NBER dates, we find that four of the previous eight (50%) and 11 of the 33 recessions (33%) since 1854 were TOD recessions; these results would increase to 63% and 36%, respectively, counting the Y2K Recession.

If recessions were totally random occurrences, we would expect 20% beginning in years ending in 9 or 0, rather than 33% (or 36% counting Y2K). Also, the TOD effect's track record is nearly 70% (or 75% counting Y2K) for the 16 turn-of-decades since 1854, when records were first kept.

Anomalies aside, many factors suggest a 2020 recession starting with inverted yield curves, trade wars and the global recession virus. Absent the unknowable "X" factor, such as impeachment efforts, which we also saw in 1999, perhaps the most telling recession indicator is the increasing uncertainty with the presidential election.

Just like President Donald Trump is reversing Obamacare, a Democratic president could do the same with recent tax cuts, regulatory relief and other pro-business reforms. This election uncertainty could reduce business investment, economic activity, stock prices, and overall confidence.

Consumer and business confidence is a critical determinant of the economy's direction. Borrowing from Nobel laureate Robert Solow's famous inflation quote, it is fair to say, "We will have a recession because we expect a recession, and we expect a recession because we have had recessions."

If we reduce the TOD effect to just turn-of-decades with presidential elections, where there is always some uncertainty, we can call this the presidential election TOD or PETOD effect. There have been eight turn-of-decades since 1854 with a presidential election starting with 1860 and every 20 years thereafter, namely 1880, 1900, 1920, 1940, 1960, 1980 and 2000.

The PETOD effect existed in five of those eight elections with recessions beginning during each election year in four cases (1860, 1920, 1960 and 1980) and in the previous year in one case (June 1899 before the 1900 presidential election). The PETOD effect of 63% (five of eight) would have been 75% (six of eight) counting the Y2K recession. Also, the PETOD effect existed for two of the last three (67%) and four of the last six (67%) such presidential elections; these results would have been three of the last three (100%) and five of the last six (87%) counting Y2K.

Considering traditional economic factors and my nontraditional anomalies, I believe there is an 80% probability a recession will begin next year.

Kenneth H. Thomas, president of Miami-based Community Development Fund Advisors LLC, taught finance at the University of Pennsylvania's Wharton School for over 40 years.