The best time to buy the Presidential Election Cycle is from September of the second year to April of the third year. Nevertheless, the fourth quarter of the third year is strong, particularly after a weak third quarter. In the past, it was better to buy near the end of October than at the end of September. How does the fourth quarter do in the third year of the Presidential election cycle? ‘Everyone knows’ that the third year of the Presidential cycle is incredibly reliable, and has returns that far exceed the other three years. Even Grantham has touted it, which I thought must be tongue-in-cheek, because he is a macro-guy. So I decided to go back and check, and found his letter written at the end of the third quarter in 2014 for GMO. It turns out he was quite serious. Regular readers know the score: +2.5% a month for the seven months from October 1 to April 30, in year three on average since 1932 (a total of +17%). This is now the 21st cycle. The odds of drawing 20 random 7-month returns this strong are just over 1 in 200 according to our 10 million trials. But 17 of the actual 20 historical experiences were up, and the worst of the 3 downs was only -6.4%, so the odds of this consistency plus the high return would be much smaller. The remaining 5 months of the Presidential year have a good but not remarkable record, over .75% per month, but the killer here is that the remaining 36 months since 1932 averaged a measly +0.2% a month!” Reference to the remaining 5 months means that Grantham views the third year of the Presidential cycle as running from September to September. More importantly, we have missed the key months from September 30 to April 30. From 2014 to 2015, that time span had the S&P 500 rising by 11.39%, which is not too shabby given what the market has done since. Yahoo Finance only had S&P 500 data as far back as 1950. So my analysis is for the 16 third years since then (see the table below). We have completed 17 years from his September to April time frame, however, and I calculated an average 19.72% return for those time periods, with a median return of 19.49%. There was only one decline of -.76% in 1978-79. But dividends have not been included. So every period actually had a positive total return. For the full calendar third year, the average return was 17.12%, with a median return of 18.08%. That’s very good also, but not as good, and that is a 12-month return versus Grantham’s 7-month return. For all years since 1950, the average calendar year gain was 9.18%. Therefore, the average gain in the other 3 years of the Presidential cycle works out to 5.69%. Out of the 16 third years, 15 were up, and one was unchanged (2011). With stocks down YTD, the odds would appear to be good that we will get a nice rally over the last three months. I say ‘appear to be good’, because statistically we can’t calculate the odds. This is a small sample. It is not a random sample. And there is no solid theory to support why the pattern of the recent past should hold in the future. Let’s see how the last three months of the third year have done since 1950. From 9/30 to the end of the year, the average gain in the S&P has been 3.04%, with a median return of 4.39%. The mean is lower because of the skew created by 1987. Third Year Pres. Cycle %ch. Oct. 31 to end of yr % ch. Sept. prev. yr to April 3rd yr % ch. Full 3rd year % ch. April to Sept. 3rd yr % ch. 9/30 to end of 3rd yr % ch. Sept. low to end 3rd yr % ch. Oct. low to end 3rd yr % ch. Sept. 30 to Oct. low % ch. Sept. low to Oct. low 1951 3.62 15.32 16.35 3.7 2.19 2.19 4.9 -2.58 -2.58 1955 7.42 17.49 26.40 15.04 4.14 6.74 11.47 -6.57 -4.25 1959 4.12 15.04 8.48 -1.23 5.29 8.61 6.95 -1.55 1.56 1963 1.36 24.04 18.89 2.72 4.63 4.63 4.38 .24 .24 1967 3.40 22.79 20.09 2.87 -.25 2.98 3.4 -3.53 -.41 1971 8.34 23.31 10.79 -5.4 3.81 4.58 8.85 -4.63 -3.92 1975 1.29 37.39 31.55 -3.93 7.54 9.86 8.75 -1.12 1.02 1979 5.91 -.76 12.2 7.43 -1.35 1.35 7.84 -8.53 -6.02 1983 .84 36.55 17.27 1.00 -.69 .43 .95 -1.63 -.52 1987 -1.87 24.66 2.03 11.61 -23.2 -20.4 9.89 -30.1 -27.6 1991 6.28 22.64 26.31 3.31 7.56 8.73 10.69 -2.83 -1.77 1995 5.92 11.24 34.11 13.54 5.39 8.28 6.65 -1.18 1.53 1999 7.8 31.28 19.53 -3.93 14.54 15.84 17.78 -2.75 -1.65 2003 5.83 12.47 26.38 8.62 11.64 11.64 9.20 2.23 2.23 2007 -5.23 10.97 3.53 2.99 -3.82 1.15 -2.15 -1.71 3.37 2011 0.34 19.49 -.003 -17.0 11.15 11.34 14.41 -2.5 -2.69 2015 11.39 -7.93 Mean 3.46 19.72 17.12 1.96 3.04 4.87 7.75 -4.32 -2.59 Med. 3.87 19.49 18.08 2.87 4.39 5.68 8.30 -2.67 -1.09 (The median date of the September low is the 21st. The median date for the October low is the 17th.) The average fourth quarter gain for all years since 1950 is 4.06% with a median of 4.92%. So the third year of the Presidential cycle has a lower average using both measures. The much lower mean is probably because of 1987, but clearly the fourth quarter of the third year is actually not as good as other years. There were 5 down quarters out of 16. They were 1967, 1979, 1983, 1987 and 2007. But all 5 years that declined from April to September 30 (1959, 1971, 1975, 1999, and 2011) had good gains in the fourth quarter . This augurs well for 2015, but 5 out of 5 does not mean we have to get 6 out of 6. The average gain for the two months following October 31 was 3.46% with a median of 3.87%. I don’t know what the comparable percentages are for all years. Two years had declines – 1987 and 2007. So the return is better for the last two months than the last three months. This should not be a surprise. I compared the October lows with the September lows, and found that on average (in the third year), the October low was 2.59% lower than the September low (see the table). October had a lower low in 10 out of 16 years. If you can identify the October low, then the average gain from there to the end of the year was 7.75% with a median of 8.30%. 2007 was the only down year with a loss of -2.15%. Locating the vicinity of the October low is not as stupid as it sounds. The median low date was October 17th. Unfortunately, the 1987 crash was on the 17th, 18th and 19th with the huge losses on the 19th (I remember it well. I was 100% invested and canoeing a river in Missouri.). Eight of the 16 lows were on the 19th or later. Three of the lows were on the second to last or last day. So if you buy at the close on the third to last day, you should be able to beat that average return dated from the end of October. The last two days in October are pretty good on average. I will buy stocks when Financial Select Sector SPDR ETF (NYSEARCA: XLF ) hits a twenty-day high (adjusted for dividend payments). The levels are posted in my Instablog. I actually buy small caps when XLF hits a twenty-day high. I compared the Russell 2000’s performance in the fourth quarter of the third year with the S&P 500 since 1987, and found that on average the S&P did slightly better. The R2000 is more volatile. In strong fourth quarters, it beat the S&P. In weak fourth quarters, it underperformed badly; e.g. 1987. I’m pretty optimistic about the last two months of the year. There is a strong possibility that October will be bad, because of all the negative macro- indicators. Risky high-yield investments like MLPs, mREITs, and junk bonds have been hammered. Sentiment is very negative as indicated by Investors Intelligence, Hulbert’s sentiment measures, Rydex, and Citigroup’s Euphoria/Panic model. I think sentiment follows the market. If October brings further drops in stock prices, then these measures will become even more negative, but that will set us up for a bigger bounce into the end of the year.