AMZN seasonality

Kresmion takes the last daily close of each calendar month, adjusted for splits and dividends, and computes each month's return against the previous month-end. For each calendar month it reports the mean, median and share of years with a rise across every completed year, the highest and lowest year, and t = mean divided by its standard error; an average with |t| under 2 is marked as within noise. The test is against zero, not against the other months, so in a rising market up months clear it more easily than down months; shuffling each ticker's monthly returns across the calendar shows how many tiles clear it with no calendar pattern at all. Only stocks and funds with ten or more full calendar years are listed. The month in progress is shown separately and is never part of an average.

• Closes are daily closing prices adjusted for splits and dividends, so a stock split never reads as a fall. One limit: a dividend paid after the stored history was last rebuilt (September 2026 for the large caps, spring 2026 for the index funds) is not yet folded into the older closes, so that month reads about one dividend low (0.1 to 0.5 points for these names).
• A month's return compares the last close of the month with the last close of the previous month. Only consecutive completed months count; the month in progress is shown on its own ("so far") and is never part of an average.
• Window: each stock or fund from its first full month of stored history (2000 for the oldest, the listing month for later ones) to the last completed month: for AMZN, February 2000 to October 2026, 25 full calendar years. A ticker is listed only with 10 or more full calendar years. Not listed yet: DIA, XLK, XLF, XLV, XLE, XLI, XLY, XLP, XLU, XLB, XLRE, XLC.
• Per calendar month: the mean and median across years, how many years rose, the highest and lowest year, and t = mean divided by its standard error (the standard deviation across years over the square root of the number of years). An average with |t| under 2 is within noise: it cannot be told apart from zero with this many years.
• The average year is the mean, across years, of the return from the start of January to each month-end; its December point is the average calendar-year return.
• The test is against zero, not against the ticker's other months. Stocks rose in most years, so an up month clears it more easily than a down month of the same size. With each ticker's monthly returns shuffled across the calendar, which removes any calendar pattern but keeps the general rise, about 25 of the 156 ticker-month averages still clear |t| of 2 (28 do on the real calendar), so a single ringed tile is weak evidence. Calendar averages describe the past; they are not forecasts and not trading advice. Only monthly statistics are shown here, never a price series.