BigCapital beta

Article · Money

Irrational money: how people trip over their own feet

Six common money habits, each run as a plain rule on real data. Missing just the 10 best of 1,869 months since 1871 took 81% off a stock investor's result; selling at every crash bottom left 1.6% of it; and a $2 Powerball ticket returns $0.32 in prizes below the jackpot. One habit, though, refused to look silly.

Updated with Oct 2026 data12 min read

Missing the 10 best of 1,869 months since 1871
−81%

$10,000 from January 1871 became $11.9B invested every month, $2.21B without the 10 best

Raises that were pay cuts
25 of 61

years 1965–2025: average hourly pay rose in dollars but bought less

The 10 best months that began 20%+ below a record
7 of 10

only 35% of all months began that deep: the best came in the depths

Last year's winner won again
26 of 64

1962–2025; yet chasing it turned $10,000 into $17.9M by Oct 2026, an equal mix into $1.44M

Waiting for a Powerball jackpot, one ticket a draw
1.87M yrs

odds 1 in 292,201,338, 3 draws a week

What it cost: six habits, $10,000 each

Each habit below is run as a plain rule on real data, starting with $10,000, and set against doing nothing special. Each test covers the longest stretch its data allow, so the periods differ: compare the bars, which show the share of the plain result the habit kept, rather than the dollars.

What $10,000 ended as, habit by habit (bar: the share of the plain alternative that the habit kept)
  1. 1
    Sold at every crash bottom, back 12 months laterLoss aversion1871–2026
    Held through$11.9BPanic seller$186M
    −98.4%
  2. 2
    Kept $10,000 in an account paying 0%Money illusion1964–2026
    Put in$10,000Buys today, in money of then$920
    −90.8%
  3. 3
    Spent $10,000 on Powerball ticketsOverweighting tiny chancesAny draw · 5,000 tickets
    Spent$10,000Expected back below the jackpot$1,599
    −84.0%
  4. 4
    Missed the 10 best monthsMarket timing · loss aversion1871–2026
    Invested every month$11.9BWithout the 10 best$2.21B
    −81.4%
  5. 5
    Sold in May, back in NovemberCalendar folklore1954–2026
    Held all year$18.6MSell in May$11.1M
    −40.4%
  6. 6
    Bought last year's winner every yearRecency bias · herding1962–2026
    Equal mix of four$1.44MWinner chaser$17.9M
    ×12.4

Each row runs over its own years, the longest its data allow. Cash before July 1954 earns 0%: the data have no rate before then. The lottery row counts the prizes below the jackpot; the 5,000 tickets also carry a 1 in 58,441 chance at the jackpot itself.

Missing the best months

Market timing promises to dodge the bad months. The catch is the good ones. From January 1871 to October 2026, $10,000 in US stocks with dividends reinvested grew to $11.9B. Leave out only the 5 best of those 1,869 months — sitting in cash at 0% instead — and it ends at $3.83B; without the 10 best, $2.21B; without the 20 best, $844M.

The best months were not calm ones. 7 of the 10 best began with stocks 20% or more below their record, although only 35% of all months began that deep, and 5 came within a year after one of the 10 worst. The very best, August 1932 (+51.3%), came 4 months after April 1932 (−23.3%).

Dodging the worst months is the dream: skipping only the 10 worst would have multiplied the result ×7.08. But the two kinds sat so close together that leaving after a bad month risked missing the rebound: skipping both the 10 best and the 10 worst ended at $15.7B, +31.7% against staying put. Loss aversion — the finding of Daniel Kahneman and Amos Tversky that a loss hurts more than an equal gain pleases — is what makes stepping out feel safe.

One of the 10 best monthsOne of the 10 worstOther months: darker, bigger move
Every month since 1871, a decade per row; return with dividends from one monthly average to the next. Arcs join a best month to a worst one less than a year before it. Hover or tap a month. Data: Robert J. Shiller, Yale University
$10,000 from January 1871 to October 2026, dividends reinvested
  • Invested every month$11.9B100%
  • Without the 5 best months$3.83B32.1%
  • Without the 10 best months$2.21B18.6%
  • Without the 20 best months$844M7.1%

A missed month earns 0%, like cash under a mattress. Percentages: the share of the fully invested result.

The 20 best and 20 worst months
Best months
#MonthReturn
1Aug 1932+51.3%
2May 1933+29.3%
3Jul 1938+20.5%
4Jun 1933+17.6%
5Apr 2009+12.3%
6Sep 1982+12.1%
7Feb 1991+11.6%
8Oct 1938+11.6%
9Oct 1879+11.3%
10Apr 1933+11.2%
11Sep 1939+11.1%
12Nov 1998+11.0%
13Feb 1975+10.8%
14Sep 1932+10.3%
15Jun 1901+10.3%
16Dec 1873+10.1%
17Nov 1935+9.7%
18Jan 1976+9.5%
19May 1997+9.2%
20Aug 1984+9.2%
Worst months
#MonthReturn
1Nov 1929−26.2%
2Apr 1932−23.3%
3Oct 2008−20.2%
4Mar 2020−18.9%
5Dec 1931−18.1%
6Sep 1946−14.4%
7Sep 1931−14.4%
8Oct 1937−14.1%
9Sep 1937−13.8%
10Oct 1930−13.4%
11May 1940−13.3%
12Oct 1932−13.2%
13Oct 1931−12.8%
14Jun 1932−12.4%
15Nov 1987−12.3%
16Apr 1939−12.2%
17Oct 1987−11.9%
18Jun 1962−11.4%
19May 1932−11.3%
20Jul 1974−11.3%

Selling at the bottom

Since 1871 US stocks fell 20% or more from a record and later climbed back to a new one 13 times (the rule of our crashes article). Picture an investor who held through each fall and broke at the very bottom: sold everything in its worst month, kept the money in cash and came back 12 months later. The stayer won every one of the 13 races: a year after the bottom the seller trailed by 9.2% to 57.4%, 23.1% at the median. The cruellest was 1929–32: in the year after the bottom the market gained +134.7% with dividends, the seller's cash 0.0%.

Repeat it at all 13 bottoms and the habit compounds. $10,000 from January 1871, held through everything, grew to $11.9B; the panic seller ended with $186M — 1.6% of it. Waiting 24 months before coming back left $34.2M, or 0.29%.

Selling in the exact worst month is the worst case — real panic sales are spread across a fall — yet the bottom is also when selling feels most urgent. Before July 1954 the data have no cash rate, so in the 6 earlier crashes the seller's cash earns nothing here; counting only the 7 bottoms since then, the seller kept 18% of what the stayer had.

Stayed investedSold at the bottom: cash, then back in
$10,000 at each bottom (the month with the lowest average price), over the next 36 months, dividends reinvested. Cash earns the fed funds rate from July 1954, 0% before. Data: Robert J. Shiller, Yale University
$10,000 at each bottom: held, or sold and back later
CrashFall12 months after the bottom24 months after the bottom
StayedSoldStayedSold
1872–77*bottom Jun 1877−47.3%$13,292$10,000−24.8%$16,240$10,000−38.4%
1881–96*bottom Aug 1896−42.1%$13,009$10,000−23.1%$14,991$10,000−33.3%
1902–03*bottom Oct 1903−29.3%$13,011$10,000−23.1%$16,293$10,000−38.6%
1906–07*bottom Nov 1907−37.7%$14,945$10,000−33.1%$17,996$10,000−44.4%
1909–21*bottom Aug 1921−37.4%$14,562$10,000−31.3%$14,158$10,000−29.4%
1929–32*bottom Jun 1932−84.8%$23,468$10,000−57.4%$23,438$10,000−57.3%
1961–62bottom Jun 1962−22.5%$13,036$10,300−21.0%$15,378$10,664−30.7%
1968–70bottom Jun 1970−29.0%$13,662$10,550−22.8%$15,249$11,049−27.5%
1973–74bottom Dec 1974−43.4%$13,804$10,622−23.1%$16,913$11,183−33.9%
1987bottom Dec 1987−26.8%$11,880$10,791−9.2%$15,469$11,849−23.4%
2000–03bottom Feb 2003−43.7%$13,889$10,112−27.2%$14,818$10,269−30.7%
2007–09bottom Mar 2009−50.8%$15,580$10,015−35.7%$17,981$10,033−44.2%
2021–22bottom Oct 2022−20.3%$11,647$10,488−10.0%$16,022$11,068−30.9%

* Bottom before July 1954: the seller's cash earns 0% (no rate in the data). Cash: effective federal funds rate, Board of Governors of the Federal Reserve System, H.15

The EFFR is subject to the Terms of Use posted at newyorkfed.org. The New York Fed is not responsible for publication of the EFFR by BigCapital, does not sanction or endorse any particular republication, and has no liability for your use.

Chasing last year's winner

Recency bias makes last year look like the future; herding makes the crowd's favourite feel safe. The test: every January from 1962, put everything into whichever of four assets — US stocks, 10-year Treasury bonds, gold or cash — did best in the year just ended. Against it: an equal mix of the four, set back to a quarter each every January.

Here the habit refused to look irrational. Last year's winner won again in only 26 of 64 years, yet $10,000 from December 1961 grew to $17.9M chasing it — against $1.44M in the equal mix and $6.45M in stocks alone. Winners often kept running for another year, a pattern researchers call momentum, and the long runs paid for the misses. The longest: stocks, picked 6 years running, 2013–2018.

The price was the ride. The chaser's worst year was 1975, when gold returned −24.2%; the mix's worst was 2022, at −7.0%. The chaser trailed the mix in 27 of 64 years. Picks over the whole test: stocks 29, gold 21, bonds 8, cash 7. It is one test of four broad assets and says nothing about single stocks or funds.

US stocks10-year TreasuriesGoldCash✓Won again
  1. 1962500−10%
  2. 1963%+2%
  3. 1964500+17%✓
  4. 1965500+13%✓
  5. 1966500−8%
  6. 1967$+4%
  7. 1968500+15%
  8. 1969Au−14%
  9. 1970$+8%
  10. 1971%+10%
  11. 1972Au+47%✓
  12. 1973Au+67%✓
  13. 1974Au+72%✓
  14. 1975Au−24%
  15. 1976500+23%✓
  16. 1977500−6%
  17. 1978Au+30%✓
  18. 1979Au+119%✓
  19. 1980Au+18%
  20. 1981500−2%
  21. 1982$+13%
  22. 1983%+3%
  23. 1984500+5%
  24. 1985%+28%
  25. 1986500+24%
  26. 1987%−5%
  27. 1988Au−14%
  28. 1989500+30%✓
  29. 1990500−2%
  30. 1991$+6%
  31. 1992500+16%✓
  32. 1993500+10%
  33. 1994Au−1%
  34. 1995$+6%
  35. 1996500+24%✓
  36. 1997500+32%✓
  37. 1998500+25%✓
  38. 1999500+22%✓
  39. 2000500−6%
  40. 2001%+6%✓
  41. 2002%+14%
  42. 2003Au+23%✓
  43. 2004Au+9%
  44. 2005500+7%
  45. 2006Au+23%✓
  46. 2007Au+28%✓
  47. 2008Au+2%
  48. 2009%−7%
  49. 2010Au+23%✓
  50. 2011Au+18%✓
  51. 2012Au+3%
  52. 2013500+30%✓
  53. 2014500+16%✓
  54. 2015500+2%✓
  55. 2016500+12%✓
  56. 2017500+21%✓
  57. 2018500−2%
  58. 2019$+2%
  59. 2020500+18%
  60. 2021Au−4%
  61. 2022500−15%
  62. 2023$+5%
  63. 2024500+30%
  64. 2025Au+63%✓
  65. 2026*Au−5%
Each tile: a year, last year's winner bought for it, and what that asset returned (December average to December average). 2026*: to Oct 2026, so far. Gold: World Bank, Commodity Price Data (Pink Sheet), CC BY 4.0 · Bonds: 10-year Treasuries with coupons reinvested, our calculation from Shiller's long-term interest rates
Last year's winnerEqual mix, reset yearlyUS stocks
$10,000 from December 1961, switched or rebalanced every December; log scale: equal steps are equal percentage changes. Data: Robert J. Shiller, Yale University
$10,000 from December 1961 to October 2026
HeldBecameA year
Last year's winner$17.9M+12.2%
Equal mix, reset yearly$1.44M+8.0%
US stocks$6.45M+10.5%
10-year Treasuries$368K+5.7%
Gold$1.16M+7.6%
Cash$242K+5.0%

Calendar folklore: Sell in May, and January

“Sell in May and go away” says the summer half-year, May to October, is the weak one. In Shiller's monthly averages since 1871 it was the weaker half: November–April gained +6.1% on average, May–October +4.5%. Weaker is not negative, though: the summer half rose in 102 of 156 years (65%), and it beat the winter half in 5 of 16 decades.

Sitting the summer out had a price. Since October 1954, when the cash rate in the data begins, $10,000 in stocks from November to April and in cash from May to October grew to $11.1M; held all year, $18.6M. The folklore kept 60% of it.

Average return of each half-year since 1871, dividends reinvested
+6.1%
November–Aprilrose in 72% of years (112 of 155)
+4.5%
May–Octoberrose in 65% of years (102 of 156)

Monthly averages: the winter half runs from the October average to the April average, the summer half from April to October.

November–AprilMay–October
Average return of each half-year by decade (the decade of the half's last month), dividends reinvested; monthly averages. Data: Robert J. Shiller, Yale University

The January effect

In this data the folklore holds up on paper: January had the highest average of the twelve months. January averaged +1.84% against +0.74% for the other eleven months, and rose in 72% of years. Its lead has faded: before 1990, +2.08% against +0.69%; since, +1.08% against +0.92%. Two cautions: monthly averages blur single days, and January's figure, from the December average to the January average, takes in the second half of December. The lowest average belongs to October: +0.03%.

November–AprilMay–October
Average return of each calendar month since 1871, dividends reinvested; monthly averages. Data: Robert J. Shiller, Yale University
The half-years by decade and the twelve months as tables
Average half-year, by decade
DecadeNovember–AprilMay–October
1870s+5.4%+2.4%
1880s+3.8%+3.4%
1890s+3.4%+3.4%
1900s+8.7%+2.4%
1910s+1.9%+3.9%
1920s+8.6%+9.4%
1930s−3.3%+6.6%
1940s+3.0%+6.3%
1950s+11.3%+7.9%
1960s+8.9%+0.6%
1970s+6.4%−0.2%
1980s+9.7%+8.3%
1990s+12.0%+5.3%
2000s+2.1%−0.5%
2010s+9.8%+3.2%
2020s+6.3%+10.7%
Average month since 1871
MonthAverageRose in, of years
January+1.84%72%
February+0.73%55%
March+0.46%62%
April+1.03%67%
May+0.56%59%
June+0.50%59%
July+1.26%70%
August+1.24%64%
September+0.57%62%
October+0.03%58%
November+1.02%65%
December+0.74%65%

Money illusion: raises that were pay cuts

Irving Fisher gave it a name in 1928: people think in dollars, not in what dollars buy. Average hourly pay of US production and nonsupervisory workers — about four in five private employees — rose in dollars in every one of the 61 years from 1965 to 2025. After inflation it fell in 25 of those 61 years: raises that were pay cuts.

The worst was 1980: pay rose 8.0%, prices 13.5%, so buying power changed −4.9%. The longest run of such years lasted 9 years, 1987–1995. Over the whole span the hourly wage grew ×12.4 in dollars, from $2.53 to $31.34 (yearly averages), but only ×1.19 in what it buys. The buying power of 1972's pay was not seen again until 2020, 48 years later. (In 2020 the average also jumped because many low-paid workers lost their jobs that spring.)

The same illusion makes idle cash feel safe: $10,000 left in an account paying nothing since January 1964 still shows $10,000, but buys what $920 bought then. It is not only history: in the 12 months to September 2026, pay rose +3.4% and prices +3.7%, so buying power changed −0.3%.

Receipt · 1980

  • Your raise+8.0%
  • Prices+13.5%

What you got−4.9%

Receipt · 1964–2025

  • Hourly pay×12.4
  • Prices×10.4

What it buys×1.19

Receipt · since Jan 1964

  • Put in$10,000
  • Balance, Sep 2026$10,000

Worth, in money of then$920

Pay raise, in dollarsPrices (CPI)Raise below inflation
Yearly averages, change from the year before; production and nonsupervisory employees, private sector. Some months of 2025 have no published CPI; those years average the published months. Pay and prices: U.S. Bureau of Labor Statistics, retrieved from FRED, Federal Reserve Bank of St. Louis
Every year as a table
YearPay raisePricesAfter inflation
1965+3.7%+1.6%+2.1%
1966+4.0%+3.0%+1.0%
1967+4.4%+2.8%+1.6%
1968+5.9%+4.2%+1.6%
1969+6.4%+5.4%+0.9%
1970+5.8%+5.9%−0.1%raise below inflation
1971+6.5%+4.2%+2.2%
1972+7.5%+3.3%+4.1%
1973+6.1%+6.3%−0.1%raise below inflation
1974+7.2%+11.0%−3.4%raise below inflation
1975+6.8%+9.1%−2.2%raise below inflation
1976+6.8%+5.8%+0.9%
1977+7.5%+6.5%+0.9%
1978+8.1%+7.6%+0.4%
1979+7.9%+11.3%−3.0%raise below inflation
1980+8.0%+13.5%−4.9%raise below inflation
1981+8.6%+10.4%−1.6%raise below inflation
1982+5.8%+6.2%−0.3%raise below inflation
1983+4.2%+3.2%+1.0%
1984+3.5%+4.4%−0.8%raise below inflation
1985+2.9%+3.5%−0.6%raise below inflation
1986+2.2%+1.9%+0.2%
1987+2.4%+3.6%−1.1%raise below inflation
1988+3.2%+4.1%−0.8%raise below inflation
1989+3.9%+4.8%−0.8%raise below inflation
1990+4.0%+5.4%−1.3%raise below inflation
1991+3.1%+4.2%−1.1%raise below inflation
1992+2.4%+3.0%−0.6%raise below inflation
1993+2.6%+3.0%−0.4%raise below inflation
1994+2.6%+2.6%0.0%raise below inflation
1995+2.8%+2.8%0.0%raise below inflation
1996+3.4%+2.9%+0.4%
1997+3.8%+2.3%+1.5%
1998+4.0%+1.5%+2.4%
1999+3.7%+2.2%+1.5%
2000+3.9%+3.4%+0.5%
2001+3.8%+2.8%+0.9%
2002+2.9%+1.6%+1.3%
2003+2.7%+2.3%+0.4%
2004+2.1%+2.7%−0.5%raise below inflation
2005+2.8%+3.4%−0.6%raise below inflation
2006+3.9%+3.2%+0.6%
2007+4.0%+2.9%+1.1%
2008+3.8%+3.8%0.0%raise below inflation
2009+3.0%−0.3%+3.3%
2010+2.4%+1.6%+0.7%
2011+2.0%+3.1%−1.1%raise below inflation
2012+1.5%+2.1%−0.6%raise below inflation
2013+2.1%+1.5%+0.6%
2014+2.3%+1.6%+0.7%
2015+2.1%+0.1%+2.0%
2016+2.4%+1.3%+1.2%
2017+2.3%+2.1%+0.2%
2018+3.0%+2.4%+0.5%
2019+3.5%+1.8%+1.7%
2020+5.0%+1.3%+3.7%
2021+5.0%+4.7%+0.3%
2022+6.4%+8.0%−1.5%raise below inflation
2023+4.9%+4.1%+0.7%
2024+4.2%+3.0%+1.2%
2025+4.0%+2.6%+1.4%

The lottery, by its own rules

A Powerball play costs $2: five white balls from 1 to 69 and one red Powerball from 1 to 26. That makes 292,201,338 possible tickets and one jackpot among them: 1 in 292,201,338. With one ticket in each of the 3 weekly draws, the jackpot would come on average once in 1.87M years. Buying every combination once would cost $584M.

Prospect theory has a name for the pull: people overweight tiny chances, so odds of 1 in 292,201,338 can feel like “maybe”. The fixed prizes below the jackpot return on average $0.32 per $2 ticket (16% of the price), and a single ticket wins nothing at all 96% of the time. Everything else a ticket is worth has to come from the jackpot — which is taxed, and shared with every other winning ticket.

One jackpot among all possible tickets, drawn to scale
  1. 293 squares of a million tickets, one marked

    292,201,338 possible tickets: 293 squares of a million (the last one only partly full). One square holds the jackpot.

  2. A million tickets as 10,000 cells of 100, one cell marked

    Zoom into that square: 10,000 cells of 100 tickets each.

  3. 100 tickets, one winning

    Zoom into that cell: 100 tickets. One wins.

As pixels, the possible tickets would fill 141 Full HD screens; the jackpot is one pixel on one of them.

Break-even, from the published odds and prizes: for a $2 ticket to be worth $2, the jackpot's lump sum would have to reach $491M before tax, with nobody to share it. After the top federal income-tax rate of 37%, $803M. With 100M other tickets in the draw the chance of sharing lifts it to $949M; with 300M, $1.29B. Advertised jackpots are the 30-payment annuity total, larger than the lump sum these figures use.

Mega Millions costs $5 and multiplies every smaller prize by 2 to 10 at random, so those prizes return more — $1.12 per ticket, 22% of the price — but its odds are 1 in 290,472,336, and its break-even lump sum is $1.13B before tax, $1.84B after.

Before tax, not sharedAfter 37% tax, not sharedAfter tax, 100M other ticketsAfter tax, 300M other tickets
Expected value of one $2 Powerball ticket by the jackpot's lump sum; dashed: the ticket's price. Prizes from $10,000 up taxed at 37%; tickets picked at random. Rules and prizes: Powerball, Mega Millions (checked November 9, 2026); figures calculated by BigCapital
Two games, one arithmetic
GameTicketJackpot oddsSmaller prizes, per ticketLump sum at which a ticket is worth its price
Before taxAfter 37% tax+ 100M other tickets+ 300M other tickets
Powerball$21 in 292,201,338$0.3216%$491M$803M$949M$1.29B
Mega Millions$51 in 290,472,336$1.1222%$1.13B$1.84B$2.17B$2.94B

Lump sums, before state taxes. Sharing: with N other random tickets in a draw, a winner's expected share is (1 − e−λ) ÷ λ, where λ = N ÷ the number of combinations: 85% with 100M, 63% with 300M for Powerball. Numbers people pick themselves cluster on birthdays and patterns, which makes sharing likelier still. California pays the smaller prizes pari-mutuel.

Questions people ask

What happens if you miss the best days in the stock market?

In Shiller's monthly data since 1871, $10,000 in US stocks grew to $11.9B with dividends reinvested; without the 10 best months it ended at $2.21B, −81%. The best months came in rough times: 7 of the 10 best began with stocks 20% or more below their record. This article measures months, not days.

Does “Sell in May and go away” work?

Since 1871, the May–October half gained +4.5% on average against +6.1% for November–April, and still rose in 65% of years. Run as a rule since 1954, with cash from May to October, $10,000 became $11.1M against $18.6M held all year.

What is money illusion?

Thinking in dollars rather than in what dollars buy — Irving Fisher's term from 1928. US average hourly pay rose in dollars in every one of the 61 years from 1965 to 2025, yet after inflation it fell in 25 of them. Cash shows it too: $10,000 kept at 0% since January 1964 now buys what $920 did then.

Is a lottery ticket ever worth its price?

On expected value, only at enormous jackpots. A $2 Powerball ticket returns $0.32 on average in smaller prizes; to be worth $2 the jackpot's lump sum must pass $491M before tax, or $803M after the 37% federal rate, if nobody shares it. Expected value is an average over many draws: a single ticket wins nothing 96% of the time.

What happened to people who sold at the bottom of a crash?

In the 13 crashes since 1871 that ended with a new record, selling in the worst month and coming back 12 months later left the seller at −9.2% to −57.4% against holding, a year after the bottom. Done at every one of them since 1871, it left 1.6% of the buy-and-hold result.

Go deeper

Stock market crashesEvery fall of 20% since 1871, how deep and how long the way back Average stock market returnEvery year since 1871 as a brick, and 1 to 30 years of holding Time MachineWhat a sum invested in any month became Inflation calculatorWhat a dollar of any month since 1871 is worth today

How the numbers are calculated

  • Stocks: Robert J. Shiller, Online Data (Yale University): the S&P Composite (the S&P 500 and, before 1957, its predecessors) with dividends reinvested every month, monthly since January 1871. Prices are monthly averages of daily closes, so single months and half-years look calmer than in close-to-close figures.
  • Best and worst months: each month's return from one monthly average to the next, with dividends; a missed month earns 0%. “Began 20% or more below a record”: the price at the end of the month before stood that far under its highest monthly average so far. “Within a year after”: up to 12 months after one of the 10 worst months.
  • Selling at the bottom: the falls of 20% or more from a record monthly average (the rule of the S&P 500 page), counted once the record was regained, so the bottom is known. The seller sells at the average of the bottom month, holds cash and buys back at the average 12 (or 24) months later; once per fall, no taxes or costs.
  • Cash: the effective federal funds rate compounded every calendar day (actual/360) and averaged by month, like the other series. The data have no rate before July 1954; cash earns 0% before then.
  • Chasing: calendar years from December average to December average; the first full year with all four assets is 1961, so the first pick is for 1962. Stocks with dividends (Shiller); 10-year Treasuries with coupons reinvested, from Shiller's long-term rates (before 1953 one rate a year, interpolated, so bond swings are understated); gold from 1960 (one price a year in 1960–1967); cash at the fed funds rate. The equal mix is reset to a quarter each every December. No taxes or trading costs.
  • Gold: World Bank Commodity Price Data (The Pink Sheet), monthly average price in US dollars, CC BY 4.0. Source: World Bank.
  • Calendar: half-years run from the October average to the April average and from April to October; a calendar month's average is the simple mean of that month's return over all years. Sell in May as a rule starts in October 1954, the first October with a cash rate.
  • Pay and prices: average hourly earnings of production and nonsupervisory employees, private sector (BLS, series AHETPI, seasonally adjusted) and the consumer price index for all urban consumers (CPI-U, CPIAUCSL), retrieved from FRED. Yearly averages; a raise is the change from the year before; “after inflation” divides one by the other.
  • Lottery: Powerball and Mega Millions rules and prize tables as published (checked November 9, 2026). Odds are computed from the number of balls and match the published ones. Mega Millions multipliers: 2× in 15 of 32 tickets, 3× in 10, 4× in 4, 5× in 2, 10× in 1 (the published 1 in 2.13, 3.2, 8, 16 and 32). After tax: 37% on the jackpot and on prizes of $10,000 or more; state taxes and Powerball's Power Play left out.
  • Shiller's notes on the latest months: Oct price is Oct 2nd close; Sept/Oct CPI estimated; Oct GS10 is Oct 2nd value.
  • These are rules run on past data and arithmetic on published odds, not forecasts. Nothing here is investment advice.