A 38-year-old with a large sum in cash cannot bring himself to invest “at current valuations.” A reader with most of a large portfolio parked in a money-market fund is waiting for “at least −10 or −20%.” A UK saver has left money in an ISA for months waiting for the crash everyone keeps predicting; a German one keeps a growing cash reserve “for the correction that must come.” The replies are always the same two: a study showing that investing at once beats drip-feeding about two times in three, and “you will still find an excuse not to go in.” Both true, neither an answer to the person's actual fear, which has two halves: what if I invest and it crashes, and what if I wait and it never does.
This story replays both halves with the same money. Take 100,000 arriving on the day of a historical peak — deliberately the worst day to invest, because that is the day the person is afraid of. Either invest it all at once, or hold it in savings at 3% and invest only after the index has fallen 15% from that level. Follow both for 36 months. Then run the same two choices through the future where the fall never comes and the market grows 7% a year.
Waiting for a crash is a bet with two outcomes and one price. The forums quote the odds. They never quote the price.
100,000 after three years, at the 2007 peak and in the no-crash future
Price index, no dividends, no taxes. The peak day is chosen with hindsight; the person in the thread does not know whether today is one. The no-crash future is an assumption (7% a year), not a forecast.
In the crash future, waiting wins. That is the whole reason the question exists.
Every row of the table says the same thing about the crash future: buying after a 15% fall beat buying at the peak. At the 2007 peak, 90k against 75k after three years. At the Nasdaq's 2000 peak, 33k against 25k — both terrible, one less so. In 2020 and 2022 the fall arrived within months and the waiter bought it. This is not a discovery. It is arithmetic: if the fall comes, the person who waited for it paid less.
The row that matters is the last one, and it is the same in every country and every year because it does not depend on history at all. In the no-crash future the wait rule never triggers. After three years the money that went in at once is worth 123k; the money that waited is worth 109k, all of it still in savings, still waiting. The cost of the bet is 13k per 100,000 over three years — and it keeps growing every year the crash does not come, which is what happened to the April 2025 sellers and to everyone who has been holding cash “for the correction” since 2023.
| Peak | Invest now, crash future | Wait rule triggers | Wait, crash future | Invest now, no-crash future | Wait, no-crash future | Peak investor after 5 yrs | after 10 yrs |
|---|---|---|---|---|---|---|---|
| 2000 · Dot-com bubble United States · Nasdaq Composite | 25k | month 1 | 33k | 123k | 109k | -59.3% | -53.2% |
| 2007 · Global financial crisis United States · S&P 500 | 75k | month 5 | 90k | 123k | 109k | -8.2% | N/M |
| 2020 · COVID shock United States · S&P 500 | 119k | month 1 | 175k | 123k | 109k | N/M | N/M |
| 2022 · Inflation & rate-hike bear market United States · S&P 500 | 108k | month 6 | 136k | 123k | 109k | N/M | N/M |
| 1999 · UK millennium peak United Kingdom · FTSE 100 | 57k | month 15 | 73k | 123k | 109k | -30.5% | -22.1% |
| 2000 · French dot-com bubble France · CAC 40 | 49k | month 4 | 60k | 123k | 109k | -35.7% | -47.3% |
The peak-day investor, five and ten years on
The threads also ask a darker version: what if I buy at the top of a bubble? The last two columns answer it for the peaks where the cycle lasted long enough. Money invested at the Nasdaq's March 2000 peak was worth -59.3% of its starting value five years later and -53.2% ten years later, price only. At the FTSE 100's December 1999 peak, -30.5% and -22.1%. At the CAC 40's September 2000 peak, -35.7% and -47.3% — the French index was further from its peak after ten years than after five, because 2008 arrived before 2000 was repaired. The 2007 investor was at -8.2% after five years and back above the peak before ten.
Those are the true worst cases in this dataset for a single day's decision, and they are price-only: the FTSE 100 paid dividends of three to four percent a year through that decade, which the price index ignores entirely. They are also the cases the forum reader is not in. The person asking has a lump sum on an ordinary day, not on the day that hindsight will later label a peak. Investing at every all-time high in history, rather than at the six selected here, has a much better record — which is the statistic the replies quote, and which is true, and which does not remove the fear.
Price the two mistakes, then choose a rule
The useful move is not to guess which future arrives. It is to write down, before looking, what each mistake would cost you and which one you could live with. Invest now and watch a 2007 happen: down to 75k at three years, in price terms, on the way back. Wait and watch nothing happen: 13k behind after three years, further every year, with the same decision still in front of you every morning. Some people can carry the first and not the second; some the reverse. A split — part now, part on a schedule, part on a rule — is a way of carrying some of each, and the calculator lets you set the shares.
What does not work is the thing the threads describe: waiting for “a good entry” with no number attached. That is not a rule, it is a mood, and the table shows what a mood costs when the market does what it usually does.
What this does not prove
- Starting on the exact peak day is the worst case for “invest now,” chosen deliberately because it is the day the person fears. It is not the typical outcome of investing at a record high.
- Price indices only. Dividends would raise every “invest now” figure and the peak-investor columns, most of all for the FTSE 100.
- The wait rule buys at the month-end level after the trigger, an approximation of execution. A real order fills at a different price.
- 7% a year for the no-crash future and 3% for savings are assumptions; the comparison between the two choices in that future depends on the gap between them.
- Six selected peaks are six shapes, not a distribution. The UK and French rows are those countries' own indices, not a global fund in pounds or euros.
- Nothing here recommends investing now, waiting, or splitting.
Method and sources
Each episode's normalized daily-close path (weekly-sampled with exact anchors for long cycles) is evaluated at month ends. The peak-day investor's value after five and ten years is read from the same path. Intraday highs and lows are excluded. Research snapshot cutoff: 14 September 2026.
- Nasdaq Composite historical data — the 2000 peak.
- S&P 500 historical data — the 2007, 2020 and 2022 peaks.
- FTSE 100 historical data — the December 1999 peak, dividends excluded.
- CAC 40 historical data — the September 2000 peak.
- Companion story: the 2025 sellers and the cost of waiting since.
- Markets Panic methodology, definitions, caveats and reproducibility notes.