Lesson 8.8 · about 15 minutes
The crash simulator
- Long history
Learn to
- See how hypothetical mixes behaved in the falls the data holds
- Measure a mix's deepest fall and how long it took to get back
- Face a moment at the bottom and see what each reaction would have meant
Ideas in this lesson
- Max drawdown
- Time to recover
- SIP through a fall versus a lump sum
- Panic costs
- Hindsight
A maximum drawdown is the deepest fall from a high to a later low. For a mix it depends on how much it holds of each class and on how those classes moved during the fall. The simulator replays real falls of the Nifty 50 price index with a mix you set, and shows the value path, the deepest fall and how long the mix took to regain its high.
Our dataset starts on 1 January 2016, so the 2008 fall is not in any chart here. The 2008 fall and the 2020 fall are in the register: about 57 to 60% from the January 2008 peak, as reported; 38.4% from 14 January to 23 March 2020 on closing levels in our data†. The 2020 figure is recomputed in this lesson from closing levels.
The real series here are daily closes from 1 January 2016 to 31 December 2025: the Nifty 50 price index (dividends are not included), NSE's Nifty 10 yr Benchmark G-Sec Index (a total return index for the 10-year government bond) and the price of an NSE-listed gold exchange-traded fund. Cash is an invented steady 4% a year, because we have no published cash series. All of it is past, not forecast.
At the bottom the simulator pauses and asks what you would do. Stay with the mix, add a tenth more, or move everything to cash. It then shows what each would have meant a year later and at the end of the fall. No answer is marked correct. Hindsight makes every one of these look obvious, and at the bottom nobody knows it is the bottom.
Try it · past data to 31 December 2025; cash is invented
Pick a fall and set a mix. Watch the value path, then answer the pause at the bottom.
Loading the long-history data…
For the starting mix, staying with it or adding ended ahead of moving to cash in every one of the falls in this data, because each was followed by a rise from its low. In a fall that kept going, the order could be different. That is why the pause is framed as hindsight, and why nothing here says which reaction is right.
A mix that fell less does not always recover sooner. In the 2021 to 2022 fall, a mix with no equity (40% bonds, 40% gold, 20% cash) fell 3.1% and took 153 days to regain its high from its low, while a mix with 30% equity, 60% bonds and 10% cash fell 7.3% and took 89 days. Time to recover depends on how fast the holdings that fell grow back, not only on how far they fell. Set those two mixes yourself to see it.
The second panel compares the same total money put in two ways: all at the equity high, or in equal monthly amounts beginning at the high. A fall flatters the monthly amounts because later amounts are invested at lower prices; a rise would flatter the lump sum. It is arithmetic about one path of history, not a plan.
Check yourself
0 of 3 right
Answer every question correctly to complete this lesson.
Facts and sources
Figures in this lesson, with their dates
Nifty 50 falls in 2008 and 2020: about 57 to 60% from the January 2008 peak, as reported; 38.4% from 14 January to 23 March 2020 on closing levels in our data
NISM reports that the Nifty 50 fell about 57 to 60% from its January 2008 peak (about 6,357), depending on the trough date. For early 2020 the same reporting gives about 38 to 39% between intraday levels of 12,430 and 7,511, but those two levels compute to 39.6%, so the reported range is approximate. Recomputed from closing levels in the Learn long-history dataset, the 2020 fall is 38.4%: from a close of 12,362.30 on 14 January 2020 to a close of 7,610.25 on 23 March 2020.
As of 23 Mar 2020 · last checked 30 Oct 2026 · next review 30 Apr 2027 · Reported by secondary sources, to verify · NISM (2008 and the intraday 2020 levels); closing levels from our own long-history dataset
Still open: Only the 2008 figure and the intraday 2020 levels come from the secondary source (reported, R). The 2020 close-to-close fall is recomputed in npm run portfolio:check from the dataset, and the 2008 fall lies before the dataset's first day (1 January 2016), so it has not been recomputed.
† To verify: reported by secondary sources and not yet checked against the primary document. A fact reviewer clears these before public launch.
Sources
Last reviewed 30 Oct 2026 by the author. Independent fact review is pending before public launch. Facts are as of the dates shown. Education only; not investment advice.