Timing Market Dips
· 2 min readWith all the Coronavirus panic the DOW has been falling off a cliff the past few days:
Which raises a valid question: if you're sitting on some cash should you buy a index fund like the DOW?
Let's backtest
This isn't the first time the DOW has dipped more than 5%. It's happened 18 times before 2020:
| Occurrence | Date | Dip % |
|---|---|---|
| 1 | 1987-10-19 | 22.61 |
| 2 | 1987-10-26 | 8.03 |
| 3 | 1988-01-08 | 6.85 |
| 4 | 1989-10-13 | 6.90 |
| 5 | 1997-10-27 | 7.18 |
| 6 | 1998-08-31 | 6.36 |
| 7 | 2000-04-14 | 5.65 |
| 8 | 2001-09-17 | 7.12 |
| 9 | 2008-09-29 | 6.97 |
| 10 | 2008-10-07 | 5.10 |
| 11 | 2008-10-09 | 7.33 |
| 12 | 2008-10-15 | 7.87 |
| 13 | 2008-10-22 | 5.69 |
| 14 | 2008-11-05 | 5.04 |
| 15 | 2008-11-19 | 5.07 |
| 16 | 2008-11-20 | 5.56 |
| 17 | 2008-12-01 | 7.70 |
| 18 | 2011-08-08 | 5.54 |
| 19 | 2020-03-09 | 7.78 |
| 20 | 2020-03-11 | 5.85 |
| 21 | 2020-03-12 | 9.98 |
Buying the dip immediately
Buying the dip on the close yields the following percent gains:
| Dip Date | 1 year | 5 year | 10 year |
|---|---|---|---|
| 1987-10-19 | 24.21% | 82.56% | 356.5% |
| 1987-10-26 | 21.15% | 78.80% | 337.4% |
| 1988-01-08 | 14.80% | 72.92% | 317.4% |
| 1989-10-13 | -6.66% | 50.82% | 314.4% |
| 1997-10-27 | 16.82% | 17.91% | 90.9% |
| 1998-08-31 | 43.64% | 24.89% | 55.3% |
| 2000-04-14 | -1.42% | .95% | 6.7% |
| 2001-09-17 | -7.99% | 29.59% | 28.1% |
| 2008-09-29 | -6.01% | 47.20% | 155.0% |
| 2008-10-07 | 2.94% | 58.10% | 179.9% |
| 2008-10-09 | 14.98% | 72.23% | 208.7% |
| 2008-10-15 | 17.31% | 78.37% | 195.4% |
| 2008-10-22 | 18.33% | 80.67% | 198.6% |
| 2008-11-05 | 9.48% | 71.12% | 176.5% |
| 2008-11-19 | 29.19% | 99.76% | 217.7% |
| 2008-11-20 | 36.62% | 11.41% | 231.2% |
| 2008-12-01 | 28.50% | 97.40% | 210.9% |
Waiting for less volatility
But wait, it might be safer to sit on cash or invest in something else during periods of volitility! A decent measure of volatility is the standard deviation over a 20 day period (1 month) divided by the close price. This volatility can be as high as 10% (like during 5% dips) and as low as .5% when not much is happening. If we wait for volatility to die down <3%, this yields the following percent gains:
| Dip Date | Buy date | 1 year | 5 year | 10 year |
|---|---|---|---|---|
| 1987-10-19 | 1987-11-18 | 11.38 | 63.69 | 309.39 |
| 1987-10-26 | 1987-11-18 | 12.07 | 65.41 | 304.70 |
| 1988-01-08 | 1988-01-11 | 12.80 | 69.92 | 310.20 |
| 1989-10-13 | 1989-10-16 | -9.76 | 45.82 | 300.70 |
| 1997-10-27 | 1997-11-14 | 10.47 | 11.50 | 80.54 |
| 1998-08-31 | 1998-09-23 | 32.80 | 15.46 | 43.66 |
| 2000-04-14 | 2000-04-17 | -4.00 | -1.68 | 4.00 |
| 2001-09-17 | 2001-10-19 | -10.82 | 25.60 | 24.21 |
| 2008-09-29 | 2008-09-30 | -10.21 | 40.62 | 143.67 |
| 2008-10-07 | 2008-12-19 | 13.36 | 74.10 | 208.27 |
| 2008-10-09 | 2008-12-19 | 14.98 | 72.23 | 208.73 |
| 2008-10-15 | 2008-12-19 | 17.29 | 78.35 | 195.36 |
| 2008-10-22 | 2008-12-19 | 17.51 | 79.41 | 196.58 |
| 2008-11-05 | 2008-12-19 | 16.63 | 82.29 | 194.56 |
| 2008-11-19 | 2008-12-19 | 20.43 | 86.22 | 196.22 |
| 2008-11-20 | 2008-12-19 | 20.27 | 86.11 | 191.60 |
| 2008-12-01 | 2008-12-19 | 22.05 | 87.50 | 195.35 |
When we use this measure we wait between a few days and a few weeks to buy. Interestingly all the dates after the housing crisis in October 2008 wait to buy until 2008-12-19 when volatility dies down.
Comparison
There are 3 cases when you could have gotten a better "sale" by waiting for even lower prices than a 5% dip:
| Date | Naive price | Delayed price | Worth waiting |
|---|---|---|---|
| 1987-10-19 | 1738 | 1939 | No |
| 1987-10-26 | 1793 | 1939 | No |
| 1988-01-08 | 1911 | 1945 | No |
| 1989-10-13 | 2569 | 2657 | No |
| 1997-10-27 | 7161 | 7572 | No |
| 1998-08-31 | 7539 | 8154 | No |
| 2000-04-14 | 10305 | 10582 | No |
| 2001-09-17 | 8920 | 9204 | No |
| 2008-09-29 | 10365 | 10850 | No |
| 2008-10-07 | 9447 | 8579 | Yes |
| 2008-10-09 | 8579 | 8579 | Yes (by 8 pennies) |
| 2008-10-15 | 8577 | 8579 | No |
| 2008-10-22 | 8519 | 8579 | No |
| 2008-11-05 | 9139 | 8579 | Yes |
| 2008-11-19 | 7997 | 8579 | No |
| 2008-11-20 | 7552 | 8579 | No |
| 2008-12-01 | 8149 | 8579 | No |
But how much better off would you have been when using the Naive strategy? This is the Naive strategy's advantage over the delayed strategy:
| Date | 1 year diff | 5 year diff | 10 year diff |
|---|---|---|---|
| 1987-10-19 | 12.83 | 18.86 | 47.19 |
| 1987-10-26 | 9.07 | 13.39 | 32.76 |
| 1988-01-08 | 1.99 | 3.00 | 7.25 |
| 1989-10-13 | 3.09 | 5.00 | 13.74 |
| 1997-10-27 | 6.34 | 6.40 | 10.36 |
| 1998-08-31 | 10.83 | 9.42 | 11.72 |
| 2000-04-14 | 2.57 | 2.64 | 2.79 |
| 2001-09-17 | 2.83 | 3.99 | 3.94 |
| 2008-09-29 | 4.20 | 6.58 | 11.40 |
| 2008-10-07 | -10.41 | -15.99 | -28.32 |
| 2008-10-09 | -1.1e-5 | -1.6e-5 | -2.9e-5 |
| 2008-10-15 | .01 | .02 | .04 |
| 2008-10-22 | .82 | 1.26 | 2.08 |
| 2008-11-05 | -7.14 | -11.17 | -18.05 |
| 2008-11-19 | 8.76 | 13.54 | 21.55 |
| 2008-11-20 | 16.35 | 25.30 | 39.64 |
| 2008-12-01 | 6.44 | 9.89 | 15.58 |
As you can tell the Naive strategy wins a decent amount more than the delayed strategy. But when the delayed strategy works, it's really the better strategy to use by a 10% margin.
| Horizon | Mean naive | Mean delayed |
|---|---|---|
| 1 year | 15.05 | 11.01 |
| 5 year | 63.22 | 57.80 |
| 10 year | 193.0 | 182.8 |
As you can see, these differences are so miniscule that timing the market doesn't really matter for people looking to invest over long time horizons.
Conclusion
If you're focused on a 10+ year time horizon go ahead and take the Naive strategy of buying an index fund on the Coronavirus dip! Any reason is a good reason to start investing, and the statistics favor it in the majority of historical dips.
Python code available on Github.