Saturday, July 27, 2019

Published 12:02 PM by with 0 comment

Why Does It Seem like Stocks Are at an All-Time High?

It often feels like stocks are at an all-time high, but there's an obvious reason for this...stocks are often near their all-time high.

It's so obvious in retrospect, but it didn't click for me fully until this week. US stocks have tended to rise over time. When something is rising and bounces around a bit in value, it will still generally be close to its all-time high value.

To test out how close, I took S&P 500 data since 1959. 'Near all-time high' is not an exact measure, so I took a few different definitions. If you define it as 'at least 90% of the all-time high', then ~60% of days since January 1, 1959 met this criteria. To calculate this, I'm simply comparing the value at each date with the highest value at or before that date. The table below has more of these values:

% of all-time high% of days at this level
>9919
>9545
>9060
>7588

88% of the time since 1959, the S&P 500 has been within 25% of its highest all-time value at that time.

To see it graphically, I put together this plot. The pink periods are those in which the S&P 500 was at least 90% of its all-time high at that point.


There is a potential gap here...why start at 1959? What if there was a giant crash right before this point that biased everything and messed up my boundary condition? It's hard to answer. The 'S&P 500' didn't officially exist until March 4, 1957. However, variants started in 1923. Further, I can only find free data sets that have daily values going back to 1950. Finally, does the market of 1940 really represent the market of 2010? How far back is reasonable to go? I thought '60 years' is a nice number so I went with it. It's nice also since the first trading day in 1959 was an all-time high so the 'what if it crashed right before' effect is theoretically minimized. It is worth noting that this avoids the great depression. These numbers would be lower if I had a full data set back to that period.

To provide another arbitrary period though, here's the last 30 years:

% of all-time high% of days at this level
>9921
>9549
>9062
7583

That's it. Just thought it was a cool way to capture this info and it made it more intuitive for me.


      edit

0 comments:

Post a Comment