A Look at Multi-Day Runs in the Stock Market

It has felt like, at least to me, that the last half year or so has brought an increase in what I’ll call multi-day runs in the stock market. What I mean is long grinding moves of either consecutive up or down days without a break. These type of relentless assaults can wreak havoc on very short-term mean-reversion strategies such as extreme RSI(2) or adaptive daily-follow through.

In this report I want to look at these multi-day runs and whether they are becoming (a) more frequent, (b) stronger, or (c) longer/shorter in length.

Question #1: Are Multi-day Runs Becoming More Frequent? (Click to enlarge)

I’m defining a multi-day run as any move of two days or more in a single direction on the S&P 500. I’ve looked at longer duration moves as well (such as a minimum of 3 days or 4 days in a single direction), but the conclusions are more or less the same.

The graph above shows the rolling 1-year number of occurrences of multi-day runs, both up and down, from 1955, and shows that no, multi-day runs are not any more common today than they’ve been over the last 50+ years.

Question #2: Are Multi-day Runs Becoming Stronger? (Click to enlarge)

The graph above shows the rolling 5-year average percentage change of multi-day runs up (green) and down (red) from 1955. Yes, one could say that there has been an increase in the strength of multi-day runs recently (i.e. the red and green lines flare out at right side of graph) however, that would only tell part of the story.

The next graph shows the same data in blue, but here we’re looking at the average absolute % change of all runs (i.e. we removed the +/- sign) and comparing it to the standard deviation of all daily changes (multi-day run or not) in red. Click to enlarge:

Note how the two lines track each other almost perfectly. Clearly, if multi-day runs are becoming stronger, it is simply a result of an increase in broader market volatility.

Question #3: Are Multi-day Runs Becoming Longer/Shorter? (Click to enlarge:)

The graph above shows the rolling 5-year average length (in days) of all multi-day runs (blue), runs up (green), and runs down (red) from 1955.

Although the observation isn’t crisp, and it’s difficult to draw a strong conclusion, it does appear that multi-day runs have become shorter over the last 30+ years, from a high of 3.5 days in the early 1970’s, to about 2.9 days today.

Putting it All Together

Putting all of our observations together we get:

Multi-day runs are not any more frequent today than they have been over the last 50+ years. They have become stronger more recently, but this is simply a result of an increase in overall market volatility. They have, possibly, become shorter in length over the last 30+ years, meaning changes in price have become a bit more violent (because they are squeezed into a shorter amount of time).

Not what I expected to find, but I call em’ like I see em’…

Comments

[abetterplace:]

GRAPHS: !*@#&^<>? Too much college professor orientation. Everyone has a graph on just about everything and still virtually no one is getting it right, especially me.
I wish someone would make a graph on how fast we can get our pants on in the morning compared to 50 years ago.
If we can't get it right at the beginning,well, I think maybe the rest of the day might be screwed too.We all may need a new perspective.

Apr 17 07:29 AM

[BigJake:]

If you are complaining about the graphs on this site perhaps its time to hang up the trader's suit and become and art teacher....

Apr 17 10:15 AM

[eddybaby:]

Interesting stuff.
One thing that would be useful to know data on the "size" of the runs. For example, does it really make any difference if the MDR is only say 100pts, but if they are say 800 at a go then it has serious implications.

Apr 17 10:20 AM

[Friar Hilarius:]

Michael

I like your research

It would be helpful to apply a smoothed median to the data

The last chart would benefit from a trend line bisecting the highs and lows (I have a specialised median which would achieve this, but I need the raw data in Excel format)

I like your analytical approach

By the way ... what were you expecting and why?

With Best Wishes

Hilarius

Apr 17 10:38 PM

[kotika98:]

<b> They have, possibly, become shorter in length over the last 30+ years, meaning changes in price have become a bit more violent (because they are squeezed into a shorter amount of time).</b>

That is not the reason - the true reason is that mean-reversion has become stronger. In fact, if you look closely on the graph comparing runs with volatility, you can see that multiday runs magnitude has been lagging - the ratio of volatility to average absolute mutliday run return is at a 50 year low! And what a low - 50% lower than usual!!!
I had come to this conclusion based on my own research, but its interesting to see that truth can be seen from many different angles.

May 14 07:30 AM