Welcome back traders! I wanted to take an in-depth look at the explosion of volatility of late in comparison to previous price action. In part 1, we focused mainly on the charts and what story that tells. In part 2 here, we will be utilizing my bitmex data collection (data found here ; with a how-to guide found here) to further analyze the price movements.
When we left off several days ago, BTC was here:
Fast forward to current (Thursday the 29th) and things looking better:
We have seen a pretty nice move off lows off last 3 days, reaching about a 25% bounce at its peak. At time of writing (thurs 22:00 utc), we have seen about a 5% sell off from its 24-hour high and currently hovering around 4200 (per my composite index). Sort of just making a little range here, and we shall see which direction it goes.
As we all are aware, BTC volatility was virtually non-existent for a lengthy stretch of time before exploding several weeks ago. My Boredom Index, as discussed in part 1 has been flat-lined since we broke through the 6k floor:
I will warn you in advance, there are going to be a lot of images of numbers and spreadsheets posted below, but hopefully I can convey the story that those numbers are telling ME, to you.
First image I would like to share is a heat-mapped display of average volume and average % range for any given day+hour of the week:
This allows us to quickly see the average volume and % range for any given hour throughout the week. More specifically, if we look at the right half of the image, we can see that overall, the average % range for any given hour is 0.81%. The highest average coming in on Wednesdays at 4pm utc at 1.54%
I wanted to show that just so we can get an idea of what an ‘average’ amount of volume and price movement is in any given hour.
The image below is snapshot of my data collection taken after Tuesday 11/13 data:
The next image shows the same data, but taken after Tuesdays 11/27:
Now this one shows the difference between the 2, just 14 days apart:
The highlighted column is what really stands out to me. If we look at the total number of hours, from 11/13 through 11/27, it increased from 5448 to 5784. A 336 hour increase (14 days x 24 hours).
That represents just a 6.17% increase in total time. But if we look at the other increases, we can see a much larger jump. For example, we have seen a 44% increase in instances of an hourly candle being 2% or greater in just these last 2 weeks.
This data comprises almost 8 entire months worth of data. Every hour of every day since April 1st. To have the total number of instances of hourly ranges >2% increase 44% in just 14 days out of 8 months is simply incredible to me.
I also noticed that after 11/13, if you look at the total number of >2% instances (304) compared to total number of >0.5% instances (2841) the difference is roughly 10x. 304 is just 10.7% of 2841. (column B)
In English: for every 10 instances of >0.5% hourly ranges, we see roughly 1 instance of an hourly range >2%.
However if we look at column H (total # increase) we can see those totals as 135 and 280, respectively. The former is now 48% of the latter figure. 10.7% → 48% (Again in English: Was 1 for every 10. Over the last two weeks, we are seeing 4.8 for every 10
Another statistic in there that stands out to me is the average length of time in between seeing >2% hourly moves. On the 13th, we averaged one >2% move every 17.95 hours. Two weeks later, that is down to 13.22
The next several images will be of the raw data. The volume totals in column H are ‘heat-mapped’ so the more green/red the cell is…the more of one extreme that total falls on.
This one shows what i consider to be a ‘normal’ day. It is roughly 2 days worth of action in mid-July. In the volume column (H), we can see that there is a pretty good mixture. We have some red totals (lower than average volume), some green totals (higher), and some uncolored totals (middle). Also, if we look at column J we can see mostly smaller % ranges, with some 1's, 1.5's, and 2+ ranges sprinkled in.
Now this snapshot is taken from October, the the middle of the 68 day range, and shows same days of the week (end of Mon, all of Tues, start of Wed). This is what I consider to be a ‘boring’ day. In the volume column, there is one ever-so-slightly green row at the bottom, 3 or 4 uncolored rows, and the remaining vast majority of instances are red. Again, lower than average volume. Looking at column J, that tells an even more dull tale. In those roughly 35 hours or so that are showing, we had only one example of an hourly range greater than just 0.5%. (Remember from above, that we see, on average, a greater than 0.5% hourly range 52% of the time. 1 out of 35 is just under 3%) In the majority of hourly instance, the % range is 0.1% or less. We can also see the point ranges in some of those…. 3, 7, 8, 5, etc.
(Ironically enough, it was around this time that I started pondering the topic of a recent Medium post on over-trading found here )
So I showed you ‘normal’, i showed you ‘boring’, now lets see what these last few weeks have looked like:
This image shows most of Monday and Tuesday, Nov 19–20. In the volume column (H), almost every instance is green. A few uncolored rows, and zero red’s. Even more eye-catching to me is the % ranges (J). Nearly every instance in this image shows an hourly range greater than 2%. In fact, this image shows 38 hourly instances, and 31 of those have a greater than 2% range. That is 81.58% of instances! (As we remember from above that as of 11/13, we averaged an hourly range greater than 2% of the time just 5.58% of instances). At least for this small snippet of time, that is a 1,362% increase in the frequencies of >2% hourly ranges!
(quick note: bottom of that image shows the hour we put in our 2nd highest % range of 12.24% in my entire data set going back to April 1. The ‘Soros candle’ still reigns supreme at 17.68%)
As evident by the charts, this extreme spike in volatility has become the new ‘normal’ over the last 2 weeks.
Above shows a snippet of raw data from the 19th and 20th. The highlighted rows are instances of >2% ranges. (Again remember: we see them on average once every 17 hours and just 5.5% of instances.) We can see that we went an incredible stretch of 22 consecutive hourly ranges >2% and 31 of 32. That is just mind-blowing levels of volatility. Most of those instances aren’t ‘barely’ over 2% also… 8 of those are over 4% hourly ranges.
Quite a difference from the ‘boring’ day example shown above, where the point ranges were all single digits. Here, almost every instance is 100+ point move. (Which, of course is a larger % move at BTC 4k than at 8k)
Let’s take a closer look at my ‘2% moves’ tab:
This tab shows the longest intervals in between seeing hourly ranges >2%. (as illustrated in my ‘boredom index’ charts discussed in part 1) For most of October and first half of November, we consistently saw days and days go by without any instances of a >2% hourly range. (Remember, we average on every 14 hours or so, occurring 7% of instances). The longest we went was just over 2 weeks at 340 hours. Over the last couple of weeks we now see >2% hourly ranges more often than not. Outstanding volatility for a short-term scalper.
In part 1 we looked at the charts, in part 2 here we looked at the volatility in my bitmex data collection. In part 3 I will compare the volatility to the volume.
