Bitcoin Forum

Whilst working on my short-term time series analysis, I tried out fitting the price to a yearly cycle. The data is too artificial though, and hindcasting is impossible with the short timeseries involved.

I have tried to document all examples of artificiality I could find in the chart below; however, there are bound to be more. Bitcoin's history is simply too short for any significant seasonal cycle to emerge.

https://i.imgur.com/wygGO93.png

Seems to be more based on the day of the week than seasonal or end/start of month from what i've observed over the years. I'd say sunday/monday is usually when it's hitting the bottom of a skid then climbs up during the week and peaks on fri/sat before it dips. Of course this isn't always the case but at least its the trend i've seen. I'm too lazy to link a good chart for price/momentum but when you look at transactions per day it gives a pretty telling picture:
https://blockchain.info/charts/n-transactions?timespan=180days&showDataPoints=true&daysAverageString=1&show_header=true&scale=0&address=

Seems theres always 25%-100% less transactions on sunday/monday than compared to its high point in the respective week.

Bump.

Again, no clear sign of any seasonal cycle.
https://i.imgur.com/p9meLDN.png

1) For short terms, there is only short circles that changes time to time. For exemple, sometimes weekends have been dump time for some weeks, sometimes pump time for some days, and sometimes a bigg sell-off used to happens at same time od a day for several day, but it changes from time to time and there is no long term pattern to predict what short term cycle we are in and what and when will be the next


2)Only long term pattern I see is a exponential grow after enough momentum is reached, coming from good news and people putting big money in, followed by correction after the peak(like all other investiments). But there is no pattern about when the spikes may happen, only that they happens after some time of stable price. In 2012(or 2011?) took about 18 months for a big spike in price happens, so there is no pattern.

Here's the chart done by weekday. Aside from the well-documented "Saturday dip", there is no perceptible trend.

https://i.imgur.com/bkOvY4L.png

Edit: Putting a bit more thought into it... We can generate this kind of time series analysis by comparing to a trended Fourier wave. I'm thinking the logarithm of Bitcoin price could be modelled as:

• Trend (ax term).
• Full-period sinusoidal.
• Half-period sinusoidal.
• Third-period sinusoidal.
• Etc.

Someone better at mathematics than I will probably know of a way to generate those sinusoidals. From them, we can determine which time period has the most pronounced cycles.

There are a lot of different cycles that could be analyzed, seasons are only one of many.

Here is a book containing the most useful cycles:

Gann Made Easy (pdf) (http://www.fxf1.com/english-books/Forex%20Systems%20Collection/Gann-1/McLaren,%20William%20-%20Gann%20Made%20Easy.pdf)
Scribd (http://www.scribd.com/collections/2470912/Gann-Made-Easy)

Contrary to most other actives, bitcoin trades also on weekends.
However, banks are closed on weekends, so it's hard for new fiat to enter the market.
I think this is the explanation for this weekly cycles.

I have considered the seasonal cycle theory as well. I think it has to do with temperature drops and decrease of sunlight in North America. These are known to affect mood and people have experienced a lot more depression during the winter months. This leads to less risky behaviour. You see a peak in the summer corresponding with hope and overcoming winter depression which results in a greater propensity for purchase.

With a seasonal cycle, why would Bitcoin have to adhere to North America. North America isn't the center of the world you know :)

Bitcoin has investors from all over the globe try factoring in
 - Percentage of active people on the other hemisphere like Australia where summer is winter and visa versa.
 - Percentage of active people following different year cycles like the Chinese years including their meaning.

Maybe Bitcoins seasonal cycles are not yet noticed because they are buried under the hypes and bad news. We might need stability before we can see seasonal cycles.

"Less risky" in the winter? Then why would the price shoot up like it did? That seems pretty risky to me, :p.

It seems to me that the dip is shifted a bit earlier, indicating that at least part of the reason there is a "weekend dip" is the expectation that there will be a "weekend dip".

it shot up in the winter because the old miners make great heaters  ;D
(which leads their interest in the coin itself I presume)
https://i.imgur.com/1srrQRV.jpg

Seems to me that we don't have nearly enough data for any correlations drawn here to hold much weight. Any thoughts on statistical significance? Is 5 winters enough? :P

That image is awesome. I actually loved having my system up in the winter to heat up the room; I never had to use my real heater!

The only reason I'm still mining is this year's late spring. We actually had a frost on the May Two-four weekend north of the city this year. Global warming is a misnomer.

Keep those satoshis rollin' in.  :)

It is pretty nice being able to earn free BTC while heating up the house too. Sure beats burning electricity/gas heating up the house and getting nothing from it but warmth.

here is heat as hell by nature. Wish I was a mad scientist, so I could make a cooling system using bitcoin miners

Here are the steps:

1) Buy tons of miners
2) Move to where it is always winter (keep moving around the world or simply move to a place like Antarctica)
3) No cooling needed!