Topic: My thesis: Bitcoin Calendar and Spillover effects (Read 2109 times)

PM sent.

Your page is amazing!, thanks.

Send me a PM and I can help point you to some relevant research.

Thanks you very much, I'll take a look over the data and see if this things fit statistically

Where can I get an intraday time series? I've only found daily data

Here are a couple tidbits that may help you:

I notice that many of the largest sell offs end between 7-10 PM US Eastern Time (most likely because some certain % of American traders like to hold cash overnight, especially in a downtrend) I have profited off this trend numerous times, it is very nice when you see it happening  . In price advances, we sometimes get a boost in the morning hours in the United States, and in the nov/dec 2013 bubble, possibly the morning in China as a lot of money flowed in from China.

During Bubble phases, such as march/april 2013 and nov/dec 2013, we experienced "weekend crashes". The massive price rises are caused by many people sending a lot of money to exchanges at once and buying; however since most banks are not open weekends, the inflow of cash is significantly stopped during the weekends, and the price is more susceptible to crashing. Monday morning would mean a lot more cash entering the exchanges and the price would rise rapidly. This "weekend crash" is the most pronounced in times where the price is rising fastest, and I assume it lessens over time as more methods for getting fiat into bitcoins over the weekend emerge.

In terms of price effects of rising difficulty: The difficulty rise once every 2 weeks is expected (as it has gone up every time for well over a year), so I assume this is already priced into the markets beforehand. But it is interesting to look on a longer time frame. Look at the hashrate on a log chart since 2009. You will notice the difficulty spikes during/shortly after the 2011 bubble, then levels out, until the 2013 bubbles which caused the hashrate to spike again. Also notice that litecoins' hashrate flatlined in september, and again in october, and the $ price of litecoin has stopped crashing at roughly the same time. It looks like price bubbles cause a mining equipment arms race, until most of the hashrate is owned by miners who break even or mine at a loss. Then the miners hold a larger % of the mined coins and the price stops falling.


Reasons for price bubbles

Although I have not heard many people mention this, the block reward halving in late 2012 coincides with the beginning of the uptrend that led bitcoin to $32, where it broke a new all time high. Therefore the block reward halving may have had a massive effect on the timing of the 2013 bubbles. (Since most people just assume it was because of Cyprus, this may not be fully priced in for the next halving  )

When the all time high is broken, a media frenzy starts, and thus the bubble is fueled. This means that most people are only interested in bitcoin for its price - everyone watches it when it is skyrocketing. Go look at the google trends chart for the word "bitcoin"; the 3 spikes you see go along with the three bitcoin bubbles.



Also main reasons for price falling since last november (idk if this will help):

A large % of mined coins are sold at market

People are scared of holding large amounts of $ on exchanges for any significant amount of time, after seeing so many exchanges fail such as mt gox

To a far lesser extent merchants instantly cashing out (although large purchases of $250k-$1m+ that get instantly dumped  at market will cause a larger effect).

Perhaps manipulation bots on mt gox using fake USD pushing price up in nov/dec, so prices got higher than they would have got in a "normal" bubble, so the downtrend looks bigger.


Also, view my profile, and view my 5th post down, concerning how to time the alt coin markets in the next bitcoin bubble (all run by pure speculation and adrenaline, or human nature) Essentially people profit from the bitcoin bubble, use some of the profits to speculate on alts starting with the ones with the highest market cap, and eventually all of the alts bubble (very generally the lower the market cap the higher the % increase).

yeah you're right, regarding "whales", there is documentation of large financial entities who use "window dressing" strategies (they change their investment portfolio at a certain date so when they have to submit their annual reports there are just "good investments") crashing some prices.

It doesn't really matter if it was the whales or all humanity, what matters is that it has to be recurrent, every year (month, week, etc) so maybe you can use it as a strategy, for example there is a "day of the week" effect in some stock markets where mondays have bad performance and fridays good performance (comparing with the average of the week).

With the little I've done with bitcoin data I could say that there is a "difficulty change" effect, where the volatility on the previous days is higher than the average.

On human behaviour reflecting price... "human behaviour" would mean all humans as a whole, it's pretty hard to extract reliable data in this sense when a private group of whales can crash the price if they really wanted to. So if they do it at xmas you would say, oh its xmas, people is cashing out to buy some presents, but whales started it. You cant never know.

no one 

Hello everyone

I'm Felipe a student from Chile and I'm working on my thesis to get my college degree, so I was working on Bitcoin trying to connect some points to elaborate a relevant thematic for economic research.

So far I got involved with the efficient markets theory and behavioral finances, where I want to verify the efficiency degree of bitcoin exchange markets, opening path to behavioral anomalies since humans are scarcely involved with bitcoin system operation so any anomalies on prices could be directly reflecting human behavior.

For example, calendar effects are related to the change of day, change of week, month, year, etc... bitcoin markets works continuously 24/7, so there are no gaps in the data, so making some analysis on calendar effects could be rewarding, also I want to propose a calendar effect related with the difficulty change.

Furthermore, bitcoin is a decentralized system, so I want to analyse spillover effects from other markets, here I have two hypothesis, on one hand we could have an isolated system where there is no spillover from other markets and in the other hand bitcoin could be so globalized that all spillover risks gets diversified... anyway both cases open path to making bitcoin an alternative to make hedging strategies. However there is always the possibility that bitcoin is related to a couple of economies and spillover effects are significantly strong.

What do you think about this?
Is everything I said possible?
The current data available is useful for this purposes?
For this I will need to use GARCH models, what are your experiences building this kind of models with bitcoin data?

I welcome every user with the related knowledge to make an opinion over this, it would help me a lot.
Gracias!