Topic: Bitcoin Price Forecast - Independent Variables Wanted (Read 1338 times)

I would argue that interest in bitcoin via google is more closely linked to Bitcoin being in the news and volatility in price then it is the actual price. 

The price of bitcoin is driven by news and other underlying factors. There have been many times this year when the price of bitcoin went up after bit merchants started accepting bitcoin (the price went up after when both overstock and dish announced they would start to accept bitcoin).

The price of bitcoin had fallen dramatically when the MtGox fiaso came about, and when the IRS announced it ruling on bitcoin being property instead of currency.

I admit that some people buy and hold (and possibly sell) bitcoin based on their faith in bitcoin, however they are becoming a smaller minority as bitcoin matures.

Google Trends could be one way way of picking up quantifiable data. The graph of search term interest for 'bitcoin' closely matches price, although I would guess that it is always lagging price movements and probably not useful as a 'forecast' tool like you want.

Anyway, I'm sure someone already has a news aggregate program out there that uses natural language text analysis to determine the positivity/negativity of articles. That could be useful, just got to find it or code it yourself. You could also maybe scrape things like these forums, especially the count of or thread popularity. Count the 'HODL' vs 'Get out now my brothers, I warned you, blah blah' posts.

It is hard to quantify something so new. You could do technical analysis based on past performance. Not sure how to quantify the other side without and incredible amount of daily research.

do an inverse of economic variables as per health of our economy today. Inverse of GDP will be bullish BTC for example..

however if you can come up with a valid model you might swell just trade forex or something else... since your model would be able to forecast based on variables that worked. I suspect you will not find this.

Bitcoin isn't driven by news, it is driven by faith
People start buying in it goes up
People start dumping it goes down
If people have faith in Bitcoin and eventually everyone does it will stabilise around a certain area
I truly believe it will hit into the mid-low four figures by mid 2015

Over the medium term, the price of Bitcoin will be based on how many companies accept Bitcoin for payment.

Similarly the level of consumer adoption of Bitcoin will also drive the price, however this variable is directly related to above.

As mentioned by previous posts the price is driven primarily by speculation.

Well, that's why I made the thread to ask this question. I think a lot is known. For example, in my opinion, a small part of the reason the value is rising over time is due to Bitcoin's scarcity and alternative currencies supply expansion. So, not only is supply of Bitcoin known, but perhaps there's a proxy for M1 expansion too (if we're talking dollars).

Another variable that's known is "volume", for example.


I guess we disagree a bit here. I would say it's easier than Forex because it's not tied to an economy. While you believe being tied to a country simplifies things (because of data - don't get me wrong, I see your point), I use to model economies, and I believe being tied to a country can complicate things. Although more data may be available, the "right" data is not and it's not very reliable. Supply and velocity are largely unknown with domestic currencies because of cash. And the demand issue remains. Bitcoin solves the supply problem because it's quantifiable. In fact, I've created weighted exchange rate models in the past and I think this is quite similar. There's no reason to think that the variables for a simple OLS regression aren't available, which could be used to predict price. I know for sure it's tried in Forex markets. Anyway, if the results end up significant with a nice R-square and coefficients following logic, it's applicable.

And from there, that model could be used to predict alt coin prices in a top down simultaneous model.

So, back to what "other" variables we could use...

What other factors apart from supply are quantitatively known?

I wouldn't say that its easier than Forex. FX long term forecasts has multitudes of macro economic data going in. For the short term, the price history depth makes autoregressive models possible. Neither of these factors are available in BTC.

How would one obtain an aggregate on the amount of bitcoin-related news? I was thinking there has to be a source for how often "bitcoin" is searched daily or mentioned on twitter. But, I'm unaware of anything.

I agree that a large portion of the price is speculative. However, I consider Bitcoin's price to be more practical to forecast than ForEx currencies, particularly when you're looking to forecast in the short term. My main reason is the supply and many other factors are known and available. Because of this, all one really needs is what's going to drive demand or how to effectively measure that - quantitatively. 

Amount of bitcoin-related news and whether it is positive or negative.
Previous price changes (because of momentum).

Much of bitcoin price is driven by speculation and non quantifiable factors. At the end of the day it is a software/code and you deem some value out of it. It does not generate revenue of its own and neither pay dividends. It is not related to any country's GDP.

Perhaps use an auto-correlation model, with factors for short term change in the price. may be think on the lines of gold.

OK. So, I would like to set up a forecast model for Bitcoin price. I have ideas on what independent variables impact the price, but I'd like to ask a crowd just in cast I'm missing something.

The model would like this:

Bitcoin's price (today or tomorrow) is a function of X1, X2, X3... Xn.

So, please chime in with the following what is your thought on an independent variable (X1) that impacts Bitcoin's price? Does the variable impact today's price, tomorrow's or both? Finally, and most importantly is the variable quantifiable and what is the source of the data? Why?

I'll give an example - X1 is Bitcoin's available supply. The available supply impacts today's and tomorrow's price. The available supply is quantifiable and the data source could be obtained from blockchain. Quantity supplied impacts price (or exchange rate) - basic econ 101.