Quote from: Wandererfromthenorth on May 21, 2015, 03:34:11 PM

Ok, so what do you use to know when to buy, hold or sell?

Ufortunately it seems that the price of bitcoin is quite unpredictable.

It is obvious that the price responds to some news and rumors. Since 2014 there have been a dozen or so moderately large price events -- steps, spikes and dips 50--100 $/BTC -- that are clearly linked to such media events: by their size and direction, timing, and comments in forums.

Most of those events originated in China, and affected only Chinese traders: the PBoC decrees of Dec/2013, the rumors and anti-rumors in Mar--Apr/2014 about further PBoC restrictions, the closure of bank accounts of Chinese exchanges, and the closure of one or two Exchanges.

There were a couple of moves due to events that affected everybody (Mark's claim of a "bug in the protocol" on 2014-02-10, the first USMS sale in Jun-Jul/2014, the rumor of "three billion hedge fund about to trade on OKCoin").

Finally there were few smaller price spikes linked to Western news and rumors, such as "Microsoft will accept bitcoin", "McDonalds will accept bitcoin on Valentine's day, and the Coinbase "lunar" teaser.

Obviously, none of these events can be predicted in advance, At best, one could recognize them when they are just starting, a couple of hours before everybody else. But, even then, it seems impossibel to predict how big the effect will be and when the change will stop or reverse.

Looking at all the history up to 2013, the main price movements can be described surpisingly well by the sum of about 10 overlapping "idealized bubbles" with simple and qualitatively similar shapes. See Figure 1 in particular. Each of those idealized bubbles has a fast exponential rise that stops suddenly, followed by a slower exponential decay. In a couple of cases, it is necessary to add a flat section between the rise and the decay. Each of those bubbles could be explained as the epidemic diffusion of bitcoin through some new "market", the saturation of that market, and its gradual shrinking.

The bubble that peaked on Nov/2013 was almost certainly caused by the opening of the "market" of amateur commodity speculators in Mainland China, a demand that was primarily served by Huobi and OKCoin, and perhaps by BTC-China.

There was a sudden rise starting around 2014-05-20, from ~450 $/Ƀ to ~680, that I cannot connect to any event. But that rise can be well described by another small bubble, peaking in early Jun/214. Since the decay rate of that mini-bubble seems to mtch that of the Nov/2013 bubble, peraps it is some new segment of the Chinese market, or just a temporary recovery of the same market.

The bubble that peaked in Apr/2013 may have been due to Chinese demand too, perhaps limited to Shanghai and/or Hong Kong, served by BTC-China and Bitfinex.

I do not know what "markets" could have been resposible for the previous bubbles (I only started following the bitcoin scene only in late 2013).

According to that model, the price now is still dominated by the tail (decay) of the Nov/2013 bubble and of the Jun/2014 mini-bubble, with smaller ups-and-downs due to news and rumors. (See Figure 7 for a model that includes the latter.)

As for predicitions, all that the model can say is that, if there are no new bubbles, the price will probably continue decaying exponentially, by the rate that has prevailed since Feb/2014. However, the bubbles do not seem to follow any pattern; as they shouldn't, if they are due to separate markets being opened. Moreover, the fast exponential rise means that each new bubble cannot be detected until it is already well underway; and its peak cannot be detected until it has clearly happened. Which would explain why no one seems able to predict the price, even in the short term.