Ding Dong! The Bubble-burst is done! $90-95 is the new normal

[solex]

Not so different from my fit curve... (orange dotted line)
"The fit curve is an exponential percentual to the chart least squares fit - in other words: I played around until it looked good."

I'm not seeing your image...

Should work now....  it's the same as on http://blockchained.com

Ah, yes. And great site. Very interesting.

[justusranvier]

It drains blood (power) from the centralized government(s) (who have ceased to act in the interest of their people).

Governments act today, as they always have throughout history, in the best interests of their people.

Anyone who thinks something has changed recently has just been excessively gullible when it comes to the question of exactly makes up "their people".

[monkeybars]

It drains blood (power) from the centralized government(s) (who have ceased to act in the interest of their people).

Governments act today, as they always have throughout history, in the best interests of their people.

Anyone who thinks something has changed recently has just been excessively gullible when it comes to the question of exactly makes up "their people".

Uh, what? Governments have been known to kill and oppress their own peoples for millenia.

[notme]

It drains blood (power) from the centralized government(s) (who have ceased to act in the interest of their people).

Governments act today, as they always have throughout history, in the best interests of their people.

Anyone who thinks something has changed recently has just been excessively gullible when it comes to the question of exactly makes up "their people".

Uh, what? Governments have been known to kill and oppress their own peoples for millenia.

He is drawing a distinction between the well connected elite and the labor stock.  The elite are "their people", not the common man.

[solex]

It drains blood (power) from the centralized government(s) (who have ceased to act in the interest of their people).

Governments act today, as they always have throughout history, in the best interests of their people.

Anyone who thinks something has changed recently has just been excessively gullible when it comes to the question of exactly makes up "their people".

Uh, what? Governments have been known to kill and oppress their own peoples for millenia.

He is drawing a distinction between the well connected elite and the labor stock.  The elite are "their people", not the common man.

Yes. And normally these people merge into society so it is difficult to spot them (apart from their late-model mercs), but occasionally they come to light in a spectacular manner:

http://www.ft.com/cms/s/0/dd63f9c2-ee02-11e2-a325-00144feabdc0.html

[oda.krell]

[...]
Yes it is simple linear regression (least squares fit). 1100 datapoints with slope projected 300 more. The data is presented raw (from bitcoincharts.com with no massaging).

I don't take this as gospel, just a rough guide. We are in the speculation forum and the best we can do in speculating is to make use of the data we have. There is no doubt that Bitcoin is on an exponential adoption phase, which has been seen with other technologies such as the take-up of mobile telephony. It will plateau at some point and skew level on charts such as the one I posted. We can have confidence that the regression line has short-term future relevance because of inertia within the Bitcoin ecosystem which is building all the time as more money and people get involved in many different countries.

If you allow, one more comment, the same one I made to [statistical modeling guy whose name I still can't remember but who has a carricature of Darwin as an avatar]: I don't object at all to regression analysis, and even extrapolating from it, but I am a bit suspicious of the underlying assumption of this particular analysis...

(aside: got into a huge discussion with the guy about whether it's an "assumption" or not. dude don't know his axioms.)

... the assumption that there is *one* underlying function generating the data. I find it hard to believe that traders' sentiment, btc fundamentals, number of market participants, bitcoin protocol related hard facts, etc. would all be identical, or in theory at least perfectly predictable, over the years, which in principle would have to be the case for the same growth function to generate the data in 2011 and in 2013.

I know, it's all simplifying assumptions. I'm not really objecting to using your method to predict the price. My own methods are in fact less sophisticated. I just like to soapbox a bit about my pet belief that a number of functions (perhaps only slightly) greater than 1 governs the btc price.

[Spaceman_Spiff]

[...]
Yes it is simple linear regression (least squares fit). 1100 datapoints with slope projected 300 more. The data is presented raw (from bitcoincharts.com with no massaging).

I don't take this as gospel, just a rough guide. We are in the speculation forum and the best we can do in speculating is to make use of the data we have. There is no doubt that Bitcoin is on an exponential adoption phase, which has been seen with other technologies such as the take-up of mobile telephony. It will plateau at some point and skew level on charts such as the one I posted. We can have confidence that the regression line has short-term future relevance because of inertia within the Bitcoin ecosystem which is building all the time as more money and people get involved in many different countries.

If you allow, one more comment, the same one I made to [statistical modeling guy whose name I still can't remember but who has a carricature of Darwin as an avatar]: I don't object at all to regression analysis, and even extrapolating from it, but I am a bit suspicious of the underlying assumption of this particular analysis...

(aside: got into a huge discussion with the guy about whether it's an "assumption" or not. dude don't know his axioms.)

... the assumption that there is *one* underlying function generating the data. I find it hard to believe that traders' sentiment, btc fundamentals, number of market participants, bitcoin protocol related hard facts, etc. would all be identical, or in theory at least perfectly predictable, over the years, which in principle would have to be the case for the same growth function to generate the data in 2011 and in 2013.

I know, it's all simplifying assumptions. I'm not really objecting to using your method to predict the price. My own methods are in fact less sophisticated. I just like to soapbox a bit about my pet belief that a number of functions (perhaps only slightly) greater than 1 governs the btc price.

Simple function works fine for Moore's law, no?

[Spaceman_Spiff]

Simple function works fine for Moore's law, no?

It does, but that does not mean that there isn't variance. And on an exponential scale a little bit of variance can cause your position to be overrun.

Yeah, I just meant for approximating LT trends.

[oda.krell]

[...]

Simple function works fine for Moore's law, no?

Yes and no. Actually, a single exponential growth function could be used to paint a pretty little picture of btc/USD development until now. Or Apple's stock. Or bacterial growth. However, that doesn't mean that in each of those cases only one function actually is behind that data -- in fact, it is extremely unlikely, given my argument (identical parameters, or perfect predictability of them) above.

In the end it's all about precision. If you want to predict whether, all things going mostly alright, btc/USD will be up a lot from where we are now within 3 years, then yes, the single function model is most likely good enough. If we want to decide whether we're in a weekly slump of a mid-term trend reversal of a mid-to-long-term bear market, then maybe the same simplification is a bit too much.

[Spaceman_Spiff]

[...]

Simple function works fine for Moore's law, no?

Yes and no. Actually, a single exponential growth function could be used to paint a pretty little picture of btc/USD development until now. Or Apple's stock. Or bacterial growth. However, that doesn't mean that in each of those cases only one function actually is behind that data -- in fact, it is extremely unlikely, given my argument (identical parameters, or perfect predictability of them) above.

In the end it's all about precision. If you want to predict whether, all things going mostly alright, btc/USD will be up a lot from where we are now within 3 years, then yes, the single function model is most likely good enough. If we want to decide whether we're in a weekly slump of a mid-term trend reversal of a mid-to-long-term bear market, then maybe the same simplification is a bit too much.

Even if applicable, the simple function is of course a result of various subfunctions, like you said.  And it's clear that this exponential growth (with its value appreciation) can't last forever.  However, as bitcoin is still relatively small, it currently remains my price guidance for the rosy scenario.

[oda.krell]

[...]

Even if applicable, the simple function is of course a result of various subfunctions, like you said. And it's clear that this exponential growth (with its value appreciation) can't last forever.  However, as bitcoin is still relatively small, it currently remains my price guidance for the rosy scenario.

actually, that's exactly my point. That's the often heard objection "price growth can't be exponential, because such growth becomes unsustainable quickly". Which is absolutely true. But only if the rate of growth is constant.

Say we have a function characterized by:

"Beginning with a monthly growth rate of 20%, this value continuously decreases s.t. it is halved every 12 months"

(which would be a sigmoid function of some sort, I guess)

Simple linear regression performed on data plotted on a log chart, i.e. what solex did, is not able to model such a function. to my knowledge at least, someone please correct me if I'm wrong.

I'm not saying that this is the most likely candidate for the btc price function, but it's an example of what is outside the realm of linear regression.

[solex]

Say we have a function characterized by:

"Beginning with a monthly growth rate of 20%, this value continuously decreases s.t. it is halved every 12 months"

(which would be a sigmoid function of some sort, I guess)

Simple linear regression performed on data plotted on a log chart, i.e. what solex did, is not able to model such a function. to my knowledge at least, someone please correct me if I'm wrong.

I'm not saying that this is the most likely candidate for the btc price function, but it's an example of what is outside the realm of linear regression.

This is correct. A simple linear trendline on a log chart will plot constant acceleration, but not account for a second-order rate of change in that acceleration. If there is a slowing in the Bitcoin price (fx rate) acceleration then the data should evidence a curve or banana-shape over a longer timescale. What is very interesting is that in the last three years the curvature in the data is quite small. It is not obvious, to my eye anyway.

This was probably different in the past. If we divide Bitcoin into the nascent (pre-Mt Gox) phase Jan 2009 to May 2010, and the growth phase of May 2010 onward, then the price acceleration from the first sale above zero, then to $0.50, was different. Trace Mayer, on runtogold, has the growth rate in 2009 at 4,867%. I have no idea where is the data is for that number, but it is way above anything seen since. Having proper exchanges has stabilized the rate of price acceleration. The current acceleration definitely has a rate of change inherent, but it is not significant yet, although in a few years it will be apparent.

[oda.krell]

So did you ever regress over the data starting from a different assumption, say, assuming two (or three) independent (exponential) functions generated the data, and search for the optimal splitting point of the data, then compare the total error. (I'm lazy, I know. Why ask others to do the work that you could do yourself. My apologies.)

[molecular]

We are blabbing and yapping here about capitulations etc... completely ignoring simple fact that BTC's value is directly proportional to the level of adoption which continues to grow very quickly and expotentially (offset by new mined coins).

Markets can ignore fundamentals only for so long. Once the market decides to price in the current and future growth you will see 4 and soon after that 5 digit USD valuations of 1 BTC.

listen to this man, he is correct.

[ElectricMucus]

We are blabbing and yapping here about capitulations etc... completely ignoring simple fact that BTC's value is directly proportional to the level of adoption which continues to grow very quickly and expotentially (offset by new mined coins).

Markets can ignore fundamentals only for so long. Once the market decides to price in the current and future growth you will see 4 and soon after that 5 digit USD valuations of 1 BTC.

listen to this man, he is correct.

Says one early gpu miner about the other.

[GigaCoin]

I am calling the 2013 bubble-burst completed. This is based upon the observation that the 2013 bubble was structurally more similar to the August 2012 event than the 2011 one.

March 28 to April 1st - pre-2013 peak. Stability in low 90s.
April 18/19 - same during period of large oscillations.
July 19-23 ongoing - more stability in the 90s.

I also believe that the reward halving at the end of Nov 2012 was the key driver in the appreciation of the BTC fxrate to a new level. Markets always discount known data with a future effect. So for four years the market lived with the information that the 21 million cap probably would be effective. The reward halving proved that it will be effective. This is an important distinction, and a successful reward halving paradigm is now properly priced in.

The repricing likely coincided with large acquisitions (such as the Winklevoss holding being accumulated), coupled with news-frenzy and feedback from the telegenic but BTC-irrelevant Cypriot bank crisis, likely caused over-compensation, so the fxrate ran 150% higher than the fundamentals warranted.

That 150% surge has now dissipated.

Any takers on this reading?

As predicted, this prediction has failed.

[molecular]

We are blabbing and yapping here about capitulations etc... completely ignoring simple fact that BTC's value is directly proportional to the level of adoption which continues to grow very quickly and expotentially (offset by new mined coins).

Markets can ignore fundamentals only for so long. Once the market decides to price in the current and future growth you will see 4 and soon after that 5 digit USD valuations of 1 BTC.

listen to this man, he is correct.

Says one early gpu miner about the other.

says one early gpu miner to another who both managed to hold on to their coins through at least 2 bubbles because they knew the fundamentals were strong.

I've just today met a guy I had met in early 2011. We were both mining at the time. He sold everything he had in late 2011. Now he's dealing in mining chips and shit trying to regain some bitcoins.

[molecular]

I am calling the 2013 bubble-burst completed. This is based upon the observation that the 2013 bubble was structurally more similar to the August 2012 event than the 2011 one.

March 28 to April 1st - pre-2013 peak. Stability in low 90s.
April 18/19 - same during period of large oscillations.
July 19-23 ongoing - more stability in the 90s.

I also believe that the reward halving at the end of Nov 2012 was the key driver in the appreciation of the BTC fxrate to a new level. Markets always discount known data with a future effect. So for four years the market lived with the information that the 21 million cap probably would be effective. The reward halving proved that it will be effective. This is an important distinction, and a successful reward halving paradigm is now properly priced in.

The repricing likely coincided with large acquisitions (such as the Winklevoss holding being accumulated), coupled with news-frenzy and feedback from the telegenic but BTC-irrelevant Cypriot bank crisis, likely caused over-compensation, so the fxrate ran 150% higher than the fundamentals warranted.

That 150% surge has now dissipated.

Any takers on this reading?

As predicted, this prediction has failed.

huh? In what way exaclty has solex' prediction failed?

[vokain]

I am calling the 2013 bubble-burst completed. This is based upon the observation that the 2013 bubble was structurally more similar to the August 2012 event than the 2011 one.

March 28 to April 1st - pre-2013 peak. Stability in low 90s.
April 18/19 - same during period of large oscillations.
July 19-23 ongoing - more stability in the 90s.

I also believe that the reward halving at the end of Nov 2012 was the key driver in the appreciation of the BTC fxrate to a new level. Markets always discount known data with a future effect. So for four years the market lived with the information that the 21 million cap probably would be effective. The reward halving proved that it will be effective. This is an important distinction, and a successful reward halving paradigm is now properly priced in.

The repricing likely coincided with large acquisitions (such as the Winklevoss holding being accumulated), coupled with news-frenzy and feedback from the telegenic but BTC-irrelevant Cypriot bank crisis, likely caused over-compensation, so the fxrate ran 150% higher than the fundamentals warranted.

That 150% surge has now dissipated.

Any takers on this reading?

As predicted, this prediction has failed.

huh? In what way exaclty has solex' prediction failed?

right. Solex concluded the same thing i did. based on its similar behavior to Aug 2012's decline and progression into the new year, this bubble is done deflating and very likely about to do the same thing. time is all it needs. i'm guessing the next big correction won't be as severe, but who knows at that point. depends how quick we go up.