Topic: Logarithmic (non-linear) regression - Bitcoin estimated value - page 12. (Read 117628 times)

@Trolololo you can post something like a "chart update" every 2 weeks or so to check how's the chart going and compare time-periods to know if it's following the logarithmic regression  

This days price was falling down but it's recovering step by step, so IMAO could be nice if we can see it updated on your chart  
Thanks!

Trolololo, I just tipped you a small amount in advance.

I'm not a big fan of Metcalfe's law as a valuation metric anyway, but I was just pointing out that this graph supports it.

I should clarify that by "a little patience is really all that is required here" I meant to see if the model works. I wasn't suggesting that I think patience is all that is needed to see high prices. I'm pretty uncertain about that.

Also, we might note that n log(n) doesn't diverge that much from n^2 if n is "small"

Arg, where's that article on Metcalfe's Law being wrong... I feel (vague, I know, I know) like it's more n log(n) than n^2.

Eh, maybe n^2 is the better model during early growth, though.

Same here. I'd love to see the actual chart updated. Will donate.

Agreed, certainly seems we where ahead of the curve during the last ATH.
Would be nice for some updated charts covering into 2015 ....

A little patience is really all that is required here. This model is calling for 1000 USD/BTC in three months or so. Sooner or later the price will either converge with the model or will diverge so much that the model becomes absurd. So far neither has happened.

There is a certain structural validity to it, if you accept Metcalf's Law as a valuation metric and observe that users have been growing roughly linearly (reported by coinbase and others).

If the guys making the charts could actually predict the market movements, they'd be rich

lol, did this debate fizzle before it even got started?

Paging a statistics expert prepared to defend the use of the ordinary least squares method.

Yes, it would be nice.

You can check today's value HERE

Nice! Thanks for the update, @lesni4ok.
Could be interesting to build that charts on an online-system that updates automatically and let everyone check in real-time   Don't you think so?

I don't know if there's something similar  

As I posted time ago, R^2 is not 0'98... but quite smaller (possibly around 0'91...).
I will recalculate it in the next update.

R^2 = 0,988 is not bad.

Oh, how i hope that this analysis is right, but BTC is, imho, one kind of market that can be used in this case

anywais in 4 months will see if the first prediction will be right!

I like how the channel goes from $15 to $60k. 

http://www.everego.com/btc-dxy.jpg

Plotting a best fitting line through a time series makes no statistical sense. It doesn't matter that you have a high R2 or that everyone else is doing it.

See here: https://bitcointalksearch.org/topic/non-spreadsheet-long-term-predictions-600802

Another problem that I also discuss is that model parameters change over time. How far back should we go to estimate the current parameters going forward?

This is actually an issue now. Because if we use the past half year or a year of data to estimate model parameters -- then the trend is DOWN.

Hey, thanks for sharing your work.


I'm not sure the linear price axis graph should look like an S curve. I believe this could be true for adoption or #users, network hashing power + others, their growth is bound by max resources, Bitcoin price however isn't bound by any resources in my opinion. That's why I prefer your model to Stephen Reed's logistic model, you don't have an arbitrarily $1000000 price limit.

EDIT : even #users isn't bound to max resources if you count bots, agents, DACs, machines, ... in the future.

Yes, that is a good question.  My opinion (but only my personal opinion) is that for something to be a long term store of value, it needs more trust than to be a means of payment.  Both are monetary assets of course, but with different functions and time scales.  I'm strongly influenced by the Austrian school and by von Mises and Rothbard's thinking.  Of course, these people didn't know about computers and internet and so on, and they are highly tied to the gold standard, which I'm personally not.
But there's something to say for gold, which bitcoin hasn't: "historical proof of trust".
There's a difference between taking a high-risk, high potential return speculation (like I do too with bitcoin) with a small part of one's savings, and to need something to build your retirement on.  There you need trust, or (and!) diversification, to hedge the breach of trust.

So my personal opinion is that a long term store of value needs trust, and trust takes time to build.  This is why I don't believe that bitcoin, as of yet, can take a large part in the market share of the aggregate demand for "store of value" in the long term.  Of course, lots of people are holding on bitcoin, but that is *speculative*.  They think they are first adopters of a monetary asset that hasn't yet reached, by far, its full potential.  As such, the tradeoff between the lack of trust (it is too new) and the potential benefit (it will "go to the moon") turns it into nevertheless an attractive asset.  But a high-risk/high gain asset.  That's not "a store of value".

Stores of value are things that need trust.  Real estate, gold, "trust-worthy" (haha) pension funds, and a diversified portfolio.  State bonds (haha).  

The high potential gain is in my opinion not an offset for the high risk, as a storage of value.

But indeed, if bitcoin works essentially as a storage of value, then its market cap (and hence the bitcoin price) will not so much be determined by M x V = P x Q, but rather by the fraction it will cover in the aggregate demand for "storage of value" worldwide.

That is essentially unpredictable for the moment.  You could just as well assume that bitcoin will completely replace gold (6 trillion worth or something), or will replace 1% of it.  As I said, for the moment I have a hard time believing that there is sufficient trust in bitcoin to replace the century-long trust in gold.  But I may be too conservative.

For the moment, it is hard to distinguish what part of holding bitcoin is speculative for "to the moon", and what part is just "holding value".  My personal guess is that it is 99.9% the first.  That not many people are considering bitcoin as a storage of value were it not for its large potential to increase.  But then, it is *not* a storage of value, but rather a high-risk high return asset.  Like shares in a start-up.


That's also how I understand it.  You should also take into account a risk aversion factor, which is especially sensitive to long-time storage of wealth, if your pension depends on it for instance.

Between something worth 100 for sure, and something worth 1000 with a chance of 10% and worth 0 with a chance of 90% (same expectation value), most people prefer the first, by a large factor.  I don't know what the risk factor is.  


The market ideally (ideally, because now it is probably manipulated by traders on the rather short term with effects on the longer term) takes into account all available information.  If the probability jumped from 1% to 10%, ideally, the (speculative) price should jump by a factor of 10.  But how could this probability suddenly jump ?
It could by an improbable event like a country officially adopting bitcoin as its currency.  The day the Swiss government, after a public voting, announces that the Swiss Franc is replaced by bitcoin, yes :-)  It could by an oil country announcing that oil will now be paid in bitcoin.  Or something of the kind.  Otherwise, only a smooth evolution is thinkable, no ?


Ok,let us take just an illustration.

Let us assume (it is just for sake of having an idea and an illustration, I don't imply anything) that bitcoin is *the* money for the drugs market, and for nothing else. (again, I'm not implying anything, it is just to get some numbers for an illustration).

The world drugs market is estimated, according to wiki http://en.wikipedia.org/wiki/Illegal_drug_trade to $300 billion in 2003.  So let us estimate that now at $500 billion.  Now, if the velocity of money in that world is comparable to the M2 velocity of money in the US according to the FED, that's between a factor of 2 and 4 grossly (see wiki http://en.wikipedia.org/wiki/Velocity_of_money).

If the entire drugs economy were bitcoin and nothing else were bitcoin, and we take that the bitcoin velocity is similar to the dollar velocity in the normal economy (highly uncertain) then the estimated market cap of bitcoin would be $500 billion / 4 = $ 125 billion.

We take the velocity of bitcoin here to be 4.  That means that bitcoins are traded on average 4 times a year to buy drugs with.  Some will be held years, others will be spend ten times a year.  Like your savings account and your cash in your wallet.  But on average, it is 4.

That is, if all bitcoins took part in that economy, with the same distribution of "holding times" as M2 dollar money in the US economy.

In other words, in that case, the price of bitcoin would have to rise with something like a factor of 25, to about $ 10 000,-, just to sustain the purely monetary aspect of this business.

In this case, bitcoin is not really used any more as long-term store of value than fiat dollars in the US economy.  It is the merchant usage of bitcoin that drives its price: the demand for bitcoins to be able to do all this trading sets the price.  It is what I considered with the quantity theory of money formula.

As I personally do not think that bitcoin can take a large part in the "store of value" market (without the speculative "to the moon" drive) for the moment (because of lack of trust), but I think it is merchant adoption that will be a fundamental to the bitcoin price, it is the above kind of estimation that gives an idea of what value to speculate for (multiplied with the probability of the scenario happening, and reduced with a risk aversion factor).

But I may be wrong, and it may be the "store of value" market that drives the price.  If you estimate that bitcoin will take on 1% of the gold market, that's good for $60 billion.  If you give that a 10% chance of happening, well, we are not very far from the current market cap.

thanks for the great insight.

I always thought that the quantity theory of money would fail to deal with the store of value function of money (or hodling) assuming that money is always neutral.

You seem to have a deeper understanding of that - so I have a few questions:

1.) I think that bitcoins function as a store value will for quite a long time be more important than its usage (even though being interdependent). Can you give more insight regarding factor T? Or provide a good (academic) article regarding that factor in this particular kind of analysis in QTM?

2.) It is obvious that the price is hugely driven by speculation - but given todays market capitalization (~5 Billion) we could say that it shows (no one knows if this is really possible on methodological level) that we give it a 1 % chance to get 1 % of the trade and store of value of the world economy (500 Billion). or do I misunderstand that?

3.) I always have the assumption that we as economist fail when time comes into play . But you tried to give some insight regarding that, when you say that you do not think that it will be an overnight ride to the moon and assume that it takes a decade. But let us simply assume that perception regarding bitcoin changes due to some factors. At question 2 we had a 1% chance at todays prices to reach 1% of the world market. for simplification let us assume we change to a 10% chance. Don't you think that prices would climb much faster, even given the same or nearly the same velocity?

4.) given your formula. Do you think it is possible to fill it with numbers, to give us a speculation free price of bitcoin - I think I do not completely get your factor t and how to use it.