Topic: rpietila Wall Observer - the Quality TA Thread ;) - page 275. (Read 907227 times)

So so while the utility of Bitcoin is predominately speculation, then my theory as presented is that price p still has significant feedback loop on n the proxy for adoption.

Fundamental adoption is growing exponentially, but I am arguing that superimposed on that trend is the reverse feedback loop of p on n.

Bitcoin and silver may be similar in that adoption is very sensitive to rising prices, because they are both predominately speculations at this stage, although Bitcoin is much more rapidly developing network effects and I posit that is why the distance between the crashes in Bitcoin is shorter than for silver (less than a year versus more than 2 years).

We had hoped to get network effects with silver, and Risto probably did more on that than anyone, getting silver retailed at the Post Office, created the most liquid market in silver for small retail investors emulating the concept of the London daily fix, and creating silverbank. I tried to do my part minting silver coins from 1000oz bars during the crash to $9 to get more supply of coins into the market. I supplied to Risto, Tulving (he didn't know it), and others. It was very risky and I think perhaps the mint I used ended up screwing over some people after I got out.

No where near as much fun as the software startups I did. Risto's serendipitous (mis)adventures was the most endearing part of it. I hate tangible ventures (love tangible for personal things such as sports and love).

http://blockchain.info/en

How can switch Blockchain.info default language away from Turkish to another language? It reverts back every time I click a page.

I would have said in May.  Adoption is a gradual thing. The next adoption bump should be much bigger in that case.

Gbianchi found the best R^2 fit was n^2.26.  I find the extra sqrt(sqrt(n)) intriguing but frustrating.  Its easiest to blame it on stochastic deviation but something bugs me about it.

p = price
n = number of unique addresses (a more relevant proxy for and than number of users)

If n at 150,000 is 33% too high above the bottoms trendline since 2012 (use a ruler on the "All time" plot), and if p is tracking n^2 per Metcalf's Law and Peter R's chart, then a 0.67 x 0.67 = 0.45 of recent price could be expected.

If I understood Aminorex's point, the correlation that Peter R's chart upthread showed is predominately n^2 driving p, because n^2 has more structural significance than p.

In other words, either just purely mathematical a n^2 term in a polynomial has more weight than a n term (because for example in the first derivative the n becomes constant, i.e. no velocity only position and I think Aminorex is also looking at it from a statistical math framework), or I added some real world interpretation noting that network effects (e.g. the creation of Bitpay and Coinbase) is reflected in those n^2 connections. Reed's law says that as n is nodes in network, maximum connections is n^2. Changes in price can't make Bitpay and Coinbase disappear. Thus n^2 has more significance than p.

Yet I am pointing out that probably p has some reverse feedback on n (the number of unique addresses, i.e. proportional to users), because some users react to price and sentiment considering that they entered because of a pumped up media blitz in November. So as p declines, n can decline on the margins.

And the correlation of the charts seems to say this is true on all the prior p crashes the n dropped (did Peter R show a chart back to 2011 so we can see if it is always proportionally to n^2 on the p declines?).

One could argue that something else beside p change caused n to drop which caused p to drop. I find that intuitively difficult to support because p is so important to the feeling of the users especially at this stage when Bitcoin is primary a speculation vehicle and not a utility vehicle like a washing machine or the internet. Do we want the internet to go up in price? No we use it because of utility. Bitcoin is not at that stage yet.

Aminorex is most convincing  You both sound incomparable.

And it all sounds wonderful but is completely impenetrable.  If BTC crashes and burns tomorrow all the theoretical thinking will suddenly change "ah...we forgot to factor in...blah blah blah"  This happens often does it not, often because the theory is divorced from reality or based on inaccurate data.  

From my simplistic PoV the chart shows address used not the number of users?
My point is every time I use BTC I create a new address to either send from or receive to but I'm still a single user.  That probably sounds illogical but I suspect my behaviour is fairly commonplace.

Since the number of tx has remained steady, why not the number of users?

Discriminating between signal and noise is a selective process.  This is the chart that meets the requirements for fitting to metcalfes law.  Value transfered per unit of time would be needed for fisher.  The laws can be validly applied and thus are useful.  I don't know of a generative model for fundamental value which is parameterized by or fit to the other charted factors, so I can't easily derive signal from them.  Lots of structural inference would be required, and need to be proven out.

The easy and sensible thing would be to do vector ARIMA.   Phase plots are likely to reveal some structure too - at least show where lag is varying.

creekbore, I am now convinced by aminorex's explanation (changed my mind) and I now think my point about unused addresses is irrelevant, as it is just part of the n^2 network effects.

However, I am still thinking p has some reverse feedback on n on the margins.

Thanks for replying ...I can always rely on you for a sensible response
And I'm not being deliberately contrarian...I hope BTC succeeds (for purely selfish reasons) but having been through three bubbles and crashes now this one feels very different.

@AnonyMint raises the points I would (individuals have multiple wallets with little in them) and addresses are even more frequent per user (I rarely use the same address twice and imagine most casual users do the same...it infers a level of uniqueness to a tx).  Do you do the same?

The other charts seem to back up the theory that we have fewer people using BTC, so claiming 'this one is the important one' could be viewed as being selective too.

Given all the other evidence that points to the contrary, it seems rather hopeful to base ones attitude (ie bullish on BTC) on a single  chart.

I understand n^2 has more structural significance than p because it represents network effects. (connections between nodes are indicative of network effects at play)

I am thinking p has some influence on n on the margins, specifically the weak hands speculators.

The degree of this feedback would be proportional the overextention of weak hands due to irrational exuberance.

I would only admit that variably lagged vector autocorrelation was evidence of a degree of feedback.  Driving is a strong word.  The feedback might drive occasional corrective jumps, rare and brief.

High IQ statement.

Unsure what the decision is.  Several laws hold usefully.  At any moment zero or more may dominate.  Metcalfe may only dominate on fairly long time scales.  Fisher may dominate with higher frequency but has less predictive power because it hides so much structure:  You  really need to do marginal analysis to get the right MV.    If you are saying metcalfe cant call a bottom, yes, we are agreed.

The trendline on the bottoms of your chart since Jan 2012 is below 100,000. Probably need to come back to that before we can move up again.

We are 33% too high.

Not if p is driving n. I think confidence and sentiment drives n, at least on the margins (there is probably a fundamental low frequency trend on n on which the marginal higher frequency is superimposed). Everyone was super bullish, now they are in the process of denial and trying to adjust to the new sentiment. Some of those n will throw in the towel, but they will be replaced by the stronger hands from the lower frequency trend. Then eventually those weak hands come rushing back in again.

Causal relationship is from n^2 to p.  p should capitulate.

Okay then you agree that chart is inconclusive for us at this decision point?

Note your chart correlated with the prior major crashes.

Lag on that factor signal should be large, blended, and volatile.  Poisson perhaps.  Nontrivial to extract and low weight in a generative model.

We already saw a decline on your chart correlating with the dump from $1200 in December. And now we see another decline on your chart starting recently. So now all aspects are corroborated.

So as I wrote upthread, the marginal price drives the adoption not the other way around. In other words as the speculative sentiment shifts, the adoption follows.

Edit: The number of unique address on your chart has diverged from the BTC price since February. One of the two has to capitulate to the other soon. Perhaps that is the confusion (Risto referred to) that has been holding the market up (or down depending on your perspective).