I've seen your model, and I like it. I consider it somewhat more motivated than a pure time-based one. Btw, I think a (roughly) similar one is being developed by an Italian guy whose name I forgot (something bianchi), who measures "adoption" based on zero addresses.

Anyway, an improvement maybe, but still based on one crucial assumption: user base is adequately reflected in total number of addresses (or transactions, or whatever info you prefer to download from blockchaininfo).

Not that it's wrong per se, but it's not obvious either. We're still pretty early into the adoption (and usage), and it's too early to say with certainty if we can use that data unfiltered to estimate the user base (I know, I know, "popular addresses excluded").

Yes. Your presentation of it.


"Buy now, because we are (almost) guaranteed to turn around anytime now, based on the log trend."

That's the essence of your sales talk, and it is doing a disservice to Bitcoin.


"Buy now, it has historically proven to be a good approximate spot, but be prepared to see your USD value go down further in case the bear market continues."

That'd be an honest presentation of the historic results, but one that I have yet to see in this form.


In case you accuse me of putting words in your mouth, here's the type of presentation I have in mind:


There is no such thing as "rock bottom" of the trendline. If the market continues to go down south, your model will pick that up long after the fact.

For fairness sake, you do add disclaimers when pressed (like: "don't invest what you can't afford to lose"), but you are still regularly abusing your model in discussions about short-to-mid term predictions (e.g. was this the bottom or not?), where this model has no place whatsoever.

This is a remarkably incoherent post for you. Guess the nerves of the hodlers are wearing thin at the moment.


>I find it hard to imagine how you can consider it anything but dispassionately objective to note that the historic trend is still holding

Who said anything about the "historic trend [not] holding"? I certainly didn't.

I simply reject the pathetically transparent attempt to turn the historic trend into something that it isn't, which is pretending that it is: (a) a guarantee that it will continue to grow like that, and (b) even when assuming point a) holds, that we see a continuation of growth *now*(ish).

Neither of the two points is a consequence of the historic trend analysis (how could it?), but it is regularly presented as such. Hence: snake oil sales talk.

This is getting a bit kindergarten-y, isn't it?


Anyway.

The problem with your log-linear model isn't that it's not updated regularly. I know you do that.

The problem is the *extreme* laggyness of the model, and how you intend to use it. By the time your regression picks up the fact that we are in a huge bear market*, it is already much too late.

The log-linear trend can maybe, if you insist, be used to get an idea of the order of magnitude of the expected price, assuming nothing about the growth function up to now has fundamentally changed. And even then you should remind people that it is no guarantee either, just a reasonably motivated model.


* EDIT not our current situation, but what would happen if we see, say, another few months of bearish market.

Agreed. The picture looks even a bit worse on Huobi than on Stamp.

But I don't buy the commonly repeated mantra "Only Huobi/China matters at the moment". Or rather: China might determine the price, but not in the way that people commonly imagine how it works.

If we had a complete, formal model of Bitcoin price discovery that includes a model of the separate but connected exchanges, then I would in fact fully expect that one of the results of that model would be something like "current price is determined to a larger degree by China than by, say, Bitstamp".

But that's /not/ the same as saying "Huobi determines the price, then the rest of the world follows it". The process is, imo, more complex than this, and in reality involves a mutual exchange of information between the exchanges as trading takes place.

The only clear cut case of China really setting the price that others follow is through arbitrage bots, but arbitrage is (obviously) not what determines the current price level in the first place.

Where I'm getting with this:

I don't believe there is a simple leading/following division between the exchanges, except for a few individual movements, and it is perfectly possible to read the trends on either of the exchanges.

In fact, I consider many signals to be better visible on Bitstamp than on Huobi, perhaps because of a worse signal/noise ratio on Huobi that has at least a doubtful amount of volume. I explain it to myself (this is pure conjecture now of course) by noting that human traders ultimately determine the price level, but that human trading is nearly drowned out on the Chinese exchanges during most of the day through automated trading.


To be clear, I don't dismiss Huobi for getting short-term signals, but (a) I believe I can read all the relevant signals (above, say, hourly time frame) from Bitstamp data, because price is highly correlated across the exchanges, and (b) because of the high volume of unclear origin on Huobi, it is perhaps actually better not to rely on Huobi data too much for analysis.

Tail firmly tucked between your legs. As was to be expected when someone calls you out on your many bluffs.

I bet it'd be fun to play poker against you.



Will you, for fuck's sake, let go already of your harebrained attempt to shoehorn a log-linear regression based trendline into short-to-mid-term trading advice?

Get it into your head: If (for whatever independent reasons) we will stay below that trendline long enough, it will look substantially different from the one you constructed right now. Happened in 2011, might happen again in 2014. Your "advice" of buying because we are currently below it is based on wishful thinking and extremely biased interpretation of the data.

But I waste my breath here. You made it abundantly clear over the years that you choose a good snake oil sales talk over objectivity any time of the day.

Huh. This is a new level of pathetic reaction to a little bit of bear market....

Let me put it in simple terms: If Bitcoin isn't resistant enough (or as some like to call it: anti-fragile) to survive the normal processes in a market economy, then it is going to be replaced by something that is able to survive just that.


tl;dr The (unwritten) rules of capitalism aren't turned off at your convenience.

The passive aggressive reaction to unwelcome news, the jerkish attitude to the messenger bearing said news (sorry y3804), and the pure bile poured over China and the Chinese in this thread (and the forum in general)  is nothing short of depressing.

Grow up, Bitcoin investors. You didn't whine when China pushed us to 1000+, so now that they put in the reverse gear, you should bear* it with some dignity as well.


* yea, couldn't resist the pun, sorry :)

I'm not that impressed by our current recovery from the April 25 downwards move. We bounced off from 50% as clean as can be, are currently back to the area between 23% and 38% (i.e. the very first retracement step), and the first level, 23%, looks pretty perforated to me. Taking into account volume as well, I'm leaning towards a re-test of 43x.

+1

At 30 days the bet is already at a real risk of ending in a tie. At 90 days, under the original conditions, that's almost guaranteed.

The way I see it, this bet is about making a rather strong, unconditional statement in the middle of a crucial time for the market. At 30 days, it does that just fine.

EDIT: I obviously have no say in this, it's not my bet. But since this is discussed in a public place, you should expect it to be discussed.

Wrote him an email, no response at all.

One more post on triangles (of doom), i.e. possible boundaries of a continuation of the downtrend that governed price action since early December (We're getting close to 5 months now. Quite the bear market, huh?)





(A) is the log trend line connecting the December ATH with the early January peak (near 1000). I don't consider it the most likely candidate for the main downtrend, for two reasons: a) too little points of contact, and b) price action has been too far away from the trendline for a long time, i.e. it's not only not touching the line, it's not even /near/ it for months now.

(B) is my main candidate for the upper bound of our downtrend. 3 points of contact, and price in general staying close to it to make it appear a relevant trend even at times when price doesn't touch the line.

(C) is the old lower boundary, the line through the 380-400 local bottoms. It's a log line as well, but only very weakly slopes upwards. That trendline was broken of course, on April 10/11, and again two days later, but it might still act as resistance. After all, we only closed one daily candle below it (another daily candle two days later is a borderline case), and we bounced off of it on decent volume, so you could argue the line was tested and held (sort of).

(D) is the log trendline between the 2013 ATH (at 266) and the current capitulation bottom (at 340). It is based on two assumptions a) 340 was indeed the final bottom (for which one can find independent reasons, such as the observation that we bounced off of the lower weekly Bollinger Band the week after the April 11 capitulation), and b) there is a very small (exponential) growth between the *high* of the previous bubble cycle and the final bottom of the *current* bubble cycle.

Taken together, those trendlines meet at 4 points:

(1) (Latest) Time of conclusion: Late May. Price point of conclusion: low to mid 400s. A possible candidate to mark the end of the December bear market. It would mean the 400 trendline held after all, but so does the more severe downtrend (B). Assuming it would  break out downwards, we would get to:

(2) Late June, mid 300s. My preferred candidate (EDIT: "preferred" as in: the most likely one) for a resolution of the bear market. It assumes the log line from previous ATH to current bottom holds, as does downtrend B, and once we hit the 300s again we might have enough force to finally break through B decisively.

(3) Early/Mid July, mid 400s. Assumes the 400 trendline holds now, but the B downtrend wasn't the relevant downtrend. Not very likely, in my opinion.

(4) Mid August, slightly below 400. The most dragged out consolidation triangle. Slightly more likely than scenario 3 I would say: it'd give the market a lot more time to take a "time out", without any major new price breakthroughs, neither up nor down. Think: No fresh fiat arriving for a long time, but the time for huge panic selling is over as well.

You sure of that? Last time I looked into it, it seemed to be working on the rule I described.

And of course, every 'buy' has a matching 'sell', but for every matched order you can say which one was the one that triggered the match. So at the very least, they should do it like that, so that a large "dump" (i.e. big market sell) counts as fiat out. It would be a bit like a crude Chaikin Money Flow...

But why am I even doing this... this site is entirely useless, and I doubt they have any intention of improving it

hehehe, welcome to the discussion from 2 weeks ago. Forgot who was the first one to suggest using a log chart for the December downtrend... but I suspect the thought appeared to more people around the same time as the bear market stretched out.

I find the most plausible log downtrend right now not the one connecting to the ATH in December (like yours seems to do), but touching the peaks on January 6 and March 3. I get more points of contact like that. That's the upper bound of my "extended triangle" from the previous page.

EDIT: Here's an illustration of what I mean:

Agreed, mostly. But I wouldn't dismiss the possibility that we saw the bottom on April 11th, and are now grinding down to a (slightly) higher low, which will terminate the bear market more clearly.

Hey, don't knock the V2. It was a brilliant piece of engineering and early aeronautic science. Just employed for, uhm, not so noble goals.

Forget fiatleak already. To my knowledge, they simply count buy order as "fiat in" and sell orders as "fiat out", which is hopelessly naive of course.

What /would/ indeed give a good idea of "fiat in/out" would be if they were able to show the fiat reserves of the exchanges and their changes... for obvious reasons, that's not happening though.

Quality TA. AAA+++. Would post again in risto's thread.

Good man.

Look at that guys...  3 green hourly candles in a row...  Time to unironically start posting those rockets huh

don't look at the volume though that'll only ruin the fun