:

[TiagoTiago]

How does your current formula match previous periods? (go back in time, ignore things after that and make a prediction then match with what actually happened)

[1714943856]

1714943856

[1714943856]

1714943856

[TiagoTiago]

I'll rephrase: how good is your system at predicting the past? Pretend we're 6 months ago, if you had the system set the way it is now, how off would the prediction be from what really happen in these last 6 months?

[dserrano5]

Awesome work chodpaba .

[DrYe5]

I could just keep it to myself, but it's not like I am really a speculator anyway. More of a buy and hold guy myself when it comes to BTC.

How's this holding up the past 24-hours? Do you have revised forcasts? $11.97's looking a bit more likely?

[proudhon]

However, I will have a reason to revise it if the price rises significantly (2x) before August. That would indicate a correlation with a double time series that has a not-pretty outcome...

Will you explain that for me?

[hugolp]

However, I will have a reason to revise it if the price rises significantly (2x) before August. That would indicate a correlation with a double time series that has a not-pretty outcome...

Will you explain that for me?

The exact same thing that picked my interest.

[Tulkas]

Nice work, chodpaba!

Do you mind to share the source code of that simulation ? 

I work with signal processing in video, we use non linear prediction like this http://www.youtube.com/watch?v=yEdEA0V0uGg a lot, i've been trying to use self organising state space models to predict variance in two dimensional data, maybe i can extend your simulation to include self adjusting auto generated patters.

[Enky1974]

The support at 10$ make sense to me, considering price/difficulty ratio. Also 9$ was a previous resistance now support.
If the network size won't implode, 9-10$ zone should hold.

[TraderTimm]

chodpaba, appreciate the work you've put in, and your willingness to share! Some of what you have posted reminds me of another forecasting technique using Fast-Fourier Transform to extract dominant cycles and apply that as a forecast from current price. I do realize you are doing a probability-based projection, just wanted to share.

Also, I'd like to ask if anyone has daily data from 06/20 to present, that would be very helpful. I haven't kept up on my end with it, and would love to update my charts. (Which I'll share, of course.)

[Enky1974]

It is not good fourier transform to extract the dominant cycle, for stock market or currency data, you should use MESA (maximum entropy spectral analisys) for instantaneous dominant cycle estimation. For a while was present a 12 day cycle, now disappeared.

[vectorvictor]

I have made some slight revisions to the forecast.
[...]

How do I know I am right?

I do not and can not know that. I only know that it is similar to what has happened before, and the current phase in the cycle represented by the forecast seems to correlate to where we are now. I am just putting it out there so the method can either fail or succeed publicly... It helps to keep me focused.

If we zoom out to the big picture, we are undoubtedly looking at a "phase transition" type of graph -- a sigmoidal S-curve (https://secure.wikimedia.org/wikipedia/en/wiki/Sigmoid_function).

This is common and normal for the uptake of popular new technology -- the same basic graph is found if you plot the adoption rate of many things (e.g. cell phones, Twitter, whatever).  Once it hits the exponential growth phase, it rises very quickly (sometimes the graph can be nearly vertical), until it finally tapers off when it reaches the point of saturation.  For example, there are now around five billion cell phones in use, so you can understand why it no longer has explosive growth (the world's population is seven billion).

Even if the new thing is much less popular (say 0.001% of the population cares about it, so it peaks at a few million users), the graph looks much the same, and it tends to find its saturation point fairly quickly.  It is a power law, and fractal (i.e. similar across scales).  Even if two such processes are substantially different from each other, they tend to look pretty similar when plotted with our usual linear bias (we weren't built to see those differences clearly).

The short story is that it will probably grow fast, then level off (and stay there unless something else replaces it).

So the real question is whether bitcoin has reached its saturation point yet.

I don't think so.  I believe there are many big stories and broader markets yet to come.  There is even a chance that it becomes ubiquitous (i.e. the most common currency exchanged in the world).  The chance of that doesn't need to be very high to have windfall potential.  For example, a 1000-to-1 chance of increasing 100000-fold in value is still a fantastic wager, and a 10% chance of increasing 100-fold is still a good bet.  [Unlike other commodities, the price will go up because there is only a finite amount.  21 million is a small number, so we're really talking about 21 billion mBC, or 21 trillion uBC.  Imagining a market cap 1000 times bigger than now is very easy -- and that still wouldn't be as big as World of Warcraft, nevermind a Samsung, or a Norway.]

The whole bitcoin idea could also flop and go to zero, as we've seen with many new technologies that never caught on.  I don't think that's very likely though, given the passionate following of people who crave freedom.  (Not to mention the support of the black markets -- it can't go to zero if arms dealers start using it!).

I would look for the basic patterns that chodpaba has identified to repeat for at least a few more cycles before they taper off (not repeat exactly, mind you, that's not how fractals work).  And it will be good news for everyone once the adoption phase is complete, as the hoarding will stop, and the currency will start acting like a currency should.

[dserrano5]

Sorry for the dumb question but, what are the two coming from the left? Is green line actual price and red line some sort of moving average?

[proudhon]

I am becoming convinced that this looks more like December, 2010 than it does March, 2011.

Correct me if I'm reading the charts wrong, but if you're right, then we're looking at possibly more downside and then a gradual increases in price, possibly over a period of a month or more, to reach the previous high.  That's as opposed to the March, 2011 scenario which would be a gradual decrease over a longish period with a very sharp increase to previous highs and beyond in a very short period of time.  Is that right?

[proudhon]

This is what I am looking at now.



Targets:

To 2011-07-18
$11.833   $14.339   $16.863   $18.028

From 2011-07-19 to 2011-08-01
$12.738   $15.244   $17.228   $19.142

From 2011-08-02 to 2011-08-12
$16.775   $17.402   $20.325   $20.952

from 2011-08-13 to 2011-08-25
$20.117   $20.830   $22.692   $31.324

On this analysis it looks like, give or take the typical fluctuations, we've consolidated and are prepared to go up again.

[SlaveInDebt]

Enjoy your analysis greatly Chodpaba as well as your difficulty analysis. Just wanted to let you know I regard yours as the best by far.

[netrin]

Enjoy your analysis greatly Chodpaba as well as your difficulty analysis. Just wanted to let you know I regard yours as the best by far.

The market awaits Chodpada's new targets before departing from $14.

[netrin]

In baby terms, you're looking at (daily period) things like this?:



...to predict what will happen after this?:



...and maybe take a weighted average against periods of time that look similar (weighted by how similar they are)? Do you scale time (for example can a slow 2 month pattern match a fast volatile 2 week pattern) or volatility (100% variance matching 1% variance)? Do you hand pick similar periods, or does every conceivable period get weighted in (maybe x day durations, iterated every day)?

[iBug]

(Offtopic: Thanks a lot chodpaba (and Enky1974), your reports and analyses are really interesting and by far the most professional in this forum !
I would really like to call them the most reliable and accurate price predictions as well, but at this moment, nobody can really tell what's going to happen next week....
Anyway, keep doing your great analyses !)

[Chris Acheson]

There are high peaks below $1 because trading was at these levels for such a long time.

Wouldn't it make more sense to use log scale for the price groupings?  There was still plenty of volatility when the price was below $1.

[dserrano5]

Wow. Just wow . Keep up the good work!

I will update this chart periodically, it will be interesting to see how it changes over time.

A daily snapshot can become a neat animation after 2 or 3 months...