Pages:
Author

Topic: Monthly average USD/bitcoin price & trend - page 7. (Read 118255 times)

donator
Activity: 1722
Merit: 1036
February 26, 2014, 02:56:04 AM
Using rpietilas formula the trend price P for x = 1876 is

P = 10 ^ (0.003073 * 1876 - 2.909514) = 716.8594


Yes, I would very much like if everybody used my formula in this thread.

Otherwise casual readers cannot understand anything, nor trust anything.

The bolded figures are the parameters of the current trendline.
full member
Activity: 238
Merit: 100
February 25, 2014, 12:29:44 PM
Hello,

I did a similar chart in r.



If the price is carrying on like this we should touch the second standard deviation soon, around 420 usd ish.

For extracting the trend I’ve used the function stl(Seasonal Decomposition of Time Series by Loess) in R


Thank you , interesting chart. Could you please post the value of the standard deviation (sigma).
newbie
Activity: 48
Merit: 0
February 25, 2014, 09:01:21 AM
Hello,

I did a similar chart in r.

https://i.imgur.com/0qdXSfK.png

If the price is carrying on like this we should touch the second standard deviation soon, around 420 usd ish.

For extracting the trend I’ve used the function stl(Seasonal Decomposition of Time Series by Loess) in R

full member
Activity: 156
Merit: 100
February 24, 2014, 12:56:56 PM
it's a very interesting chart, could you please add more values (5% interval) on the y axis? thanx very interesting indeed.
newbie
Activity: 39
Merit: 0
February 23, 2014, 05:24:01 AM
it looks nice, but I didn't udnerstood a damn thing. can you eleborate on it a bit more? how "trend" line is calculated (is it just average of all prices? but it's percentages?)

Re: trend line, watch this: https://www.youtube.com/watch?v=LI0QYkv7--4

Given rpietilas current trend line you can calculate the trend price at x where x is the number of days since 2009-01-03.

You can calculate x from a given date, e.g. x from Feb 22 in Python:

Code:
>>> from datetime import date
>>> x = (date(2014, 2, 22) - date(2009, 1, 3)).days
>>> x
1876

Using rpietilas formula the trend price P for x = 1876 is

P = 10 ^ (0.003073 * 1876 - 2.909514) = 716.8594

On Bitstamp Feb 22 high H = 619.88. That is H / P = 87.09% of Feb 22 trend price.

What abs. value when L=0,3809 will be crossed?
What is these gray areas?

The absolute value of L = 0.3809 depends on x:

absv(L, x) = L * 10 ^ (0.003073 * x - 2.909514)

E.g. for Feb 22: 0.3809 * 10 ^ (0.003073 * 1876 - 2.909514) = 273.0518

The grey area is between highs and lows on a 3-day chart, the typical price is defined as (H + L + C) / 3. BitcoinWisdom.com 3d chart with "Settings" -> "Lines" yields a similar graph.

For me it's just an opportunity to draw funny graphs in R, don't read too much into it.
member
Activity: 89
Merit: 10
February 22, 2014, 09:08:41 AM
FWIW: This graph shows the Bitcoin price relative to the trend. It also contains quantiles of highs and lows since 2012. The price is between 38.09% and 213.45% of the trend price in 90% of all days since 2012.

More quantiles:

L:Q(0.1) = 0.4113907
L:Q(0.25) = 0.5348829

H:Q(0.75) = 1.198960
H:Q(0.9) = 1.709515

Median of the typical price T:Q(0.5) = 0.6922865
it looks nice, but I didn't udnerstood a damn thing. can you eleborate on it a bit more? how "trend" line is calculated (is it just average of all prices? but it's percentages?)


What abs. value when L=0,3809 will be crossed?
What is these gray areas?
newbie
Activity: 39
Merit: 0
February 22, 2014, 06:51:48 AM
FWIW: This graph shows the Bitcoin price relative to the trend. It also contains quantiles of highs and lows since 2012. The price is between 38.09% and 213.45% of the trend price in 90% of all days since 2012.

More quantiles:

L:Q(0.1) = 0.4113907
L:Q(0.25) = 0.5348829

H:Q(0.75) = 1.198960
H:Q(0.9) = 1.709515

Median of the typical price T:Q(0.5) = 0.6922865

https://i.imgur.com/U5xw3Ea.png
full member
Activity: 238
Merit: 100
February 21, 2014, 01:43:46 PM
I found this chart a while ago, don't remember where, but it is update daily. If you switch to the s-curve that will blow your mind even more.
https://docs.google.com/spreadsheet/ccc?key=0ArD8rjI3DD1WdFIzNDFMeEhVSzhwcEVXZDVzdVpGU2c#gid=1

impressive data in terms of how you collected avg. price points daily

though as we see gox is not much suitable for the price, perhaps it should be switched to stamp(or better start weighting price from some point, like November or someting, with decreasing role of Gox) ?

Just to clarify, it's not my chart and I'm not the one updating it, I just stumbled upon it in a tread somewhere, bookmarked it, and remembered it now and thought I should share it with you.
The logistics model was created by "SlipperSlope", he has another thread on this as well.
legendary
Activity: 2282
Merit: 1767
Cлaвa Укpaїнi!
February 21, 2014, 12:50:06 PM
I found this chart a while ago, don't remember where, but it is update daily. If you switch to the s-curve that will blow your mind even more.
https://docs.google.com/spreadsheet/ccc?key=0ArD8rjI3DD1WdFIzNDFMeEhVSzhwcEVXZDVzdVpGU2c#gid=1

impressive data in terms of how you collected avg. price points daily

though as we see gox is not much suitable for the price, perhaps it should be switched to stamp(or better start weighting price from some point, like November or someting, with decreasing role of Gox) ?

Just to clarify, it's not my chart and I'm not the one updating it, I just stumbled upon it in a tread somewhere, bookmarked it, and remembered it now and thought I should share it with you.
member
Activity: 89
Merit: 10
February 21, 2014, 11:12:12 AM
I found this chart a while ago, don't remember where, but it is update daily. If you switch to the s-curve that will blow your mind even more.
https://docs.google.com/spreadsheet/ccc?key=0ArD8rjI3DD1WdFIzNDFMeEhVSzhwcEVXZDVzdVpGU2c#gid=1

impressive data in terms of how you collected avg. price points daily

though as we see gox is not much suitable for the price, perhaps it should be switched to stamp(or better start weighting price from some point, like November or someting, with decreasing role of Gox) ?
legendary
Activity: 2282
Merit: 1767
Cлaвa Укpaїнi!
February 21, 2014, 06:02:45 AM
I found this chart a while ago, don't remember where, but it is update daily. If you switch to the s-curve that will blow your mind even more.
https://docs.google.com/spreadsheet/ccc?key=0ArD8rjI3DD1WdFIzNDFMeEhVSzhwcEVXZDVzdVpGU2c#gid=1
legendary
Activity: 1470
Merit: 1000
Want privacy? Use Monero!
February 21, 2014, 05:53:11 AM
Holy shit. That blows my mind - just the numbers involved and the speed with which it can happen - and it also shows how BTC has so many unique and unpredictable aspects vs. any "traditional" tech innovation/adoption/value cycle - even compared to another disruptive game-changer.  Shocked

The first time I saw the log chart my mind was also blown. I still like watching those charts Smiley

So far, the log trend is still active. Lets hope it will continue for at least a year:
10000 USD in march 2015 Smiley

newbie
Activity: 39
Merit: 0
February 20, 2014, 05:23:55 AM
I've plotted the new trendline with typical daily price for Bitstamp and MtGox starting from 2012.

thanks. can you change log. base for = 2 ?

The graph has lines for 2, 3, ..., 9, 10, 20, ..., 90, 100, 200, ..., 900, 1000, 2000, ..., 9000, 10000.

Why do you need base 2?

cause it's hard to manually count this 100,200,300...
I mean it's not hard but it's not handy.

not user friendly

I'm also a fan of base 2 log plots. I think it's the most practical for easily reading off precise amounts on a log plot.

I prefer the log10 grid. Anyways, here is the same version with a log2 grid: https://i.imgur.com/JeTTfUQ.png
legendary
Activity: 1008
Merit: 1000
February 19, 2014, 09:50:43 AM
I've plotted the new trendline with typical daily price for Bitstamp and MtGox starting from 2012.

thanks. can you change log. base for = 2 ?

The graph has lines for 2, 3, ..., 9, 10, 20, ..., 90, 100, 200, ..., 900, 1000, 2000, ..., 9000, 10000.

Why do you need base 2?

cause it's hard to manually count this 100,200,300...
I mean it's not hard but it's not handy.

not user friendly

I'm also a fan of base 2 log plots. I think it's the most practical for easily reading off precise amounts on a log plot.
member
Activity: 89
Merit: 10
February 19, 2014, 07:49:19 AM
I've plotted the new trendline with typical daily price for Bitstamp and MtGox starting from 2012.

thanks. can you change log. base for = 2 ?

The graph has lines for 2, 3, ..., 9, 10, 20, ..., 90, 100, 200, ..., 900, 1000, 2000, ..., 9000, 10000.

Why do you need base 2?

cause it's hard to manually count this 100,200,300...
I mean it's not hard but it's not handy.

not user friendly
newbie
Activity: 39
Merit: 0
February 19, 2014, 03:15:09 AM
I've plotted the new trendline with typical daily price for Bitstamp and MtGox starting from 2012.

thanks. can you change log. base for = 2 ?

The graph has lines for 2, 3, ..., 9, 10, 20, ..., 90, 100, 200, ..., 900, 1000, 2000, ..., 9000, 10000.

Why do you need base 2?
member
Activity: 89
Merit: 10
February 19, 2014, 12:35:51 AM
I've plotted the new trendline with typical daily price for Bitstamp and MtGox starting from 2012.

thanks. can you change log. base for = 2 ?
newbie
Activity: 39
Merit: 0
February 18, 2014, 04:47:31 PM
I've plotted the new trendline with typical daily price for Bitstamp and MtGox starting from 2012.

https://i.imgur.com/sPfQbIz.png
EDIT: update graph
legendary
Activity: 2282
Merit: 1767
Cлaвa Укpaїнi!
February 14, 2014, 12:07:44 PM
Thank you Smiley

Next question is - what will happen next?

A) Bounce from the trendline towards new highs
B) Something else (what?)

B) If it follows the pattern from earlier bumps it will stabilize and we will have a period of perhaps six months or so with little movement and then another giant bumb.
member
Activity: 89
Merit: 10
February 14, 2014, 09:30:50 AM
please use log. base =2 for the graphs, it's almost impossible to differ values from 100 to 1000 on such scale.

see my example (with my own random assumptions for dec & jan since it's not published averages)
more lines  = better understanding

Pages:
Jump to: