Author

Topic: Google Trends vs Mt Gox USD: Graphs (Read 4693 times)

sr. member
Activity: 420
Merit: 250
July 09, 2011, 12:33:09 AM
#17


Looks like we're headed back down to JPMorgan parity. Not too bad.

haha - oh man - back to slumming it up with the Morgan folk.
full member
Activity: 196
Merit: 101
July 09, 2011, 12:25:25 AM
#16


Looks like we're headed back down to JPMorgan parity. Not too bad.
sr. member
Activity: 277
Merit: 250
July 09, 2011, 12:14:03 AM
#15
If that timescale is correct, than you are completely wrong. Google trends follow every time there is a big spike or low in bitcoin prices. People talk about how much they won or lost in their zany gamble game.

full member
Activity: 182
Merit: 100
July 08, 2011, 11:44:29 PM
#14
looks like a recent uptick in the search volume in google.  Lets see if the price will do the same. lol
jr. member
Activity: 56
Merit: 1
June 30, 2011, 05:15:15 PM
#13
Since this thread got revived, I've included a more recent 30 day chart in the original post.

PS, google trends is only delayed about two days.
donator
Activity: 714
Merit: 510
Preaching the gospel of Satoshi
June 30, 2011, 03:53:11 PM
#12
> Also the fun part is that the mere popular belief that Google trends are indicative of a future increase (or decrease) of prices might actually be a self-fulfilling prophecy.

Except for the part that their stats delayed, roughly for 2 weeks or so whatever was trending we would know about it before it shows up in their trend graphs.

Good point, I actually never used Google Trends until now.
So Google trends is quite useless then.
donator
Activity: 714
Merit: 510
Preaching the gospel of Satoshi
June 30, 2011, 03:51:50 PM
#11
Correlation is not causation
Don't bring empty phrases from Statistics 101.
And don't jump into conclusions, please read and interpret what I wrote.
I never said that google trends are causal for the increase of the prices, in fact, what I wrote it implies that we might never know the real cause of the increase of prices, and yet IF it is predictive it is useful.

Correlation is correlation, period.
And if the correlation is predictive, it doesn't matter if it causal or not.

What simple minded people see is A -> B
What most educated people see is: A <-> B
What I always consider is A <-> X <-> B
(Or even worse: Y <-> A <-> X <-> B)

The X-Y factor might be unknowable but what matters is the end result of B.
If X-Y is moderately constant, the correlation between A and B might seem constant, and sometimes X is too complex to understand or to be perceived.

Anyway, if whenever happens A later happens B, or viceversa, if whenever happens B later happens A, it doesn't matter if we know the real underlying causes of it for our purposes.
Whenever happens A I'll be ready for B.

zef
member
Activity: 90
Merit: 10
June 30, 2011, 03:07:27 PM
#10
From when i learned about bitcoins, it took about a week and a half before i tried mining, and probably 3 weeks before I purchased any coin through the market.  I would consider that a very fast turn around time for someone going from knowing nothing about bitcoin to investing in it.  I spent most of my time learning about the technology, asking questions, and gathering opinions before I actually jumped in. I would not expect people to act that fast, particularly if they are not tech savvy.
legendary
Activity: 1050
Merit: 1000
June 30, 2011, 02:38:26 PM
#9
> Also the fun part is that the mere popular belief that Google trends are indicative of a future increase (or decrease) of prices might actually be a self-fulfilling prophecy.

Except for the part that their stats delayed, roughly for 2 weeks or so whatever was trending we would know about it before it shows up in their trend graphs.
hero member
Activity: 630
Merit: 500
Posts: 69
June 30, 2011, 02:27:34 PM
#7
With Bitcoin kind of only existing on the same land as Google, this is kind of a neat.
member
Activity: 112
Merit: 10
June 30, 2011, 02:20:00 PM
#6
Correlation is not causation
donator
Activity: 714
Merit: 510
Preaching the gospel of Satoshi
June 30, 2011, 02:08:42 PM
#5
Essentially you have discovered that:

The more people that invest in a commodity, the greater its value? And its reverse is true as well.

The majority of Google trends hits will not lead to investment.

A lot of people are following the bitcoin phenomena purely out of curiosity. I think your observation is interesting, but definitely an oversimplification.



It's not an oversimplification.
The correlation is undeniable, now the question is if it has predictive power.

We might not really know the underlying mechanism of what drives them to purchase, but knowing that the ultimate consequence is an increase of prices it might be useful.
Also the fun part is that the mere popular belief that Google trends are indicative of a future increase (or decrease) of prices might actually be a self-fulfilling prophecy.
legendary
Activity: 1694
Merit: 1006
June 15, 2011, 11:46:57 AM
#4
Essentially you have discovered that:

The more people that invest in a commodity, the greater its value? And its reverse is true as well.

The majority of Google trends hits will not lead to investment.

A lot of people are following the bitcoin phenomena purely out of curiosity. I think your observation is interesting, but definitely an oversimplification.

full member
Activity: 224
Merit: 100
June 15, 2011, 11:03:24 AM
#3
You probably should shift Google Trends to the right by about 5 days for a better correlation with price, since the time to get setup with Dwolla and to fund a Mt. Gox account is about 5-7 days
sr. member
Activity: 504
Merit: 250
June 15, 2011, 10:44:03 AM
#2
This shouldn't be a surprise if you understand how a speculative bubble inflates.
jr. member
Activity: 56
Merit: 1
June 15, 2011, 08:58:33 AM
#1
UPDATE:

A more recent 30 day chart, you can see that the trend line I drew earlier was broken, and there hasn't been much upward movement, it still matches price very well. If movement continues downward, there is a case for a lower price:





I think that popularity is the only factor driving bitcoin prices. To drive the point home I put together a couple of charts juxtaposing google trends and Mt Gox USD prices.




The blue line represents the google trends graph, scaled to match. I find it interesting that a dip in popularity in mid may seems to have lead directly to a dip in prices. Looking at the year graph, you can see that the google trends line (which isn't fully updated to the present), is about to cut into the price line deeply. Looking at the last month, you can see that for the first time in these graphs the price exceed popularity quite a bit. This seems to have led to a quick over correction and then a bounce back.

In the month graph the google line is once again not up to date, but the pink line gives a generous trend. If the pink line were accurate, this would mean we are priced correctly at the moment, since at the scale I am using, google trends has historically been slightly above the USD price.

As always, if you found these graphs useful, please send a tiny tip to the address below. Using the newest version of the client you can get transaction fees down to a super tiny sliver.
Jump to: