Bitcoin Forum

In case anyone else is interested.  I generated the following graphs using Google Insights search return for "bitcoin" for the last 53 weeks and MtGox weekly price data from bitcoincharts.com over the same time period.

The first graph plots the MtGox weekly close along with the Google Insight relative search indicator.  The second graph gives a better representation of the correlation.

From the top graph someone might say that bitcoins are are overpriced.  But looking at the bottom graph might give bulls incentive to pimp bitcoin as much as they can.  Here is a test.  Do you think there a correlation or not?

Pearson's r = 0.8817653

http://img846.imageshack.us/img846/1407/pricesearch.jpg

http://img835.imageshack.us/img835/3873/corcx.jpg

chodpadpa reported that the best correlation is between prices at day - 3 (that is price is leading trends by 3 days).  Personally I did not check this.

Correlation between what? That the price today is explained by the price 3 days ago?

Edit:

I could not find a user by the name of chodpadpa but I did some tests.  Finding the Pearson's correlation between today's price and the price 1, 2, 3, 5, 10, and 30 days prior.

minus 1 day ---->       r=0.9823598
minus 2 day ---->       r=0.9531210
minus 3 day ---->       r=0.9472155
minus 5 day ---->       r=0.9199642
minus 10 day -->        r=0.8384874
minus 30 day -->        r=0.5698149

So if I understand what you are saying correctly then no, the best correlation is between prices at day - 1, obviously.

If you mean the best correlation is between prices at day - 3 and the search index then I don't know.  Google only gives me data in weekly format.  I can do a week prior.

r = 0.9010368 versus my original data of r = 0.8817653, but there is no significant difference between the correlation strengths (p=0.64).

Between the prices and google trends.

what bitcoin needs is more offshore gambling

Bitcoin needs a lot more usability.

Bitcoin needs a reason. It's not time yet.

Google search trends do not take into account accumulated interest very accurately because those already using Bitcoin in one way or another rarely google "Bitcoin". This is why I think that it will become less accurate as an indicator as time goes on. I think it's already more inaccurate than it used to be. Not that I think it can be used to predict anything anyway, mostly it seems that the trends follow price, not the other way around.

My two favourite indicators are transaction count and mining cost per transaction. The former shows me what the actual usage of Bitcoin is (much more accurately than node count which is useless these days because server-based clients have become commonplace) and the latter seems to be a decent indicator of bubbles. I don't claim these indicators are good for predicting either but I think it's still good to keep an eye on them.

I'll link the 7-day avg versions of both because I think they are more useful. From the transaction count one can clearly see that network usage increased right after the Good Wife episode and has been on a strong uptrend to this date. Mining cost per transaction is almost at the level it was pre-bubble (the $7.2 bubble), which seems to indicate that as far price is concerned we're at the bottom or very near the bottom.

Transactions: http://blockchain.info/charts/n-transactions?showDataPoints=false&timespan=&show_header=true&daysAverageString=7&scale=0&address=

Mining cost per transaction: http://blockchain.info/charts/cost-per-transaction?showDataPoints=false&timespan=&show_header=true&daysAverageString=7&scale=0&address=

That is why I use insights.  It is smart enough to not only use bitcoin but also other bitcoin related searches.  If you check out the results or bitcoin (http://www.google.com/insights/search/#q=bitcoin&date=today%2012-m&cmpt=q) and download the csv file.  You can see that mtgox is number 15 and SR is number 17.

The only time trend really follows price is the current time period after October 2011, which I feel bitcoins will come down to about $4.00 each.

Bitcoin needs Pokerstars to support bitcoin payments.  Americans would love it.  The US government would surely find a way to sabotage Pokerstars however.

Send a request that they use it.

I've never even heard of Insights before this thread, thanks for that. Seems to be a lot better than Trends.

I certainly will. Now that PokerStars has an EU license it's tax free for me to play at PokerStars.eu, which means I will be playing there again. It's still part of the same network as everyone else so that's good. They have the best tournaments.

Your indicators are worthless, Technomage. They have nothing to say about price, and for no timeframe.

You remind me of bitcoinbull who predicted that 10$ was the bottom using historical Difficulty vs. Price ratios: https://bitcointalk.org/index.php?topic=7427.msg325241#msg325241

I guess history repeats.

--

Adding to the thread, the search trends are not useful for price projections or indications either, as they follow price, not lead. At least that is what I have experienced.

I also don't like the cost of mining charts.  They lag.  It is like the price of mining any commodity.  For example, the price of rare earth metals (http://cleantechnica.com/2011/12/15/chinas-rare-earth-monopoly/) is not based on the difficulty in mining them or their rarity around the world.  It is based on the demand and supply of them.  Since China has almost a monopoly on rare earth metal mining they control the supply.  As we need more rare earth metals for our gagets the demand increases.

The graphs I posted are just an attempt to measure demand.  Another interesting chart to measure demand would be some kind of velocity measure of money.

To measure supply I would suggest estimating how many bitcoins each of the large mining pools generate and what is the volume on the exchanges.

Is it legal to gamble on pokerstars if I live in the US?

You seem to have no content in your arguments. You've claimed many times how easy it is to fake transactions, yet you have provided 0 proof that someone is actually doing that in purpose. I don't see why someone would do that because that indicator is not generally thought of as very important. Next time have some proof when you mention this.

For me the transaction count is not worthless, it's the most relevant indicator that I know of. It's the most accurate universal indicator of how much Bitcoin is used. Price itself is very important but price is speculation, it tells us how much faith the market has in Bitcoin at this moment. Trading volume is fairly relevant as well and I see that as one indicator of how much usage Bitcoin has. Volume has been at high levels just like the transaction count.

The transaction count is best as a long term indicator and it's the main reason I'm not at all worried about any price drops. The only times we have been at prices of $1 or below are when transactions were at 2k or less per day, recently we have peaked over 10k per day and even the 7-day avg has gone over 8k. When we bottomed at $2 the transaction count was around 5k per day.

The logarithmic graphs are interesting as well:

Transactions log: http://blockchain.info/charts/n-transactions?showDataPoints=false&show_header=true&daysAverageString=7&timespan=&scale=1&address=

Price log: http://blockchain.info/charts/market-price?showDataPoints=false&show_header=true&daysAverageString=7&timespan=&scale=1&address=

I don't think it can be used for predicting anything but it gives some loose guidelines. It tells me that a price of $1 or even $2 makes no sense right now, it would be outside any correlation. Price between $3 and $5 makes a lot more sense however. One of my favourite numbers is also the market cap, which is always the number I end up thinking about when I need to make sense of where Bitcoin is at. Current market cap makes sense for what Bitcoin is right now but none of the potential is priced in. It will easily go much higher once something positive starts happening again.

Difficulty vs Price has little to do with the mining cost per transaction chart that I linked earlier. Please do look at the charts before you throw around irrelevant arguments.

Simply compare these charts to see the similarities:

Mining cost per transaction: http://blockchain.info/charts/cost-per-transaction?showDataPoints=false&timespan=&show_header=true&daysAverageString=7&scale=0&address=

Price: http://blockchain.info/charts/market-price?showDataPoints=false&timespan=&show_header=true&daysAverageString=7&scale=0&address=

Again I'm not saying this can be used to predict future price, it actually looks like these graphs change fairly simultaneously. The correlation is very strong. I wouldn't say it's worthless, it's interesting if nothing else.

Why don't the graphs have units on the x and y axis?

Currently, I believe it isn't (for real money).  Pokerstars could enable bitcoin payments and allow players to *wink* *wink* claim they're in Canada.  No need to send out cheques or get/send money from/to American credit cards.  I suspect however that PS would be afraid to touch a pseudo anonymous currency to bypass USA laws.

No they can´t. It would be easy to check for the IRS if they are doing so, also they would need to do proper KYC. But they could accept Bitcoin as a payment.

Hehe! Is that almost $23 of mining efforts (electrical cost) to confirm a single transaction back in June?  W :o W

Obviously, that would be because the number of txs were so low compared to the network computational power, but damn!