Hacker News Reader

Comment by paul | Hacker News Reader

Is anyone tracking how much money miners are pulling out of bitcoin on a daily basis? I assume they have to sell some fraction of their earnings in order to pay for hardware and power. This should create a constant downward pressure on the price, since there always has to be new money flowing into the system, but very few people "need" to buy bitcoin in the way that miners "need" to sell (most buying is for speculative purposes).

This is almost the opposite of public companies, which rarely issue new shares, and have a constant buy pressure coming from 401k plans and other automatic investment vehicles.

View on HN

If the market were efficient, it would already price in the dilution from newly minted coins, as it does for share dilution in the stock market (which is actually quite common due to employee stock options). However, every indication is that the bitcoin market is not at all efficient. It is largely driven by speculators playing a greater fool game, not investors who are betting on intrinsic value per bitcoin based on a future estimate of coins outstanding. The result has been a series of manias and panics, which will continue until real buyers enter the market.

An alternative view: if BTC as a major currency is worth $100,000, and BTC that isn't a major currency is worth $0, the market price will be roughly $1,000 if BTC has a 99% chance of failure (ignoring the risk premium). As in reality the odds that BTC will become a major currency are difficult to estimate, volatile, and likely to be affected by the price of BTC itself, volatile prices for BTC, which we wrongly interpret as "manias" and "panics," actually tell us very little about the efficiency or inefficiency of the market.

It's more than manias and panics (though they played a large part). The legal and institutional framework surrounding it is still in constant flux, and its future value will depend heavily upon that.

Right now the institutional news all looks negative -- e.g. the failure of mt gox and the unfavorable tax treatment in Japan (and now the US).

3600 coins are currently added to the system per day. There are currently ~12.6M coins in circulation.

So that's an increase of .02% per day.

This doesn't come close to accounting for the price swings we've seen in the last 6 months. There are much larger forces at work. What these forces are exactly is...difficult to determine.

Percentage of total coins isn't the correct comparison, it's the percentage of total volume. We're on a pretty high volume day right now and the 24hr volume is at 150k. Some days it's a fraction of that, there have been multiple < 20k days recently. Depending on the day the mined coins are a significant percentage of the total being exchanged (not that all mined coins would be exchanged, but it's a fairly high number otherwise most people wouldn't have any BTC unless they mined it).

Hrm, interesting. I'm now deeply uncertain whether % of coins or % of volume is more important. I will have to think about this more.

but once the volume has been exchanged, it's already too late for most people so it would probably be better to find the cause of high % of volume?

Regarding price, percentage of volume is the important metric. It is not relevant that ~12M bitcoins exist, if only, say, 1% of these are for sale on exchanges. Only coins for sale on exchanges help stabilize the price.

But what exactly does "for sale on exchanges" mean? Presumably a great many coins that didn't trade hands yesterday are still for sale at some price.

>> There are much larger forces at work. What these forces are exactly is...difficult to determine.

How about: "lots of people who bought at the top of the BTC hype are now abandoning the BTC market because they lost faith in their get-rich-quick scheme?"

The meteoric rise and slow but steady deflation in the BTC price you see now has all the properties of a classic, media-fueled bubble. At one point people who were never interested in monetary systems or cryptocurrencies at all, were asking me how they could buy BTC because they heard on the news it went from $100 to $500 in a few weeks. These are not idealists or miners, but just hoped to ride the hype train up and make a few quick bucks without actually having to work for it.

I don't think it's 'difficult to determine' why it's fading now. Maybe if you completely live in a kind of echo-chamber that HN appears to be on topics like this, but looking from the outside it all makes perfect sense.

Ya, I mostly agree with you. The devil is in the details though. Why is it fading at the rate it is? Why did it drop to ~350 but then bump back up to ~425? It's seemingly pretty random (at least to me).

The short answer is that miners are hoarders, and this is based on a few things:

1. In theory difficulty and hash rate should follow the bitcoin price, but they don't. You'd think that as price decreases difficulty re-adjusts as miners pull out but difficulty and hash rate continue to increase and are completely disconnected from the price. Charts:

Difficulty:

https://blockchain.info/charts/difficulty

Hashrate:

https://blockchain.info/charts/hash-rate

Price:

https://blockchain.info/charts/market-price

Because of the disconnect, you can conclude that most mining business models are based on the future price of bitcoin. The miners are either true believers in the future potential, or they are effectively stuck because they have sunk upfront real USD but are currently mining at a loss and have no choice.

2. Most mining at the moment is being done at a loss. This continues from point 1, but if you look at a mining calculator your can work backwards and calculate at which electricity price point mining is currently profitable at.

As an input for our calculation, lets assume all miners are running the equivalent of a KnC Neptune, a new 20nm-based board that hasn't been released yet but it due out in a month or two (hah!):

https://bitcointalk.org/index.php?topic=347597.0

We will normalize their figures based on price per TH hashrate, and price per GW output. They promise 3TH for $10k, and they promise 2,300W.

So for each TH it is $3.3k upfront, and 733kw/h to power that TH.

Plug those figures into a calculator:

http://www.bitcoinx.com/profit/

Your gear is generating $32 per TH per day, so $96. To pay off only your hardware would take 104 days, this is with 0 electricity cost. You'd need to find electricity at a few cents per kw to break even[0]. The sums aren't good.

[0] This is possible in only a couple of places, mostly scandinavian nations with geothermal electricity supply or nations where electricity is subsidized like China and India. Even at a low 8c/kw your hardware (which is being delivered) is paying itself off in 118 days at todays difficulty rates (which won't hold).

3. A big problem with mining is that your investment decision is inelastic (not sure if that is the right term). You decide to invest at todays difficulty and bitcoin rate but outside of cloud-based mining you don't actually get started mining after placing your order for at least a month, often 5-6 months. So that means when there are little openings in the window for mining profitability everybody piles on, and you find when you get up and running that your mining estimates are an order of magnitude off because a lot of other people had the same bright idea.

The hashrate and difficulty charts show that with their increases. If you plotted the announcements of new generation mining gear against those charts in the same way Google News does you'd likely find that each big spike is equal to in time as new_mining_gear_announcement_date + new_mining_grear_delivery_date

4. Newly mined coins can't be transferred for at least 100 confirmations. This was an impromptu patch applied by the bitcoin project and a number of markets to defeat some sort of bug, but I can't recall or can't seem to find the details at the moment and can't recall what it was fixing or if the 100 number is correct. If anybody knows the details of this, i'd be interested in finding it again.

So when looking at the blockchain for newly minted coins, you'll never see them move immediately, there would be an at least 100 transaction delay (which isn't long in real time terms).

5. On non-volatile days the markets combined trade around 15M coins. On highly volatile days like today the combined volume exceeds 120M coins. 3,600 new coins are mined per day, so their input ranges from 0.01% of coins traded to 0.002%.

Note that this is traded volume, rather than percentage of coins, and the low trading days are often only a few million transactions per day.

6. You can follow the newly mined coins on the Blockchain. Ghash.io is anywhere from 35-50% of the total hash rate (which in itself is controversial) and their newly generated coins go to this address:

https://blockchain.info/address/1CjPR7Z5ZSyWk6WtXvSFgkptmpoi...

You can see that most of the coins remain unspent, and GHash.io is a unique case where you'd expect more of the coins would be spent, which i'll explain in my next point.

7. The hash rate and return itself has now become commoditized and can be traded like a derivative. cex.io was a pioneer here and they have an active market for buying and selling GH/s mining. Their pool is the GHash.io pool mentioned above.

Their current price for a GH/s of mining hashrate is 0.01058690 BTC, or around $4.03 USD. If you compare to the yet unreleased Neptune above, it is $4030 per TH/s vs the Neptunes $3,333 per TH with no electricity costs - and you can move in/out as you wish.

I have previously profitably traded on cex.io using an automated trading bot. It hasn't been active in a long time because I haven't updated it to suit the new conditions.

Here is my referral link to signup to cex.io, I get 10% of whatever you purchase:

https://cex.io/r/0/nikcub/0/

Here is the base open source version of my bot, which doesn't have any of my strategies built in it, but which can be extended:

https://github.com/nikcub/cexbot

I'll likely bump the version currently there with my new strategy interface in the next few weeks since I believe there are new opportunities in GH trading again on CEX. If you're interested in trading strategies, get in touch. The most basic mechanism in that bot will take your mining return and pour it back into purchasing GH/s so that your returns are compounded.

A lot of the return you get in CEX is from the other coins that are mined alongside Bitcoin when you purchase GH/s.

Note that nothing in the bitcoin world really makes sense, so the rules from normal public stock markets or strategies that might work there don't apply in the bitcoin world. You can see from the disconnect between hashrate, difficulty and price that a large part of Bitcoin price and market movement is speculation and emotion driven. You can still take advantage of this, though - as a lot of the price is driven by news events and emotional reactions to them.

Since it is so easy to move in/out of mining on CEX, you'd assume their mined coins are more liquid than the miners who setup long-term projects and are looking for a return based on future bitcoin price, but even at the CEX.io address you see only a small number of coins being moved.

8. In theory it would be possible to go back through the blockchain and measure/quantify what percentage of newly mined coins in a certain time period are being spent. Anecdotally i'd estimate the percentage is very low, and that most coins in circulation are greater than a year old (eg. i'd estimate 99.5% of traded coins were mined +1 year ago).

tl;dr: the current mining business model is based on future bitcoin price, most newly mined coins are held.

A masterful analysis - I learned more about the starte of the Bitcoin universe from this than I have from last 100 or so articles I've read. If you're not doing professional investor analysis, you should be.

Unfortunately, the claim that "Most mining at the moment is being done at a loss." is unsubstantiated and highly debatable. It's nearly impossible to establish the efficiency and USD-production-cost of mining technology being operated in secret - say, coincidentally nearby a 14nm Intel fab.

That was a brilliant write up thank you.

The hole in your theory is that a rational miner may think the price will multiply by 100 (for example), so buying $1000 mining equipment will be profitable because it will yield $700 worth of BTC (70% figure estimated from nothing and is really an arbitrary number as long as it is less than 100%), which will eventually be $70,000. The problem is that buying $1000 worth of bitcoins would yeild $100,000 in a $1000 investment rather than $70,000 in a $1000 investment.

The real reasons people continue to mine are one or more of these: -It is fun -It secures the network -They wrongly think they will make a profit

That's not rational though. There's been no rational reason to think bitcoin can yield those returns. You're talking about a 10000% return on investment.

There is no investment in human history that works like that.

> There is no investment in human history that works like that.

Satoshi may disagree with you

> You're talking about a 10000% return on investment. > There is no investment in human history that works like that.

Sure there is. Just none that is not also extremely likely to fail.

The obvious example would be a lottery ticket.

ummm, well bitcoin has done that. At one point you could get a thousand bitcoins for a $1 then it was 1 bitcoin for a $1000. So a return like that is possible and it has been possible in the last 4 years.

I am not saying it is rational I am just saying you are wrong :)

The whole VC industry is based around the idea that there are 10000% return on investment deals to be made..

Bitcoin has gone up 1000% every year, and 10,000% last year.

Also, every successful startup ever has had those kinds of returns at some point. So I suspect you are only considering traditional or stable investments, like post IPO stock or whatever.

Berkshire Hathaway has returned about 800,000% since inception.

Oracle has returned around 80,000% since IPO. Dell, Cisco, Microsoft, AOL, numerous others all pulled a greater than 10,000% return post IPO.

Las Vegas Sands managed around a 7,000% return from the 2009 bottom to recent top.

What do you suppose Andy Bechtolsheim's $100k check / investment into Google returned?

There are in fact plenty of examples in recent history of extraordinary returns on par with or far exceeding that 10,000% level.

I really don't get why you would invest in mining equipment. If your "mining business model is based on future bitcoin price", it makes a lot more sense to just buy some [BTC] and hold on to them ?

(Edited for clarity)

>A big problem with mining is that your investment decision is inelastic (not sure if that is the right term).

I think illiquid is the term.

You are using the wrong units units for power and energy. Neither "kw/h" or "$ per kw" make any sense in the places you put them. I think you mean "kw" and "$ per kwh" respectively.

kw=kilowatt, a unit of power; kwh=kilowatt-hour, a unit of energy

Electricity is billed per unit of energy.

> 1. In theory difficulty and hash rate should follow the bitcoin price, but they don't.

This ignores efficiency increases in mining hardware. We've basically seen an increase in efficiency of at least 10,000 times, over the past years, from CPU mining (~5000W/GH) to ASIC mining (~0.2W/GH). This completely disconnects the Bitcoin price from mining, and this will continue to happen, as developing more efficient mining chips won't stop, although the rate of efficiency increases will decrease.

So difficulty and price will never correlate, because mining keeps getting more and more efficient.

the other issue i keep thinking about related to miners are the people who sank millions into mining rigs. if you look at the hash rate / difficulty of finding new blocks, compared with the cost of the equipment and time required, seems like the incentive for mining is going to drop pretty significantly.

so perhaps they hold onto their supply, instead of selling at what they think is an artificially low price. but perhaps they panic and decide to unload - prices falls further. either way, the calculation that got them started is much less attractive.

plus, if the incentive to mine drops, the processing power of the network drops, could eventually increase the likelihood of a hack.

And that, to me, is the strangest thing about Bitcoin:

https://blockchain.info/charts/hash-rate

The miners, never, stop. Completely disconnected from the market pricing.

You assume they're all paying for the power. Once you remove that assumption, it makes more sense.

The reason is much more Machiavellian, namely if you keep the price of entry high and you loose money to mine for 80% of the time then the 20% of the time that it's extremely profitable you have a bigger multiple since the competition is weaker then it would've been otherwise.

is mining competitive, though?

Absolutely. A fixed reward (25 BTC) is rewarded each time a block is mined. Each block is based on the prior blocks in the chain, so when one miner mines a block, all the other miners must start over. Blocks are mined on average once every ten minutes, so basically, a random miner is selected approximately once every ten minutes to receive a reward.

Saying they have to start over is implying that you're somehow making progress. The only progress your are making is just ruling out a few inputs from a massive search space. You don't really lose any work when it resets because the work you've done is useless.

I understand the semantic difference, but I'm not sure the distinction matters in any way. The implication that there is progress is tenuous at best.

Of course they lose work. They have to pay the set-up cost of bundling a new set of transactions for the next block.

Currently, mining is still profitable so it wouldn't make sense for them to stop yet. Once Bitcoin dips below a certain value (around $100 I believe?) then we will have to watch for the drop off.

They will only stop once the electricity is more expensive than the mined bitcoins. We're not there yet.

Many won't even stop then because of the belief that the price can/will return to previous levels if not higher. Quite the interesting dynamic.

but that is effectively irrational isn't it?

if mining 1 bitcoin costs you 500 dollars and they are sold on the market at 400 why don't you switch off your gear for a while and just buy them ?

Maybe the point is people are merge-mining multiple bitcoins, or the electricity bill for them is way lower than we think it is.

maybe they don't have the money to buy them because it's been all spent for mining gear?

Then how can they afford the electricity?

They will earn more coins by paying the electricity and using the already bought mining gear, than if they stop paying for the electricity and use that money to buy coins.

Not if buying the coins is cheaper than mining them, which is what the GP is saying.

that would require a $500 dollar cost of electricity per coin, however in reality the $500 is made up of the investment in the hardware and electricity. You could not just switch off the machine and have $500 to spend on coins, you would need to sell the hardware. of course in most cases selling the hardware for USD and spedning it on BTC will generate more BTC than the hardware will mine.

There's also the question of how much of the electricity is paid for. E.g. how many miners are mining on "free" electricity at their parents house, or have access to rack space with "free" electricity (our racks at work come with a certain amount of power, for example; someone putting a Bitcoin mining rig in there would not increase our costs as long as the power doesn't exceed certain limits), or has a miner in a quiet corner at work or similar?

Or any number of other ways of getting free / outright stealing electricity.

Pretty humorous how monetary cranks rant nonstop about the fed printing money being theft and about how we're turning into Zimbabwe and then goddarn it Paul Krugman was right and there's no hyperinflation and to top it all off your Ayn Rand bucks start suffering "downward pressure" due to all the Ayn Rand buck printing...

Krugman was right, but I remember during the Bush years, he was predicting inflation caused by the deficits that Bush created. He's a really smart guy but can be pretty partisan.

That said, at least he adjusted his opinion in the face of evidence -- Econ 101 says we should be seeing inflation and instead we're seeing near-deflation while the Fed is literally printing as much money as they can politically get away with. The monetary cranks keep telling us we're going to see it 'any minute now', because they can't abandon their theories in the face of evidence -- and yes, eventually some inflation will happen at some point, but it's hard to say that'll have made the cranks correct.

I'm not here to canonize Krugman, in fact I'll link the sort of article you're referring to:

http://www.nytimes.com/2003/03/11/opinion/a-fiscal-train-wre...

I will say that deficits increasing interest rates in a time of expansion and deficits and QE not increasing interest rates during a recession when the taylor rule is at -5% or so[1] are not exactly the same thing.

[1] http://www.newyorkfed.org/newsevents/speeches/2012/dud0524_1...

That's exactly the column I was thinking of, I specifically remember the mortgage reference. Thanks.

And I agree on your distinction there, and the historical record bears it out, with inflation rising from 2003-2007 and then near-deflation from 2008 onwards. Thank god we didn't listen to the fiscal hawks in 2009, most of whom were just trying to score political points and make things harder for Obama anyways. Could have been a deflationary spiral similar to what happened to 2000s Japan or 1930s US.

Hm, I think I just bumped Krugman up another notch in my head -- that 2003 article looks a lot less partisan if you look at it in the context of 2003-2008 rather than 2008-2014.

I believe that at this point the amount of bitcoin being mined is so small in comparison to the amount that is out there, that this should be a negligible effect.

Nopes. Bitcoin's yearly inflation is about 1 million BTC for this year and next few years. So to keep the current price stable, almost 350 million (as of this writing) of new money has to move in BTC.

That's assuming everyone who mines coins puts them up for sale and that's far from true.

It assumes they will be for sale eventually, which is true. At that time supply will increase and price will therefore decrease unless ongoing growth is experienced.

He said "to keep the current price stable". Stable implies not changing over time starting from present day right? Or did you mean starting "eventually" that much money will need to enter the market to keep the price "eventually" stable? If so, I don't know what the relevance is of how many coins are being mined this year or next. If you're just talking about eventualities with no timeframe, then you're not saying anything quantifiable at all.

Someone please correct me if I'm wrong, but looking at hardware prices, might also support the case that "miners are hoarders".

(Please excuse the sketchy math in the following post. these figures are just based on searches on Ebay/Google. Most of all I don't know who's doing the majority of the mining and at what scale.)*

(Low End) ($490 investment in mining) At the current level of mining difficulty a machine that has a hash rate in the 200/g/. Returns about $6.00 worth of BTC a day. http://www.alloscomp.com/bitcoin/calculator.

I searched "200 gh/s" in a site that crawls Ebay, The machines for sale at this hash-rate costs about 490. You would recoup the cost of the machine in 81 days.

http://www.thepricegeek.com/results/200+gh%2Fs?country=us

If it takes 81 days to break even on your investment, even if bitcoin doubles in that time frame you basically made the price of the machine as profit ($490 in 81 days). (meh)

(High End) (18k investment in mining)

Machines in the single digit tera-hash realm come in at under 20k, they peak at 15-20K per machine. Let's average the price at 18k. They make $90 a day according to the calculator. This would mean you would need 200 days to recuperate the cost of the machine.

In this case it takes 200 days to break even on your investment. If bitcoin doubles in that time frame you made 18k in 200 days. (the online calculator said power costs in this time frame are around $70.)

I'm assuming individuals in both these positions are have long-term holding in mind. My rationale is: getting substantially richer in either case requires the price per coin becomes drastically higher. You're also not hemorrhaging money in either case (power costs seem reasonable). It makes more sense for miners to hold as much of what they mine as possible. Both positions seem to hedged on a BTC being priced a lot more higher, in this sense hard to imagine miners to be selling in great quantities.

I also feel that people DIRECTLY buying and selling bitcoin would dwarf the selling pressure of the majority of miners. The barriers of entry for this also seem much lower than setting up the hardware.

personally I have been mining Litecoins for 6 months and am just about ready to turn off my rig. Electricity cost vs. difficulty level isn't make it worth it anymore. For the record, I have not sold and don't plan on selling any of my LTC. I will be holding onto them until virtual currency becomes as mainstream as is portrayed on "Almost Human".

Another theory is that big players were buying in 2013 and the same big players are selling in 2014.

Another theory is there's no liquidity so the true price can't be determined.

> This should create a constant downward pressure on the price

You need also to take demand into account.

Slowly plunging demand.

Not that different from how paper currencies like the US dollar work in terms of money supply though.

Of course, Bitcoin is not issued by a QUANGO whose chair appointing government has 10 aircraft carriers, over 5000 nuclear missiles and so forth. Anyone can start up a Bitcoin like currency (Litecoin, Dogecoin etc.)

People overestimate the monetary value of military hardware. The Russians had a pretty powerful military when the Ruble collapsed. When people lose faith in the dollar no amount of firepower will save it.

The exchanges have some good numbers about the flows of dollars. I haven't seen where any of them are sharing.

Reply on news.ycombinator.com