Fire in the Hole!

Welcome to my third post in a series about bitcoin mining. In previous posts, I compared mining to buying bitcoin as an investment, and gave details about how to estimate mining revenue. This time, we'll look at operating expenses and the mining bottom line. Bitcoin mining is a business, and to manage a business you need to understand your numbers. That means (ominous chords) accounting. A small crypto mine has only a few accounts to track, complicated only by the fact that revenue comes in crypto, but expenses go out in fiat currency.

Fantasy Bitcoin Mining

You got your mine up and running, burning electricity and producing coin. What now? Let's back up a bit and look at how you got here, based on a completely made-up example.

A grown man wearing brown shorts brought Chad a box containing a brand-new Acme HashSettler 454 ASIC mining machine, including power supply, cables, and helpful DVD. Chad paid $1000 for the HashSettler, which boasts a 4.54 TH/s hashrate and uses 550 W. Electricity in Chad's area costs $0.12/kWhr. Burning 0.55kW for 24 hours therefore costs $1.584, payable only in USD. Chad might incur some additional minor expenses, but other than an out-of-warranty HashSettler repair bill, anything else is in the noise compared to $47/month for power. Chad paid for the HashSettler on a day when bitcoin cost $10,000; he could have bought 0.1 BTC instead.

Chad fired up the 454 and instantly noticed it sounds like a shop vacuum, and blows warmish air around the room. He relocated it to the basement and let it roar. Good thing he doesn't live in an un-air-conditioned studio apartment.

Having done his homework, Chad signed up at SharkPool to combine his 4.54 TH/s with their thousands of other members' mining power. SharkPool charges no percentage fee for their service, but they keep all block transaction fees. Chad therefore estimated that SharkPool would pay out based on the standard mining formula,

R = B H / D / (49710 hash/s) = 0.00114162 / D,

where R = daily gross revenue in BTC,
B = 12.5 BTC = the block reward,
H = 4.54 TH/s = the mine's hashrate, and
D = the current mining difficulty in trillions.

Chad happened to start the HashSettler the day of a biweekly difficulty adjustment, and started off at D = 1.408 trillion, on its way to 1.591 trillion. Watching his payouts over the next week or so, he observed daily payouts within a few percentage points of 0.00071760 BTC, as estimated for the new difficulty. To eliminate the leading zeroes, let's use satoshi instead of bitcoin (100 million satoshi per bitcoin); Chad's initial daily payout was 81,082 satoshi. The outlook on Day 1 was therefore gross revenue of 81,082 satoshi per day at $13,541/BTC, or $10.98/day. After paying $1.58 for power, he would net $9.40/day. That suggested it would take 106 days to recover Chad's $1000 equipment cost (assuming no change in exchange rate or difficulty!).

Paydirt, At Last!

Over the next two weeks, Chad logs the following:

The table shows the USD exchange rate, difficulty, and revenue, operating expenses, and net earnings in both satoshi and dollars. In the first two weeks of operations, the mine produced a net of 865,462 satoshi with a value of $153.51 at the close of Day 14. As expected, the difficulty (and consequently the satoshi revenue) remained constant for most of the two week period. On Day 13, the difficulty increased again to 1.873 trillion (the Day 13 value represents an average difficulty due to part of the day at the old figure). The increased difficulty drove the gross and net satoshi earnings down, and the fiat results followed. That happened because the exchange rate fell as the difficulty rose. The key factor driving fiat earnings, exchange rate divided by difficulty (E/D in the previous post), decreased. Note the opposite behavior between Days 1 and 2.

The table also illustrates one of mining’s key pitfalls: the fact that the mine produces bitcoin, but pays operating expenses in fiat currency. Even if a mine produces bitcoin at a constant rate, and pays expenses at a constant rate of fiat currency, its net earnings (measured in either bitcoin or fiat) fluctuate with the exchange rate. To make sense of the results, you have to look at them either in crypto or fiat, not a mix of both.

To summarize the first two mining weeks, the mine netted .00865462 BTC after paying for the electricity to extract them. That looks like $154 of income, but Chad spent $1000 to buy the mine. The mine has a way to go before becoming profitable, but remember on Day 1 we predicted 106 days to reach breakeven.

One Year in the Bitcoin Mine

The price and difficulty data in the table were downloaded directly from blockchain.info/stats, for the period 25 Nov through 8 Dec 2017. To run the model for a full year, I created price and difficulty data series as follows: Starting with 71 days of actual performance (25 Nov 2017 through 3 Feb 2018, setting Day 1 at 6 Dec), I simply repeated the price series to get 365 days' worth. That provided a sequence of up and down markets, with no overall up- or down-trend and a totally unrealistic periodicity. For the difficulty, I started with the same 71-day period of actual data, then increased the difficulty 5% every two weeks until it doubled to 5.206 trillion. I reasoned that although the difficulty seems to trend ever upward currently, at some point declining mining production will discourage further investment in additional network hashrate. I arbitrarily chose twice the current difficulty as the trigger for that. None of this is meant as a prediction of what I believe will happen in the future. I have no idea. It's just fake data assembled for illustrative purposes.

Here's a chart showing the fantasy mine model results for 365 days. The dark blue and red curves show the exchange rate and difficulty inputs. Note in the legend that the various data sets were scaled by factors of ten to show clearly on the 0 – 20,000 y-axis. The mine did produce bitcoin (light blue) in both up and down markets, as predicted, although the increasing difficulty drove the rate down from over 69,000 satoshi/day to below 4,000 satoshi/day. The yellow curve for accumulated bitcoin also shows this trend in the way it gradually levels off, ending the year at 7.64 million satoshi (0.0764 bitcoin). Although the mine failed to accumulate within a year the 0.1 BTC Chad could have bought with $1000, dramatic price increases did allow it to reach dollar breakeven around day 220 (green).

It's Only a Simulation

Chad's story, of course, samples only a tiny slice of the outcomes possible in the real world.
As with all simulations, different inputs give different results. For example, a more optimistic prediction of difficulty would give a faster breakeven, while a more pessimistic outlook would easily force a mine shutdown (light blue curve touching zero). The model does provide insight into the interacting factors affecting mining profits. Unlike investment forecasting, predicting mining performance requires a crystal ball capable of foreseeing both the exchange rate and the mining difficulty many months ahead. Perhaps you hold a rock-solid conviction that bitcoin will double in value this year. Even if events prove you correct, mining will disappoint if the difficulty triples. It could also happen that the price triples while the difficulty doubles, which would make for awesome mining.

Finally, although I cast Chad's tale as a hardware mine, I used cost figures from real-life cloud-mining ($220 s/TH for the cost of the mine and $0.35 s/TH-day operating cost). Real mining hardware is somewhat more energy efficient than the HashSettler 454, but might cost more. Chad's HashSettler will faithfully continue to mine on Day 366, but cloud contracts expire. You can easily add capacity to a cloud mine by sending more money. Most homes lack the electrical wiring to run more than a few ASIC miners. If you run more than a few machines, your projections have to include the cost of cooling them. Expanding beyond one's home entails additional costs (rent, for example). Thanks for reading!

Previous post: So You Want to Start a Bitcoin Mine

Next post: How Much Does Mined Bitcoin Cost?