How Prohashing earned 111.58% of Nicehash on June 9
Posted: Wed Jun 10, 2020 3:59 pm
A customer contacted me yesterday and asked about the comparison figures on the site's homepage. The inquiry was about whether the posted 111.58% figure for scrypt mining on June 9 meant that mining at Prohashing was actually 11% more profitable than mining at Nicehash that day, or whether there was some sort of error. I thought that I would write a more detailed explanation of how the comparison is performed.
The profitability of Nicehash is obtained by pulling data from the public API that they make available to everyone. The base unit for all of Nicehash's calculations is bitcoins, while the base unit for Prohashing's internal calculations is dollars. Therefore, the price of bitcoins can impact the perceived profitability if the price increases before the end of the day, when payouts are calculated. However, the price of bitcoins was stable, within a 2% range, on June 9. If the price of bitcoins had fallen, then Prohashing's advantage would appear to be reduced because of how our accounting is performed.
Since Nicehash does not post how the calculation is obtained, I can't speak to how they calculate their figure, although it apparently does not include losses caused by their minimum withdrawal amount, nor does it include their withdrawal fee. Nicehash further doesn't clarify whether their figure represents what a perfect miner achieves or what a miner with an average number of rejected shares achieves.
Prohashing's profitability calculation accounts for all fees, but is based upon a perfect miner's performance. Rejected shares could lower that number, and can be caused by latency or miner misconfiguration. Nicehash operates servers across the world, so theoretically some miners' latency to NiceHash could be lower, particularly if the miner is located in China or the Far East. Because Prohashing's mining servers need to communicate with a central database to avoid distributing too much dynamic hashrate across smaller coin networks, and because terabytes of costly SSD space would be required to host duplicate coin daemons, setting up a second mining server in a different region would not improve earnings. Shares would need to be rejected anyway if the notification of new coin assignment from the central server was delayed. Nevertheless, even for customers with a satellite-like latency difference of 200ms, the pool would need to reassign work every 2s or less to decrease profitability by the entire difference.
Therefore, while the exact percentage increase would differ from miner to miner, and the comparison isn't exact, it's very likely that Prohashing has been more profitable than Nicehash, at least for scrypt and a few other algorithms, for a while.
Because most large pools don't offer multiple coins for mining, and because some don't offer merge mining, Prohashing tends to be more profitable than most of the major pools. About six months ago, I tested Antpool using real miners (not APIs) and found that we were 2% more profitable. While I would expect us to slightly ahead of the major pools, I thought that we would never be able to compete with Nicehash because the fact that they essentially operate illegally should permit them to cut operating expenses. For example:
With this extra overhead, I would have been happy if we had been a few percentage points less profitable than Nicehash. Offering an honest and trustworthy business that adheres to the law, which cannot suddenly disappear and which is accountable in the case of theft should be worth some small fee.
But it seems that scrypt and SHA-256 mining is currently significantly more profitable at Prohashing than at Nicehash - and that's even before we fix many coins that are in error due to unreliable exchange APIs later this week. To answer the customer's question: yes, the difference is real, minus the rejected share rate of your miner if Nicehash's API actually returns the average rather than perfect miner's profitability.
The profitability of Nicehash is obtained by pulling data from the public API that they make available to everyone. The base unit for all of Nicehash's calculations is bitcoins, while the base unit for Prohashing's internal calculations is dollars. Therefore, the price of bitcoins can impact the perceived profitability if the price increases before the end of the day, when payouts are calculated. However, the price of bitcoins was stable, within a 2% range, on June 9. If the price of bitcoins had fallen, then Prohashing's advantage would appear to be reduced because of how our accounting is performed.
Since Nicehash does not post how the calculation is obtained, I can't speak to how they calculate their figure, although it apparently does not include losses caused by their minimum withdrawal amount, nor does it include their withdrawal fee. Nicehash further doesn't clarify whether their figure represents what a perfect miner achieves or what a miner with an average number of rejected shares achieves.
Prohashing's profitability calculation accounts for all fees, but is based upon a perfect miner's performance. Rejected shares could lower that number, and can be caused by latency or miner misconfiguration. Nicehash operates servers across the world, so theoretically some miners' latency to NiceHash could be lower, particularly if the miner is located in China or the Far East. Because Prohashing's mining servers need to communicate with a central database to avoid distributing too much dynamic hashrate across smaller coin networks, and because terabytes of costly SSD space would be required to host duplicate coin daemons, setting up a second mining server in a different region would not improve earnings. Shares would need to be rejected anyway if the notification of new coin assignment from the central server was delayed. Nevertheless, even for customers with a satellite-like latency difference of 200ms, the pool would need to reassign work every 2s or less to decrease profitability by the entire difference.
Therefore, while the exact percentage increase would differ from miner to miner, and the comparison isn't exact, it's very likely that Prohashing has been more profitable than Nicehash, at least for scrypt and a few other algorithms, for a while.
Because most large pools don't offer multiple coins for mining, and because some don't offer merge mining, Prohashing tends to be more profitable than most of the major pools. About six months ago, I tested Antpool using real miners (not APIs) and found that we were 2% more profitable. While I would expect us to slightly ahead of the major pools, I thought that we would never be able to compete with Nicehash because the fact that they essentially operate illegally should permit them to cut operating expenses. For example:
- Nicehash continued to do business in the United States as an insolvent company after its $60m hack on December 6, 2017. They continued to pay out money solely earned after the hack. It is illegal under Federal law for a bankrupt company to give preferential treatment to a certain group of creditors (in this case, those earning money after the hack.)
- When I last questioned them about the issue last year, Nicehash had no stated policy against renting their services to conduct 51% attacks - thus condoning criminal activity. Many 51% attacks have been criminal acts intended to steal money from exchanges. Not only does Nicehash make no mention of surveillance conducted to thwart such attacks but they also do not even include the term "51%" in their terms of service.
- Nicehash doesn't promise that their miners will actually submit blocks when blocks are found. They only validate shares, and I did not receive a response when I asked how they validated that there were no "low luck" miners using their service. Their terms of service state that there is a "no-refund" policy once hashrate is purchased. This policyh is one reason why Prohashing has not implemented protocol features required for Nicehash miners to connect to some algorithms.
- Nicehash does not send forms 1099-MISC to US customers who earn more than $600 per year, a violation of IRS regulations - with fines costing at least $500 per user. Corporations based outside the US are still bound by the regulation to send forms to their customers located in the US - and Nicehash operates servers in the US anyway (which are probably at risk of seizure.)
With this extra overhead, I would have been happy if we had been a few percentage points less profitable than Nicehash. Offering an honest and trustworthy business that adheres to the law, which cannot suddenly disappear and which is accountable in the case of theft should be worth some small fee.
But it seems that scrypt and SHA-256 mining is currently significantly more profitable at Prohashing than at Nicehash - and that's even before we fix many coins that are in error due to unreliable exchange APIs later this week. To answer the customer's question: yes, the difference is real, minus the rejected share rate of your miner if Nicehash's API actually returns the average rather than perfect miner's profitability.