At what point (hash) is solo mining viable?

[DrG]

I have had almost 0.2% of the network before, but that was back in the early days long before ASICs.

You realistically need to earn on average a block a day or more to consider solo mining.  Look at the largest pool behind Ghash and DiscusFish - BTCGuild.  They have had 8 hour spans with no blocks found.  That's with 11 PH at one point.  Even if you had 150TH/s you could easily go an entire difficulty and not get a block - you would want to shoot yourself.  Then the difficulty would go up and it would be even harder for the next 2016 blocks.

Today, only people with large private farms consider solo mining unless you seriously love gambling.

[1713487307]

1713487307

[Relnarien]

I would surmise that having 0.5% of the total network hashrate would be enough to find at least 3 blocks a week. Luck would obviously play a huge factor, but 0.5% might be consistent enough as is. It would be much more prudent to at least have 1% though, since the network hashrate is continually increasing. Anyway, if my information is accurate, then 0.5% of the current total network hashrate would amount to over 677 Ths. That would require an investment of over a million dollars, not including electricity, maintenance and housing costs.

[minerpumpkin]

Guys, you do realize that it really doesn't make sense do mine with such a low hash rate you are all suggesting (<10 GH/s), the chances are so incredibly low (and are getting worse every 2 weeks) that you won't find a block. If it took you 150 years today, it will take you 1000 years this time next year. Not going to happen.

[Crossbow376]

I would surmise that having 0.5% of the total network hashrate would be enough to find at least 3 blocks a week. Luck would obviously play a huge factor, but 0.5% might be consistent enough as is. It would be much more prudent to at least have 1% though, since the network hashrate is continually increasing. Anyway, if my information is accurate, then 0.5% of the current total network hashrate would amount to over 677 Ths. That would require an investment of over a million dollars, not including electricity, maintenance and housing costs.

With 0.5% of total network hashrate, you should expect to find 0.5% of the future blocks and so you should get 10 blocks every 2016 blocks (usually, 12-13 days) on average.

[Relnarien]

I would surmise that having 0.5% of the total network hashrate would be enough to find at least 3 blocks a week. Luck would obviously play a huge factor, but 0.5% might be consistent enough as is. It would be much more prudent to at least have 1% though, since the network hashrate is continually increasing. Anyway, if my information is accurate, then 0.5% of the current total network hashrate would amount to over 677 Ths. That would require an investment of over a million dollars, not including electricity, maintenance and housing costs.

With 0.5% of total network hashrate, you should expect to find 0.5% of the future blocks and so you should get 10 blocks every 2016 blocks (usually, 12-13 days) on average.

That's a theoretical estimate based on proportional percentages. I did not base my estimate on that because it is unreliable when small numbers are involved. I was surmising based on a squared-percentage principle, which from my personal perspective, provides a better estimate for real world values. Essentially, my assertion is that one's estimated found blocks are more likely to be squarely proportional to one's controlled percentage of total network hashrate rather than directly proportional. For example, having 0.1% of the total network hashrate will net you an estimated (0.1 * 0.1 = 0.01)% rather than 0.1%. Obviously, that is due to luck or lack of it. However, like I said, I find that it functions as a better estimate for real world values. It's all irrelevant anyway. Anyone who can do fourth grade math can easily calculate percentages on their own if they believe that probability is the same thing as statistics.

[ensurance982]

I would surmise that having 0.5% of the total network hashrate would be enough to find at least 3 blocks a week. Luck would obviously play a huge factor, but 0.5% might be consistent enough as is. It would be much more prudent to at least have 1% though, since the network hashrate is continually increasing. Anyway, if my information is accurate, then 0.5% of the current total network hashrate would amount to over 677 Ths. That would require an investment of over a million dollars, not including electricity, maintenance and housing costs.

With 0.5% of total network hashrate, you should expect to find 0.5% of the future blocks and so you should get 10 blocks every 2016 blocks (usually, 12-13 days) on average.

Yeah it's prudent to assume a certain percentage of the current total network hash rate. It doesn't take too much bad luck to find less blocks than anticipated, I think it's better to join a pool for a steadier payout. Also keep in mind, that the variance can be a major pain! You may not find a block for a multiple of the expected interval!

[DeathAndTaxes]

It all comes down to how much time you are willing to accept between paydays and how much variance.  Remember bitcoin mining is a poisson distribution ( http://stattrek.com/online-calculator/poisson.aspx ).  Lets assume you are not willing to go one difficulty period without finding a block.

If you have 1/2016th (0.05%) of the network you should find on average 1 block per difficulty period.
Expected = 1 block
Probability of 0 blocks = 36.7%
Probability of 1 bock =  26.4%
Probability of 2+ blocks = 36.7%

You have a ~1 in 3 chance of doubling (or more) your return but also a ~1 in 3 chance of earning nothing. That is probably more risk than most miners are willing to take.

If you have 2/2016th of the network (~0.1%) you should find on average 2 blocks per difficulty period.
Expected = 2 blocks
Probability of 0 blocks = 13.5%
Probability of 1 block = 27.1%
Probability of 2+ blocks = 59.4%

Looking better but lets look at 5 times as much hashrate, 10/2016th of the network (0.25%).

If you have 10/2016th of the network (~0.5%) you should find on average 10 blocks per difficulty period.
Probability of 0 blocks = ~0% (1 in 22,026)
Probability of 1 to 5 blocks = 6.7%
Probability of 6 to 9 blocks = 39.1%
Probability of 10+ blocks = 54.2%

So less than a 7% chance of earning less than half of expected.  How about a nice even 1% of the network.

If you have 20/2016th of the network (~1%) you should find on average 20 blocks per difficulty period.
Probability of <6 blocks = ~0% (1 in 13,906)
Probability of 7 to 13 blocks = 6.6%
Probability of 14 to 19 blocks = 45.9%
Probability of 20+ blocks = 52.9%

Now we are looking good.  Only 7% chance of earning less than 70% of expected.  Of course your risk tolerance will vary but the same distribution can be done at any percentage of the network.  As the difficulty growth goes down you may not necessarily care about how many blocks in each difficulty period but rather how many blocks per month since the powerbill is due each month.  There are ~4,320 blocks per month which means the probability of extremely bad or good luck over that longer period of time decreases (quite a bit).

[ensurance982]

If you have 20/2016th of the network (~1%) you should find on average 20 blocks per difficulty period.
Probability of <6 blocks = ~0% (1 in 13,906)
Probability of 7 to 13 blocks = 6.6%
Probability of 14 to 19 blocks = 45.9%
Probability of 20+ blocks = 52.9%

Now we are looking good.  Only 7% chance of earning less than 70% of expected.  Of course your risk tolerance will vary but the same distribution can be done at any percentage of the network.  As the difficulty growth goes down you may not necessarily care about how many blocks in each difficulty period but rather how many blocks per month since the powerbill is due each month.  There are ~4,320 blocks per month which means the probability of extremely bad or good luck over that longer period of time decreases (quite a bit).

Really? Only 1%, I figured it would've been a bit more to be already that safe in terms of variance. I guess solo mining actually is a viable option for some large mining operations. The biggest problem that remains is that you have to keep up with the pace. You have to bring new and better mining equipment online with every difficulty increase.

[Tammy Chan]

If you have 20/2016th of the network (~1%) you should find on average 20 blocks per difficulty period.
Probability of <6 blocks = ~0% (1 in 13,906)
Probability of 7 to 13 blocks = 6.6%
Probability of 14 to 19 blocks = 45.9%
Probability of 20+ blocks = 52.9%

Now we are looking good.  Only 7% chance of earning less than 70% of expected.  Of course your risk tolerance will vary but the same distribution can be done at any percentage of the network.  As the difficulty growth goes down you may not necessarily care about how many blocks in each difficulty period but rather how many blocks per month since the powerbill is due each month.  There are ~4,320 blocks per month which means the probability of extremely bad or good luck over that longer period of time decreases (quite a bit).

Really? Only 1%, I figured it would've been a bit more to be already that safe in terms of variance. I guess solo mining actually is a viable option for some large mining operations. The biggest problem that remains is that you have to keep up with the pace. You have to bring new and better mining equipment online with every difficulty increase.

I just lol at the "only 1%".

You need about 1600 TH/s to have 1% of network hashrate currently, which is roughly 1600 AntMiner S2 or 530 KnC Neptune.
And exactly as you said, to keep up with the uprising difficulty and network hashrate, you will need to keep buying more and more miners to maintain the 1% "market share".

[odolvlobo]

By my calculations, assuming a 10% rise in difficulty each period, you will need about 7.3 TH/s in order to solo mine an average of one block per year over the next year. However, you are most likely to find a block earlier than later because the longer you go without finding a block, the harder it becomes.

[midnightmagic]

DaT, people are misinterpreting your post to suggest that miners can expect to earn less when they solo with smaller hashrate. The expected return is identical for any hashrate between solo'ing, and pool-mining, no matter the combined hashrate, ignoring all other factors (e.g. central pools stealing fees or just plain stealing, or latency tricks like what p2pool does to achieve lower orphan rates than anybody.)

I recognise you are trying to explain that variance in return is less with pooled mining or higher hashrate, but this is meaningful only because humans think that way because of our weird broken minds. That is, a lower variance in return, and especially a more-frequent payout, feels like it is worth more to humans just because we all know we might get hit by a bus tomorrow.

What you are doing is showing in a calculation a way for people to calculate payout schedules they are more-comfortable with.

People should recognise that mining in a pool like P2Pool which has at the moment a daily payout, can be expected over time to payout identically to higher-hashrate collectives.

[DeathAndTaxes]

You are correct.  Anyone taking what I wrote and saying the expected return is less is misinterpreting it.  The expected return is based solely on the block reward (plus average fees) and the miners share of the overall network.

[BunsenBurner]

I recognise you are trying to explain that variance in return is less with pooled mining or higher hashrate, but this is meaningful only because humans think that way because of our weird broken minds. That is, a lower variance in return, and especially a more-frequent payout, feels like it is worth more to humans just because we all know we might get hit by a bus tomorrow.

Indeed, I believe a big proportion of people are indeed risk-seeking (enjoys taking risk).
Just see how many people go to Vegas or gambling sites to play those negative EV games.

[RoadStress]

It all comes down to how much time you are willing to accept between paydays and how much variance.  Remember bitcoin mining is a poisson distribution ( http://stattrek.com/online-calculator/poisson.aspx ).  Lets assume you are not willing to go one difficulty period without finding a block.

If you have 1/2016th (0.05%) of the network you should find on average 1 block per difficulty period.
Expected = 1 block
Probability of 0 blocks = 36.7%
Probability of 1 bock =  26.4%
Probability of 2+ blocks = 36.7%

You have a ~1 in 3 chance of doubling (or more) your return but also a ~1 in 3 chance of earning nothing. That is probably more risk than most miners are willing to take.

................

Please teach me how to calculate the probability of 0, 1 and 2+ blocks per difficulty period with a certain hashrate (1Th/s for an easier example) and for the current difficulty (make it 20 bil for easier example). I have some Jupiters ready for some gambling. Thank you.

[realbtcdealers4real]

It's plain useless.

[RoadStress]

It's plain useless.

It may be, but I would like to see what are the chances that a guy with ~160Th/s finds 3 blocks in 1 day http://bitsolo.net/pool-stats/

[TheDragonSlayer]

It's plain useless.

It may be, but I would like to see what are the chances that a guy with ~160Th/s finds 3 blocks in 1 day http://bitsolo.net/pool-stats/

No matter how lucky a guy can be its still consider gamble unless you have 1 PH... Pool is the way!!!

[lynn_402]

It's plain useless.

It may be, but I would like to see what are the chances that a guy with ~160Th/s finds 3 blocks in 1 day http://bitsolo.net/pool-stats/

No matter how lucky a guy can be its still consider gamble unless you have 1 PH... Pool is the way!!!

However this gamble might be worth it, instead of simply mining a few dollars worth of coins in the machine's lifetime.

[RoadStress]

It's plain useless.

It may be, but I would like to see what are the chances that a guy with ~160Th/s finds 3 blocks in 1 day http://bitsolo.net/pool-stats/

No matter how lucky a guy can be its still consider gamble unless you have 1 PH... Pool is the way!!!

4 blocks in a day

[wasserman99]

It's plain useless.

It may be, but I would like to see what are the chances that a guy with ~160Th/s finds 3 blocks in 1 day http://bitsolo.net/pool-stats/

No matter how lucky a guy can be its still consider gamble unless you have 1 PH... Pool is the way!!!

4 blocks in a day

I don't think you need to have enough mining capacity to expect to mine 4 blocks a day to be able to viably solo mine. I would argue that the figure would be closer to between .75 and 1.25 blocks per day (expected value). Once you get closer to the lower end of this range, you will likely be much better off mining on a pool because your variance can potentially hurt your return a lot.