Re: Research on the profitability of the main mining pools

[mikeywith]

The mining board admin deleted a very interesting topic which was actually very mining-related that I feel it shouldn't be in "Service Discussion", but well, he runs that board and he does what he thinks best even though I don't agree with many of the things he does including the deletion of this topic.

OP:

Hi. I want to share a good post about profitability of the main mining pools.

In research participated next pools:
· ViaBTC
· F2pool;
· Binance;
· Antpool;
· BTC.com;
· Huobi.

Pools were evaluated for 3 months

Test mining machine: Antminer s17+ 67T

Test date:
1.June 12, 2020, 8.00 — June 25, 2020, 8.00
June 7-June 16: ViaBTC; F2pool; Antpool; BTC.com — difficulty 13.72T
June 1
2.July 30, 2020, 8.00 — August 8, 2020, 8.00
July 30-August 8: ViaBTC; F2pool; Antpool — difficulty 16.85T
3. September 24, 2020, 8.00 — October 18, 2020, 8.00
September 24-October 14: F2pool; Antpool; ViaBTC; Huobi; BTC.com
October 15–17: Binance.
Difficulty 19.31T

Main conclusions:
Profitability in PPS +: ViaBTC > F2pool;
Profitability in PPS +: Binance > Antpool > BTC.com > Huobi

Here you can read the full article  https://interhash.medium.com/research-on-the-profitability-of-the-main-mining-pools-viabtc-f2pool-binance-antpool-btc-com-b19b1dd2426e

Replies to this topic can be found here (Thanks to LoyceV  ).


My reply to Artemis3's post:

Pools don't have a steady income, even if the fiction of PPS makes it look like it, their income is variable because mining bitcoin depends on luck, which is influenced by hashrate.

That is true but the article studies the miner's rewards on all those different pools and not the pool's reward itself, PPS earnings are not subject to luck variance as far as the miners are concerned, so theoretically a 100th on a 100 different PPS pools should yield the same reward.

With that being said, since this is PPS+, then luck will have an impact since transaction fees are paid in PPLNS structure on most pools, the impact however will be minimal if the study was conducted on all pools simultaneously and for a long period of time, which isn't the case with this study, since block fees range by A LOT, mining for a few days at a certain pool when fees are 2BTC per block would make it results look much better than the rest even if it was the worst pool in that comparison given that the margin of error is very tiny.

The pool fees are also another huge factor which the study ignores, I believe Binance had a 0% in most of the testing duration which is why the study "assumes" that Binance outperformed the other pools, we also don't know if the miner was online 24/7, or was the connection quality to all pools similar? the miner could have had a bad connection to Huobi pool and had a lot of stale/rejected shares and that's why it has the worst results, there are a ton of factors that can make this study way too far from accurate.




   

[philipma1957]

yeah that was a nice thread.  I don't feel like I want to go crazy but that thread should have should.

I put in a lot of effort to show a better way to test six pools.  Spent 20 minutes writing it up for what?

Let me gather myself a bit and wait to deal with it..

[mikeywith]

yeah that was a nice thread.  I don't feel like I want to go crazy but that thread should have should.

I put in a lot of effort to show a better way to test six pools.  Spent 20 minutes writing it up for what?

Let me gather myself a bit and wait to deal with it..

Chill out man, I know it's annoying, many good topics and comments get deleted for no reason, but the discussion should continue, I think we should do this more often, whenever an interesting/helpful topic gets deleted from the mining board we move the discussion to the next closest board, even to off-topic if we had to, use Loyce's tool to get the deleted content > https://loyce.club/archive/posts/

here is your reply which I merited before it got deleted

Unfortunately testing methods are not good.

In order to properly test the time tested must match

Ie:  all pools need to have been tested the exact same time

six pools test from Jan 1 to sept 30

means six pieces of equipment  one on each machine.

and then to account for machine differences you rotate the machines on each pool.

week 1
1 on pool a
2 on pool b
3 on pool c
4 on pool d
5 on pool e
6 on pool f

week 2

6 on pool a
1 on pool b
2 on pool c
3 on pool d
4 on pool e
5 on pool f

week 3
5 on pool a
6 on pool b
1 on pool c
2 on pool d
3 on pool e
4 on pool f


the method above will account for differences in the machines
and it makes sure every pool had gear on it was fees were super high.

the method done was not making sure that each pool was getting mined on when fees were 2 btc a block.


If I had the authority I would make the people show this issue in their article.

As I look at it it clearly shows they use one machine and move it pool to pool. This absolutely guaranteed wrong results since the high fees occur only some of the time.  The pool mining on the day of high fees wins.

[philipma1957]

yeah that was a nice thread.  I don't feel like I want to go crazy but that thread should have should.

I put in a lot of effort to show a better way to test six pools.  Spent 20 minutes writing it up for what?

Let me gather myself a bit and wait to deal with it..

Chill out man, I know it's annoying, many good topics and comments get deleted for no reason, but the discussion should continue, I think we should do this more often, whenever an interesting/helpful topic gets deleted from the mining board we move the discussion to the next closest board, even to off-topic if we had to, use Loyce's tool to get the deleted content > https://loyce.club/archive/posts/

here is your reply which I merited before it got deleted.


I went back to the post and fleshed it out. I will do that here.

Unfortunately testing methods are not good.

In order to properly test the time tested must match

Ie:  all pools need to have been tested the exact same time

six pools test from Jan 1 to sept 30

means six pieces of equipment  one on each machine.

and then to account for machine differences you rotate the machines on each pool.

week 1
1 on pool a
2 on pool b
3 on pool c
4 on pool d
5 on pool e
6 on pool f

week 2

6 on pool a
1 on pool b
2 on pool c
3 on pool d
4 on pool e
5 on pool f

week 3
5 on pool a
6 on pool b
1 on pool c
2 on pool d
3 on pool e
4 on pool f

week 4
4 on pool a
5 on pool b
6 on pool c
1 on pool d
2 on pool e
3 on pool f

week 5
3 on pool a
4 on pool b
5 on pool c
6 on pool d
1 on pool e
2 on pool f

week 6
2 on pool a
3 on pool b
4 on pool c
5 on pool d
6 on pool e
1 on pool f


the method above will account for differences in the machines
and it makes sure every pool had gear on it was fees were super high.

the method done was not making sure that each pool was getting mined on when fees were 2 btc a block.


If I had the authority I would make the people show this issue in their article.

As I look at it it clearly shows they use one machine and move it pool to pool. This absolutely guaranteed wrong results since the high fees occur only some of the time.  The pool mining on the day of high fees wins.

In the original post I went back and filled it out. loyce scraped the unedited one. I think he scapes every 15 minutes





I wanted to show a more clear methodology.

all six pools tested none stop for six months.
a machine rotation once a week means all pools have an equal share of each machine.
so pools are more correctly tested this way.

I was going to say to artemis that I disagreed with him about his concept of the fiction of pps.

never got to it.

I also replied to someone maybe the op when he asked what pools i like.

I answered I mostly used viabtc.

[Artemis3]

I didn't mean for the profits to be fictional, only the way those pools give it as if they were getting it in the same proportion. In the pool side they are not getting paid PPS, because bitcoin mining is variable. To keep this "illusion" they have to keep a lot of money in reserve to hand out when they get bad luck, in theory this comes from the periods of good luck + fees. Of course, they also (slowly) adjust their payments if there is a prolonged bad luck or just higher difficulty.

Don't try to make sense of the forum moderation, its closer to the random entropy of the universe, space radiation flipping bits to a cpu, or something quantum.

Also, some of these pools let you switch to "pplns mode" with a different (lower) pool fee. This should also be taken into consideration.

[philipma1957]

I didn't mean for the profits to be fictional, only the way those pools give it as if they were getting it in the same proportion. In the pool side they are not getting paid PPS, because bitcoin mining is variable. To keep this "illusion" they have to keep a lot of money in reserve to hand out when they get bad luck, in theory this comes from the periods of good luck + fees. Of course, they also (slowly) adjust their payments if there is a prolonged bad luck or just higher difficulty.

Don't try to make sense of the forum moderation, its closer to the random entropy of the universe, space radiation flipping bits to a cpu, or something quantum.

Also, some of these pools let you switch to "pplns mode" with a different (lower) pool fee. This should also be taken into consideration.

thank you for your reply.



It does appear we lost the op on an interesting thread.

he had posted the questions and links along with a follow up post.

all deleted

TomValkauskas19

https://bitcointalk.org/index.php?action=profile;u=2881348

0 posts

0 merits  as the merit I was going to give to  him is now impossible to do.