Weekly pool and network statistics

[organofcorti]

That's what I was thinking. The average D would be a good indicator as to whether a pool's hashrate consisted of many low hash rate miners, a fewer high hashrate miners.

[eleuthria]

It wouldn't really show average hashrate of the pool's miners, since each pool will have slightly different formulas for determining when to adjust miner difficulty, and what to adjust them to.  I think BTC Guild is fairly aggressive in terms of keeping variance low, so my target rate is still 1 share per 3-6 seconds.  Other pools might only be targetting 1 per 10 seconds, or 1 per 20.  Others might target 1 per second.  These would lead to drastically different average difficulties, even though they may share the same miner speed distributions.

It just sounds like number porn to me.  It doesn't really provide any meaning, it's just another statistic on the table.

[organofcorti]

It wouldn't really show average hashrate of the pool's miners, since each pool will have slightly different formulas for determining when to adjust miner difficulty, and what to adjust them to.  I think BTC Guild is fairly aggressive in terms of keeping variance low, so my target rate is still 1 share per 3-6 seconds.  Other pools might only be targetting 1 per 10 seconds, or 1 per 20.  Others might target 1 per second.  These would lead to drastically different average difficulties, even though they may share the same miner speed distributions.

It just sounds like number porn to me.  It doesn't really provide any meaning, it's just another statistic on the table.

You mean you don't like number pr0n? Shame on you!

You are right though - if the same dynamic difficulty isn't used in each pool you can't compare one pool to another. You can compare a pool against it's history, though I'm not sure how useful that is. It gets a bit complicated and I dont have time to really think about it atm.

[organofcorti]

30th September weekly pool statistics

The big story this week has to be the first challenge to DeepBit's dominance - 50BTC.com has almost taken the top hashrate spot!

However,  since 50BTC does not publish blocks stats or a hashrate history, 50BTC's current blockrate (170 blocks/week) means their weekly average hashrate 95% confidence interval is 2950 to 3987 Ghps - so they may have already beaten DeepBit's average weekly hashrate, or be not even close. Please provide a block history csv, 50BTC!

Thanks again to everyone who donated over the last fortnight - regular small donations help keep this ship afloat.

You can view all charts at http://organofcorti.blogspot.com

[organofcorti]

6th October weekly pool statistics


Well, it seems that Deepbit has lost the top place for the first time in a year! Or maybe not - the 95% confidence interval for the weekly hashrate of 50btc.com is +/- 500Ghps. Still, it's very close.

The top three pools' total hashrate is just over 50% of the total network hashrate for the first time in a while, and the network hashrate has dropped, too - along with another increase in difficulty.


You can view all charts at http://organofcorti.blogspot.com

[organofcorti]

14th October weekly pool statistics

I hope everyone got some use out of the Variable pool difficulty post. If anything's not clear, please post a comment or contact me with your questions.

Most pools lost some percentage of the network hashrate this week - BTCGuild and Bitminter were the only pools to experience a continuation of hashrate increase.

If there's significant (successful) pool hopping going on at Deepbit it's being hidden by the general variation in pool hashrate there at the moment. Pool hopping at slush's pool has reduced significantly.

There is something odd happening with my calculations of BTCGuild's luck - the shares per round I'm estimating is far too high. I'll try to figure out  how to fix it this week.

Two unexpectedly large donations over the last fortnight - thank you very much to my mysterious benefactors! Smaller donations are just as welcome though: 12QxPHEuxDrs7mCyGSx1iVSozTwtquDB3r

You can view all charts at http://organofcorti.blogspot.com

[Graet]

cheers ooc
but you know how we like changing things?
Ozcoin is 0fee on DGM till next Sunday - as a small way to compensate for recent downtime
PPS will remain at 4%
Cheers
Graet

[organofcorti]

Thanks for the reminder, Graet - I'll make sure the Mining pools list is updated properly.

[Graet]

Thanks for the reminder, Graet - I'll make sure the Mining pools list is updated properly.

I only made the announcement a few hours ago, not sure if you had seen it
Thanks again for the great work.
G

[eleuthria]

14th October weekly pool statistics

There is something odd happening with my calculations of BTCGuild's luck - the shares per round I'm estimating is far too high. I'll try to figure out  how to fix it this week.

You're not wrong.  BTC Guild's luck this last month has been HORRIBLE.  To the point where I've had to pull money out of the cold wallet for the first time in 5 months to avoid the hot wallet running dry.  Since October 1st the pool is 700 BTC in the hole over what it has paid out to miners accounts.  Shit happens, just hoping for a good positive swing soon to offset this last month.

The numbers for this difficulty (previous difficulty stats are off because Stratum wasn't recorded in a way my script can read):

Difficulty: 3054627
Blocks Found: 203
Blocks Expected: 234.01
Difference: -31.01 (-13.25%)


EDIT:  Hmm, maybe your numbers are wrong.  The luck has been absolutely terrible, but I don't think it's been 1.48 times difficulty in the last week.

[organofcorti]

Thanks for the reminder, Graet - I'll make sure the Mining pools list is updated properly.

I only made the announcement a few hours ago, not sure if you had seen it
Thanks again for the great work.
G

I was up early today, saw the post. It's good of you to do that. Reminds me of a story my dad told me about a primitive culture somewhere (I forget) where the shaman gets tribute while the tribe is healthy. As soon as a tribe member got sick, the shaman would receive no tribute from that family until he made them better.

[organofcorti]

14th October weekly pool statistics

There is something odd happening with my calculations of BTCGuild's luck - the shares per round I'm estimating is far too high. I'll try to figure out  how to fix it this week.

You're not wrong.  BTC Guild's luck this last month has been HORRIBLE.  To the point where I've had to pull money out of the cold wallet for the first time in 5 months to avoid the hot wallet running dry.  Since October 1st the pool is 700 BTC in the hole over what it has paid out to miners accounts.  Shit happens, just hoping for a good positive swing soon to offset this last month.

That's really incredibly bad luck. I'm starting to revise my idea of what bad luck is, after your last three weeks:

Code:

                  Date: Sep.23.2012  Sep.30.2012 Oct.07.2012 Oct.14.2012
Negative binomial CDF: 0.0443282499  0.99989794  0.99990376   0.9999975

[eleuthria]

I edited my post a few times organofcorti with some extra numbers, just letting you know.  I think you're right, your numbers do seem to be exaggerated in terms of how bad it is.  The last two weeks (approx.) I'm only 13.5% behind of how many blocks are expected given the shares submitted.

Would it help if I gave you an API page that shows the shares submitted vs blocks found as of a given difficulty?  You could use the delta from your last poll to determine the changes since your last run of the API to determine how many shares were submitted in the last 7 days + blocks found.

[organofcorti]

I edited my post a few times organofcorti with some extra numbers, just letting you know.  I think you're right, your numbers do seem to be exaggerated in terms of how bad it is.  The last two weeks (approx.) I'm only 13.5% behind of how many blocks are expected given the shares submitted.

I'm glad you made the edits, and thanks for looking at your data. I'm also very glad it's probably my script that's doing something weird rather than my idea of what's probable being utterly wrong!

[eleuthria]

Sent you a PM with a new API for you to use to build your BTC Guild stats.  It's the purest form of data I can provide you .

[organofcorti]

Sent you a PM with a new API for you to use to build your BTC Guild stats.  It's the purest form of data I can provide you .

Sweet! I do appreciate you going to the trouble of doing that.

[organofcorti]

20th October weekly pool statistics

This week Deepbit  continues a consistant slide in hashrate that has occurred since the start of the the chart. BTCGuild continues a steady increase in hashrate, and may well be in second place next week. Ah, BTCGuild - always the bridesmaid, never the bride.

Also, thanks to eleuthria, the BTCGuild pool op, for providing me an API for BTCGuild's statistics. Starting next week they'll be much more accurate, and I've deleted the last three weeks of (shares per round) / Difficulty which were inaccurate.

Finally, I've removed the "luck plots' since they were unnecessary, and added a "loss to pool hoppers" plot. This last plot shows the equivalent PPS a miner can expect at either DeepBit or Bitlc, ignoring any pool fees. this week, for example, fulltime DeepBit prop miners can expect to earn 0.99 x PPS before fees, and at Bitlc 0.68*PPS.

You can view all charts at http://organofcorti.blogspot.com

[DrHaribo]

Deepbit seems to be in 2nd place, not 3rd?

Awesome stats though, I like the new "loss to poolhopper" stats.

[organofcorti]

Deepbit seems to be in 2nd place, not 3rd?

Awesome stats though, I like the new "loss to poolhopper" stats.

Crap. Thanks for pointing that out. I had to redo do everything last night when I realised the BTCGuild script I was using was slightly wrong. I didn't redo the intro though.

[organofcorti]

Deepbit seems to be in 2nd place, not 3rd?

Awesome stats though, I like the new "loss to poolhopper" stats.

Crap. Thanks for pointing that out. I had to redo do everything last night when I realised the BTCGuild script I was using was slightly wrong. I didn't redo the intro though.

Post is edited to reflect results.

Glad you like the new chart - I know you have an interest in this area