Topic: [POOL][Scrypt][Scrypt-N][X11] Profit switching pool - wafflepool.com - page 287. (Read 465769 times)

Ahh. I guess I just didn't realize what the data actually represented. Not sure I have any interest in it now.

I've never been more happy to be wrong about something.  And it does not surprise me that you've implemented a more efficient profitability algorithm.  You should likely do the smart thing now and edit out your response in order to maintain a slight edge over the other pools, and then I'll follow suit in my quote above.

I had not bothered to scratch below the surface of dogecoin because it may very well fall out of favor after block 100000.  But I'm glad you did.  And I take notice of the fact that in a world where all dedicated single coin mining pools give way to opportunistic multi-coin mining pools, coins like doge may never gain popularity, as everyone will simply stop mining when a low reward block is presented.  Fortunately for those who are positioned to exploit such small edges, the real world is never that efficient, and someone else can slog through those less profitable blocks instead.

If there were more smart people in the world, I would give it a week before waffle reaches 10GH/s, but it will probably take closer to a month.  Which is good, because you probably still have a lot more you would like to implement, and it's easier to get certain tasks accomplished when you're smaller and more nimble.

There's 2 reasons it would be changing.

1) Its a rolling 24hr period (down to the minute).  So if a high value block slides off the end of day1, its now in day2 (day1 drops in value, day2 goes up).
2) And this might be an issue with not explaining the chart properly.  Is that the 7day line, is data for the last 7 days.  It is _not_ data from day 6 to day 7.  Meaning that high luck today (day1) would increase (slightly) the day7 stats as well.


On that note, can a few people take a look at this:
http://wafflepool.com/stats2
And let me know if the first two graphs are an easier method of explaining that data?

Why do previous daily BTC/MH values change? I understand being dynamic on day 1 (current day?), but shouldn't it be static after a 24 hour period?

Congratulations on breaking 1GH/s already  

Its been a low-luck day.  Adding luck stats is definitely on the list, just a bit behind some other stuff we're working on.  As we grow, we're definitely going to have some slow days unfortunately.  Today (so far) seems to be one of them...


Yep, just a matter of tracking coin switches and times across them (mining time per coin since last block).  Not hard, just a bit of time to do right (and tons of stuff to do).


The plan is to allow for a minimum vardiff set via password (d=64 or some such).  But need to get the new endpoints up before we add it in.
Hashpower shouldn't make us more profitable, only reduce variance (luck smoothing)

Literally working on it as we speak.  Spooling up stratum endpoints in US East (NY), US West (California), and Europe (Netherlands).  Hopefully up within a few days (maybe late tonight?)

is there eu server for this pool?

Had some stuff to attend to today, sorry for the delay in responses!  Down the list we go!

This is the way most pools do it, and you're absolutely right that statistical averages work out to 500k/block (using doge for the rest of this example).  However, the block reward for the block you're _actively_ mining for is known when you're mining it.  It is based on the last known block hash.

For example (and its a really ELI5 style example - the actual code is much more complex).  But say the next block reward is worth whatever the last digit of the previous block hash was (assume 0-15 for last digit values).  Statistically you'd earn 8 coins for the next block, however, if we know for a fact that the last block hash ended in "F" (15), we know that the next block reward is definitively going to be worth 15 coins.  Which makes for a _much_ different profitability calculation for if the next block is worth mining.  Since the function for generating block rewards is known (in the client), and all the inputs are known (previous block hash, etc), we can use the _actual_ block reward, rather than the statistical average in our calculations.  Long term, this should give us better performance in earnings. 

Let me know if I can explain it any better


Yep, per-worker stats/graphs are on the list.  Working on the new endpoints first though

I'm pretty sure is getting it right.

The luck statistic I've been asking for is not just display fluff. It lets people judge whether the observed earnings they are seeing are higher or lower than normal due to luck. That is important, especially for a small pool.

There is certainly variance but also MEOW has been one of the most profitable coins for the past several days. If middlecoin is mining (we aren't) it that could explain a discrepancy in profitability.

Raw hash rate does not make a switching pool more profitable in fact can make it less profitable because you burn your own opportunities.  This pool does have a function of not switching all the hash rate to low diff coins, which helps a bit, but it is going to happen.  Raw has rate does reduce variance though.

As things stand one or two days is probably not a long enough time to get a realistic measurement of output on this pool.

Check the stats page for average daily BTC/MH. Yesterday and today have not been as good as previous days.

I also have not seen any pool response issue.

Payout for me is also 3 to 4 times lower :S

I really like the transparency of wafflepool, vardiff should have a minimum of 64 tho.
waffle is just missing the hashpower to be more profitable, if it goes up to middlecoin levels the profits will also increase.
In the end it's all about the money 

Haven't seen that here either. I'll watch for it though.

I've been on Middlepool since it first started, but since I couldn't get a stable connection the last few days, I thought I'd give Wafflepool a try, to see if I could use it as failover at least. I'm mining with 330 kh/s and 95 kh/s, but where on Middlecoin I'd expect 0.005-0.0075 BTC/day, it seems on Wafflepool I'm only getting 0.003-0.004 BTC/day so far. Any smalltime miners that have compared the two pools and can say whether they are comparable?

Havnt got that, and i have been on and off today.

Getting "pool not responding" messages quite a bit today.  Anyone else is is it just me?

Hopefully poolwaffle chimes in here, but he has addressed this a few times now. According to him, the block reward isn't random when it is being mined. It isn't being assumed. The very moment we are mining it, it is known what the block reward will be so those stats are exactly accurate. Here's a quote from poolwaffle on the wafflepool subreddit:



As for individual worker stats, he's addressed that a couple of times and it is in the plans but profitability tweaking and some other features are higher on the agenda (rightly IMO) and as for historical hashrate charting, we can already use wafflestats at http://waffles.wilschrader.com to do this for us thanks to the added API.

It has only been a few days, but it is pretty easy to tell that wafflepool will be very successful if it stays on its current track, and I want to be a part of it.  At some point it will indeed overtake middlecoin, which has so many glaring and outstanding problems, but none of which I choose to point out.

Wafflepool is not perfect though, and the only thing that bothers me is over/under valuing the profitability of coins like dogecoin, which have variable rewards by design.  If the reward for solving a block is truly unknown until after it is solved, then it is improper to assume any block reward value based upon past results.  This is basic statistics and simple game theory otherwise known as expected value.  For dogecoin, the proper assumed value should likely be statically set to 500,000 when deciding which coin to mine.  As I type this message, the expected reward value is currently set to 736,000, which will disappoint almost two times more often than not, and meanwhile we might have passed up mining a more profitable coin with a stable reward value as a result of a bad input into the profitability algorithm.

Wafflepool guy, from what you've been able to put together, you seem like someone I would trust to get this right.  If you give it some thought and run it by some of your smart friends and associates, I am confident that you will ultimately come to the right conclusion.

Also, it would be nice to see hash rate per worker broken out in the stats, as well as an historical hash rate chart, but this request is merely display fluff as compared to my other concern.  Oh, and the best pools always disclose their current PPLNS target.  Thank you for all your hard work!

Edit: 500000 average reward for blocks 1-100000, 250000 average reward for blocks 100001-200000, 125000 average reward for blocks 200001-300000, and so on as per spec.

Seen some good profitability so far with this pool, nice work!

Hey poolwafle, could you tell us a bit about the servers?