[CLOSING DOWN] ScryptGuild Auto-Switching Pool

[BorisAlt]

What guytechi said. What would be the difference if only LTC DOGE left on ScryptGuild ? I love the site, more so than BTC Guild. You must have other reasons obviously but from user stand point ScryptGuild works fine.

[eleuthria]

After a lot of consideration, the merger is likely not going to happen.  ScryptGuild is going to remain a separate entity from BTC Guild.

However, there are going to be some major revamps in how ScryptGuild operates over the next month, to make the process more user friendly, reduce the number of settings, and make it easier to track things like BTC per MH/s/day.  These changes will also be targetting making it easier to add/remove coins from the system.  Some more details will be posted over the weekend.

ScryptGuild is still going to be doing a server migration over the next month to get it connected to the BTC Guild colo for higher uptime and fewer networking issues which currently plague OVH.


The plan is removing the ability to pick a specific coin to mine, and having the pool stop tracking shares submitted to each coin.  Instead the pool will simply track shares submitted, and pool the rewards from all the blocks mined in a given shift and split it that way.  This will reduce the database overhead *significantly*, and also make it easier to add or remove coins to the system.  Users will be able to observe what coins were mined, the exchange rate that was paid for each coin, and then how that reward was split.

Users will be able to pick BTC or LTC for their converted amounts.

Part of this change will also be modifying the coin switching backend as it relates to lower difficulty coins, only moving a percentage of users to a low difficulty coin for example when it becomes the most profitable, something which is not possible to do right now since the system is specifically allocating a user's shares to a specific coin for reward splitting.

[BRADLEYPLOOF]

And I shall begin to spread the word!  Let us grow ScryptGuild to gigahashes!!!

[guytechie]

This is awesome news!  Thanks for reconsidering!

[Destiffmast]

Great news! keep up the good work man :-)

[Felas]

Yeeeeeeesssss  - Thank you - thats all I can say at the Moment

Can´t wait to see the new Features

After a lot of consideration, the merger is likely not going to happen.  ScryptGuild is going to remain a separate entity from BTC Guild.

However, there are going to be some major revamps in how ScryptGuild operates over the next month, to make the process more user friendly, reduce the number of settings, and make it easier to track things like BTC per MH/s/day.  These changes will also be targetting making it easier to add/remove coins from the system.  Some more details will be posted over the weekend.

ScryptGuild is still going to be doing a server migration over the next month to get it connected to the BTC Guild colo for higher uptime and fewer networking issues which currently plague OVH.


The plan is removing the ability to pick a specific coin to mine, and having the pool stop tracking shares submitted to each coin.  Instead the pool will simply track shares submitted, and pool the rewards from all the blocks mined in a given shift and split it that way.  This will reduce the database overhead *significantly*, and also make it easier to add or remove coins to the system.  Users will be able to observe what coins were mined, the exchange rate that was paid for each coin, and then how that reward was split.

Users will be able to pick BTC or LTC for their converted amounts.

Part of this change will also be modifying the coin switching backend as it relates to lower difficulty coins, only moving a percentage of users to a low difficulty coin for example when it becomes the most profitable, something which is not possible to do right now since the system is specifically allocating a user's shares to a specific coin for reward splitting.

[naska21]

T

[Sir Alan]

...the human intellect is far more advanced than any AI system

Ah - you haven't read much on these forums, have you? 

I don't really see much problem with letting the system do everything: it already endeavours to maintain the highest profitability by swapping the currently mined coin at intervals which may be just a few seconds; nobody is going to keep up with that manually.  No system will ever be perfect, but as long as it keeps turning in satisfactory earnings I am content to let the machine take the strain while I get on with my life.

I see the pool's hash rate is already creeping higher than it has been for some time - possibly the good news is already spreading?

[BorisAlt]

Awesomeness. Putting back my meager 10mhs and start spreading the word.

[usao]

The auto-switch has always been a concern of mine.
It's based on the idea that you can mine the currently best priced coin based on something similar to USD/MHs/Day.
I admit that I don't know the exact algo being used, but the issue isn't with the algo, it's with the fact that if you mine any coin, it takes a while to get any return at all, unless you are throwing a new ASIC rig at it.
Then, it has to be converted to USD to capture the actual end-user value.
I don't see the good in mining the most profitable coin now when I won't be able to cash it in until later.
Seems to me better idea is to mine a set of coins that the user likes to mine, then after you have some in your wallet sell them if/when you see a price spike.

[guytechie]

The auto-switch has always been a concern of mine.
It's based on the idea that you can mine the currently best priced coin based on something similar to USD/MHs/Day.
I admit that I don't know the exact algo being used, but the issue isn't with the algo, it's with the fact that if you mine any coin, it takes a while to get any return at all, unless you are throwing a new ASIC rig at it.
Then, it has to be converted to USD to capture the actual end-user value.
I don't see the good in mining the most profitable coin now when I won't be able to cash it in until later.
Seems to me better idea is to mine a set of coins that the user likes to mine, then after you have some in your wallet sell them if/when you see a price spike.

First, to nasaka21: you wouldn't want to set the coin you want to mine manually anyways because you'll essentially be solo mining.  When everyone else auto switches to another coin, you (and maybe a few people) will be the only one in the pool mining that coin.

Now to usao, I don't understand your concern.  I think letting the AI pick the most profitable coin THEN auto convert to BTC would give you the best return because it gets converted while it's considered profitable.  Plus, it's not converting to USD.  Profitability may be based on BTC not USD.

Also, you can turn off the auto convert if you rather keep and sell yourself at a later time.

[usao]

My concern is that there is a large time lag between the mining and the conversion to BTC.
By the time conversion occurs, you generally don't get the peak price it was while you were mining it.

It doesn't do any good to mine the most profitable coin at any point in time, you really want to be selling the most profitable coin at any point in time.

The turnover between the most profitable coins is very fast (short), so mining coin XYZ when it's already selling at a high point doesn't do any good if I can't sell it till an hour or day later when I finally get it in my possession.

[naska21]

...the human intellect is far more advanced than any AI system

Ah - you haven't read much on these forums, have you?  

I don't really see much problem with letting the system do everything: it already endeavours to maintain the highest profitability by swapping the currently mined coin at intervals which may be just a few seconds; nobody is going to keep up with that manually.

[Ktulu1]

I agree with naska. I too use Scryptguild to mine individual coins. The ability to simply switch my workers over to a different coin is way easier than switching command line parameters on all the miners. As far as I know Scryptguild is the only pool that does this. In my opinion this feature is the true greatness of Scryptguild.

I haven't checked recently and a haven't checked against Scryptguilds auto-switching in a while, but the times I have compared my numbers against something like multipool's btc/mh/day numbers I've done better mining one shit coin.

[naska21]

"

[eleuthria]

Regarding manual vs auto:

Most coins are only worth mining for maybe 15-30 seconds at a time.  Due to all the "anti multipool" difficulty switchers, coins will change by factors of 5-10x in their difficulty, while their price moves only a fraction of that amount.  Manual switching cannot be done fast enough for proper profit switching, unless you're *guessing* on a coin having a rally, which is no different than gambling.

The new system, once deployed, will be able to balance the pool hash rate over multiple profitable coins at once, where only a portion of the pool is moved based on what the difficulty of the highest profitable coin is.  This system will also prioritize moving miners with lower stale rates, meaning that less hash rate will be wasted due to moving people who have horrible configurations (intensity too high).


Example:

1) DOGE is the most profitable coin at a difficulty of 700.  100% of the pool is set to DOGE.
2) WDC becomes the most profitable at a difficulty of 13.  The pool moves 30% of its miners to WDC, leaving 70% of DOGE.
3) WDC difficulty adjusts, and that 30% is moved back to DOGE.
4) CSC becomes the most profitable at a difficulty of 0.8.  The pool moves 10% of its miners to CSC.
5) LOT becomes even more profitable due to a high block reward, difficulty of 0.7.  The pool moves 20% of the pool from DOGE to LOT, putting the split at 70/20/10.
6) The LOT block gets solved, new reward puts it much lower in profitability, all those miners return to DOGE.
7) CSC difficulty adjusts back up, those miners return to DOGE.


These are round numbers, but that's the idea of the new system.  The pool will factor in the total speed of the pool compared to the difficulty of the newly profitable coin, to determine a base amount of speed to move to the coin.  It will add a modifier based on how much more profitable the new coin is.



Since the new system no longer tracks per coin shares, the fact that you individually did not have your miner moved to CSC or LOT in that time frame does not mean you didn't benefit from the pool mining it.

[naska21]

The new system, once deployed, will be able to balance the pool hash rate over multiple profitable coins at once, where only a portion of the pool is moved based on what the difficulty of the highest profitable coin is.  This system will also prioritize moving miners with lower stale rates, meaning that less hash rate will be wasted due to moving people who have horrible configurations (intensity too high).


Example:

1) DOGE is the most profitable coin at a difficulty of 700.  100% of the pool is set to DOGE.
2) WDC becomes the most profitable at a difficulty of 13.  The pool moves 30% of its miners to WDC, leaving 70% of DOGE.
3) WDC difficulty adjusts, and that 30% is moved back to DOGE.
4) CSC becomes the most profitable at a difficulty of 0.8.  The pool moves 10% of its miners to CSC.
5) LOT becomes even more profitable due to a high block reward, difficulty of 0.7.  The pool moves 20% of the pool from DOGE to LOT, putting the split at 70/20/10.
6) The LOT block gets solved, new reward puts it much lower in profitability, all those miners return to DOGE.
7) CSC difficulty adjusts back up, those miners return to DOGE.

[eleuthria]

In your future algo u need to include one more factor: pool hashrate vs net hash rate for coin. There is no use to probe net of 200-300Gh/s with the current 2Gh/s of SG. For example, in the shift ID = 2560 SG,  having `2Gh/s has used more than 15% of its time to mine LTC of ~ 230 Gh/s and all in vain, no block was found. If instead this time was spend for CSC (~90 Mh/s) some extra BTC would flow to our wallets

The pool already penalizes high difficulty coins if the pool speed is low.  For the pool to hit LTC, it needs to be at least 15% more profitable than the next best alternative.  Variance is a bitch, but not mining LTC at all because the pool speed is low is going to hurt your income in the long run.

[usao]

But it takes minutes to hours to even mine the coin, then you have to wait for exchange, then you have to cash-in the BTC to get USD.
So, the fact that a coin is most profitable at any given moment doesn't mean much to me if I don't already have that coin ready for trading to USD.
By the time I actually get the mined coins and go through the exchange process to BTC, it may not be very profitable any longer.
I don't mind mining the most profitable coin, but what I really need is the ability to have coins in my wallet(s) ready to sell when the coin is the most profitable.

[Ktulu1]

Regarding manual vs auto:

Most coins are only worth mining for maybe 15-30 seconds at a time.  Due to all the "anti multipool" difficulty switchers, coins will change by factors of 5-10x in their difficulty, while their price moves only a fraction of that amount.  Manual switching cannot be done fast enough for proper profit switching, unless you're *guessing* on a coin having a rally, which is no different than gambling.

The new system, once deployed, will be able to balance the pool hash rate over multiple profitable coins at once, where only a portion of the pool is moved based on what the difficulty of the highest profitable coin is.  This system will also prioritize moving miners with lower stale rates, meaning that less hash rate will be wasted due to moving people who have horrible configurations (intensity too high).


Example:

1) DOGE is the most profitable coin at a difficulty of 700.  100% of the pool is set to DOGE.
2) WDC becomes the most profitable at a difficulty of 13.  The pool moves 30% of its miners to WDC, leaving 70% of DOGE.
3) WDC difficulty adjusts, and that 30% is moved back to DOGE.
4) CSC becomes the most profitable at a difficulty of 0.8.  The pool moves 10% of its miners to CSC.
5) LOT becomes even more profitable due to a high block reward, difficulty of 0.7.  The pool moves 20% of the pool from DOGE to LOT, putting the split at 70/20/10.
6) The LOT block gets solved, new reward puts it much lower in profitability, all those miners return to DOGE.
7) CSC difficulty adjusts back up, those miners return to DOGE.


These are round numbers, but that's the idea of the new system.  The pool will factor in the total speed of the pool compared to the difficulty of the newly profitable coin, to determine a base amount of speed to move to the coin.  It will add a modifier based on how much more profitable the new coin is.



Since the new system no longer tracks per coin shares, the fact that you individually did not have your miner moved to CSC or LOT in that time frame does not mean you didn't benefit from the pool mining it.

I look forward to trying the new system, but I'm going to miss being able to mine individual coins with such ease.