Topic: [GLBSE] DEEPBIT Miner Variance Protection (Read 5232 times)

see you soon on marketcap

Hope this coin will be growed in the future.

MEI.DEEPBIT.B to date: + 8.4 % compared to proportional, and the ROI is + 17.4% compared to DeepBit PPS.

MEI.DEEPBIT.A1 is now active and asks have been removed. If you wish to purchase any bonds you will need to place a bid. If anyone wants to find out what I will accept for a bid at any point, please PM me.

Unfortunately some MEI.DEEPBIT.A bonds were not returned so I created a new ticker with a new contract. MEI.DEEPBIT.A1 is available now will start paying coupons from blockheight 191520 for 488 DeepBit rounds, at intervals of every 122 rounds, at 0.1016 per bond. It is also properly callable, with a return of one satoshi for each bond on expiry.

Those people buying and selling MEI.DEEPBIT.A bonds, please stop it. This bond has expired and will pay no more coupons. Please send me your bonds so I can retire the security.

MEI.DEEPBIT.B to date: + 2.7 % compared to proportional, and the ROI is + 11.1% compared to DeepBit PPS.
The fourth coupon is the extra to get us up to date.

Due to a problem obtaining data from blockchain info, my script has been assessing DeepBit round length data from two days early. The correct payment to date should be + 4.171% greater than expected. This means a bonus of 0.00278145 btc per bond will be paid at the next coupon.

The script that checks DeepBit now uses blockexplorer to find start blocks, and is working correctly. I apologise for this error, and in order to ensure it doesn't happen again, I will be performing a code audit and manually checking all results until I'm confident that data from blockexplorer fails gracefully.

Note: MEI.DEEPBIT.A was not affected by this error.

Below is the actual data for this bond so far:

MEI.DEEPBIT.B to date: - 0.79537 % compared to proportional, + 9.20% compared to DeepBit PPS.

MEI.DEEPBIT.B Coupon 1 claim: 0.02398039 Btc per bond ( - 4.08 % compared to proportional, + 5.92% compared to DeepBit PPS, and - 5.59% ROI )

MEI.DEEPBIT.A Claims paid / 100 starting bonds: 9.944681 Btc ( - 0.55319 % of expected and -2.10319% ROI).
MEI.DEEPBIT.A is now inactive; MEI.DEEPBIT.B will start paying claims at the end of this quarter.

Whoever it is with a bid of 1 for 0.05 on DeepBit.A, you're offering too much. The expected remaining value is 0.025, and the coupon will pay out shortly after which the bond expires and has a value of zero on recall. I would certainly accept that bid, but you'd be guaranteed a ~ -50% ROI.

A good guide is 0.075 btc after the first coupon, 0.05btc after the second and 0.025 after the third. However, because risk is increased and also some of the blocks in the quarter will have already been solved, you need to take that into account. PM me if you want a quote, and please be careful if you're using bots to buy MEI securities.

MEI.DEEPBIT.B has activated, and will start paying coupons at 123 DeepBit.net rounds solved.
I have removed my asks now, so if you wish to purchase bonds, place a bid or pm me..

I expect DeepBit to finish the 530 round run (which started with the last Difficulty period) later today. If this is the case  MEI.DEEPBIT.A will pay its last coupon and expire, at which point I will call the bonds for 0 Btc (the expected bond value - the expected coupon value).

I hope all users are finding the insurance useful.

MEI.DEEPBIT.B will activate in about 36 hours, and after this point no more insurance will be sold for the Difficulty period. MEI.DEEPBIT.A will soon pay its last coupon and retire until the following Difficulty period.

MEI.DEEPBIT.A Coupon 6 claim: 0.02347875 Btc per bond ( - 6.09 %)
MEI.DEEPBIT.A Claims paid / 100 starting bonds: 7.455261 Btc ( - 0.597 %)

MEI.DEEPBIT.B available now at GLBSE.com for 0.1016 btc. They will be removed from sale at block height 189503, after which we will start paying coupons.

The bonds will be valid for 492 rounds, and pay coupons every 123 DeepBit rounds.

It's ok. I rewrote the code last night. Next coupon pays 132 rounds, the final one pays 133 rounds.

If it is easier do one more on the current schedule then move to the every other schedule, so three more.  Then from then on just do every other one.  Might make the math easier.

We're exactly mid way through at the moment, with 5 payments out of 10 done. Unless any other customers want the original payment scheme, I'll make two more payments, each for 2.5 of the usual payment.

In the meantime I'll contact GLBSE support to see what can be done about voting and the contract.

I would say doubling the current period (or maybe 4x the current period) would be just fine.  Your current payout comes way faster than my mining payouts.

Yes, I thought I'd done it before but I can't find the option.


And this would be the best option by far. However, since the bond would only last about two weeks at a time, I didn't think I'd need to make it a voting type of security, so the "Motions" button does nothing for me.

I'll email GLBSE Support and see if I can organise a vote regardless.


What would your preferred payout period be? Here are some options for you and others (to make it a bit easier):

1. On the expiry of the bond (1 payment only)
2. Weekly (two payments)
3. Every three to four days
4. Keep as is (every 1.4 days or so).



MEI.DEEPBIT.B goes on sale tomorrow, 14th July 0:00 UTC. It will become active from block 189504 onwards for two weeks, when A will be active again. Any changes in payout frequency for A will also hold for B.

You should be able to edit your OP and add a poll but better yet you might put it up as a vote on the asset itself on GLBSE.  At any rate I would vote for the longer payout period since it is a lot of book keeping with the current frequency.

I've had a request to reduce the frequency of payouts in order to reduce bookkeeping. Does anyone have a preference? Unfortunately I'm not sure I can add a poll to an existing topic, so just post your preferences here.

MEI.DEEPBIT.B will IPO on 14th July 2012 at 0:00 UTC, and start paying claims after block 189503.

MEI.DEEPBIT.A and B will run on alternating Difficulty periods.

100 bonds will insure approximately 1.40 Ghps in a 1:1 ratio at the current estimated next Difficulty.

It has been an difficult Difficulty period for DeepBit.net miners so far, and holders of insurance are doing much better than they would have under a proportional payout.

MEI.DEEPBIT.A Coupon 5 claim:
DeepBit's mean round length for the period was 1.071551 x Difficulty, resulting in a  coupon payment of 0.01071551 Btc per bond, 107.1% of expected.

MEI.DEEPBIT.A coupon 4 claim: 0.01071551 Btc / bond (+ 7.1 %)

MEI.DEEPBIT.A average claims paid / bond: 0.10214772 ( + 2.04 %)

MEI.DEEPBIT.B will IPO on 14th July 2012 at 0:00 UTC. MEI.DEEPBIT.A and B will run on alternating Difficulty periods to

The next MEI.DEEPBIT.B will start paying claims after block 189503.

100 bonds will insure approximately 1.25 Ghps in a 1:1 ratio (this will vary with difficulty and pool hashrate)


MEI.DEEPBIT.A Coupon 4 claim:
DeepBit's mean round length for the period was 1.022654 x Difficulty, resulting in a  coupon payment of 0.01022654 Btc per bond, 102.3% of expected.

MEI.DEEPBIT.A coupon 4 claim: 0.01022654 Btc / bond (+ 2.3 %)

MEI.DEEPBIT.A total claims paid / bond: 0.0402127 ( + 0.53%)

MEI.DEEPBIT.A Coupon 3 claim:
DeepBit's mean round length for the period was 1.056887 x Difficulty, resulting in a  coupon payment of 0.01056887 Btc per bond, 105.7% of expected.


MEI.DEEPBIT.A Coupon 3 claim: 0.01056887 Btc.

MEI.DEEPBIT.A Coupon 2 claim:
DeepBit's mean round length for the period was 0.89105 x Difficulty, resulting in a  coupon payment of 0.0089105 Btc per bond, 89.1% of expected.

MEI.DEEPBIT.A Coupon 1 claim results:

DeepBit's mean round length for the period was 1.051 x Difficulty, resulting in a  coupon payment of 0.01051 per bond, 105.1% of expected.

Halfway through the first coupon period and average round length is 1.183 * expected. Bad luck for Deepbit so far.

Block 187487 has been solved,  so the variance protection is live. First coupon to be paid in approximately 1.4 days..

Well, we're up to block 187319, and we're waiting on block 187487 to start. So only a day or so to go before we start the insurance and another day or two (estimated 1.4 days) before the first coupon payment.

I'll update here with results and show how much a miner would have earned with and without the insurance. Anyone have any preferences as to how I should do this?

Also, I'm considering recalling unsold shares at IPO and have replacements sold at 0.9*IPO price as soon as the first coupon has gone out, at 0.8*IPO price when the second coupon has been paid, etc. Do any of my current investors have an opinion about this?

I wanted to report on organofcorti's character. I don't mine and foolishly bought a few shares of this bond, and I PM'd corti about it with some more clarifying information and eventually I realized this wasn't for me. I asked him for a buyback and he agreed to take them back from me. He should be sending my bitcoins back soon! Many thanks! I'm sure this insurance bond has real utility for his target audience

The variable is Deepbit's luck in finding blocks. Only miners at Deepbit have any reason to buy this bond, anyone else is simply not the target market.

You did make a good point though:
If miners diversified across pools, they'd have much less need for insurance like this.

Ok, I deleted the nasty post but your post advertising you services in some else's thread is still there.

Yours is a good idea, but not for me. I know about probability and variance in Bitcoin mining. I'm not an investment expert. I don't offer investments, I don't offer insurance for investments. All I do is offset the variance in pool earnings that miners experience. I know my skillset and I'm not going to risk other people's coin doing more than I know how to do.

So let me make this plain:

MEI.DEEPBIT.A  is not a mining company. MEI.DEEPBIT.A doesn't invest in mining companies. MEI.DEEPBIT.A does not hedge against changes in btc value. No company I solely run will invest in mining companies, or anything else where the risk is not completely and utterly defined beforehand..

I would have liked to have been able to offer insurance without using GLBSE, but I lack the skillset to do so.

Also, are you really offering earnings insurance on round lengths for pooled bitcoin mining? I couldn't find mention of it on your thread or website - can you post a link?


MEI.DEEPBIT.A is is not exposed to variations in Difficulty.

It's not huge and it's exactly quantifiable.

How can the market move against me - did you read the OP? Insurance expires after two weeks and I pay coupons that have an expected value of the cost of the bond minus my fee. If DeepBit is unlucky, I pay more; if DeepBit is lucky I pay less. Coupon payments mimic PPS, and the insurance is only useful for non-PPS miners to offset their variance. The value of the bond during it's two week life is also exactly quantifiable beforehand. Paying more or less for the bond than it's worth at a given point in time might happen, but since this is not an investment it doesn't affect coupons or my business model.

I can think of a way to describe it without using an example other than insurance against earnings variance in pooled bitcoin mining. Perhaps it is a hedge, but I'm not hedging against anything other than the unluckiness of a pool in terms of pooled mining round lengths - not the price of bitcoin, or the value of a market. Because I think this service is most similar to employment insurance I describe it as such.

As far as backing goes, I concede your point. There is no public indicator that I can pay for variance when the pool is unlucky, and I can't expect people to take my word for it. I'll remedy that soon and place a known amount of coins in a low risk and liquid investment.



You really didn't read the OP before posting, I think. Long term Difficulty changes cannot affect the insurance paid. Most of the lifetime of the bond is in one difficulty period, and if D does change it doesn't affect coupon payments. Read the OP. If the difficulty skyrockets there'll be just the same complaints about variance as there are now and were when D was an order of magnitude lower than it is now. It's why pooled mining exists - to reduce variance in mining earnings.


That's a non sequitur if you understand what the insurance does. The bond is not affected by ASICS (unless the ASIC owners use pooled mining in which case they'll be customers), and there is no "long term positive return" and I can show you exactly what the lifetime return probabilities are.

I am grateful you posted and brought up these questions - I'm sure others have thought the same thing probably due to the fact that I had to designate these as a bond rather than an insurance premium. I want to reiterate that the insurance does nothing that a Pay Per Share payment doesn't do - at a much lower fee. It's not an ongoing investment, just insurance against DeepBit's bad luck if you mine there. And also thanks for the heads up about publicly held backing.

Yes, that's right. This is for miners at DeepBit, and also DeepBit's pool operator (since they are also also a PPS pool, bad luck for the pool means extra payouts for him).

This is a bond not a share, and expires within two weeks after all coupons have been paid.

So unless you're a miner you shouldn't buy this bond? Basically, for shareholders to ROI, what are they to bank on?

DeepBit Miner Earnings Insurance bonds are available now, for 0.10155 btc at IPO. They will be valid for 530 rounds, and pay coupons every 53 rounds, starting from block height 187487.

For those of you thinking that Deepbit's variance wouldn't be a problem for miners, atm its not good:


If you had miner earnings insurance, I'd be paying you 181.5% of the usual coupon for that day and a half. For pools with low variance, when bad luck does occur miners seem to feel it even more - they're not used to it.

The difference is variance, and my risk. Some bonds will cost me, and some will pay me. If bonds cost 0.1 btc at IPO I'd have to manage the variance myself, but by charging 0.1015 I'm able to build up a buffer against variance. If I don't have the reserves available I'll buy more coin with which to pay them.

If a substantial number of shares are sold I will be investing them with PatrickHartnett while for the duration of the bond. I'll be abe to reduce the next IPO price if I can do this.

Since some bonds will last more than one Difficulty period, having an "A" bond and a "B" bond mean that I can re - IPO one while the other finishes paying off. There will only be MEI.DEEPBIT.A and MEI.DEEPBIT.B. The same goes for other pools, once I IPO bonds for them, for example MEI.SLUSH.A and MEI.SLUSH.B.

If the bonds don't sell at IPO I had assumed that it would be cancelled and I would wait until the next Difficulty period. I think your suggestion is much better though. If GLBSE allows it (and I'm a GLBSE virgin, so I'm not sure) I would just decrease the cost of the bonds based on the number of coupon payments that would have been paid * IPO price * 10 :

Bought before:        cost of the bond            
1st coupon                   IPO price
2nd coupon               0.9 * IPO price
3rd coupon                0.8 * IPO price
4th coupon                0.7 * IPO price
5th coupon                0.6 * IPO price
6th coupon                0.5 * IPO price
7th coupon                0.4 * IPO price
8th coupon                0.3 * IPO price
9th coupon                0.2 * IPO price
10th coupon              0.1 * IPO price

This is a completely new category of bond, so I'm expecting lots of questions and perhaps a slow start. But I think it's a worthwhile service and once I'm able to offer it for smaller pools I think insurance bonds will be very useful in reducing variance for miners.


Edited to include IPO price in above table.

I'd still like to know where the 0.11166 - 0.1016 = 0.01006 BTC per share from your example are actually coming from. Also I'd still like to know what you will do if these bonds don't sell at IPO or if you can issue additional bonds in the same time frame (it seems you might want to have MEI.DEEPBIT.A, MEI.DEEPBIT.B, ...).

No, although with the plethora of mining bonds available I can see why you'd think so. This is insurance, pure and simple. It pays like a PPS based on DeepBit's round statistics. The number of rounds for the bond's life time is selected so that the bond will pays coupons every one tenth of the lifetime round of the bond.

The amount the bond pays is dependant on the number of shares per round / Difficulty . The greater this number - the worse the pool luck - the more I pay. The expected value for the lifetime of the bond is 0.1 btc, but the return can be much greater if the pool has bad luck. Conversely if the pool has good luck the bond will pay less than 0.1 .

After DeepBit has completed the estimated two weeks worth of rounds and all 10 coupons have been paid, the bond expires and a new one is issued.

An example: DeepBit's average weekly hashrate at IPO is 3170 Ghps, giving an expected number of rounds for two weeks of 510 rounds. Difficulty for the round is 1750000 and an IPO price of 0.1016.

Every 51 rounds I add all submitted shares and divide by (difficulty x number of rounds). I multiply this result by 0.01 and this is the coupon payout for those 51 rounds.

I simulated a life time of rounds below. Each number is calculated using total simulated shares / 1750000 / 51.


Each coupon payout would be:


The total payout would be 0.11166 . The return for this bond (including actual IPO price of bond) would be:

(0.11166-0.1016)/0.1016*100 = 9.9%

This would offset the loss the miner would experience from DeepBit's "bad luck", and is in effect a PPS based payment.

If the round lengths were short, the miner would have earned more (being a proportional miner) and the lifetime value of the coupon would be less than 0.1

I hope this answers your questions, Sukrim.

Where does the money for the dividends come from? Actual mining done by you directly or via contract or some other means (*coughpiratecough*)?

Also - as this lead to some issues in the past - what's your policy on issuing more shares/bonds/"floaters" and how will they be priced? If they are priced differently than at IPO time, how will you handle buyback?

In the end what I think you do is selling a mining contract for 10 difficulty periods that is dependent on deepbit's hash rate. Is this assessment correct?

I agree. I wanted to try it on the pool with the largest number of miners first, and at a 1.7% fee instead of the 10% fee deepbit charges, so miners from Deepbit would still benefit.

But I'm open to suggestions and if there are miners who do want me to offer insurance for their pools, I'll be happy to oblige.

Interesting idea indeed. But Im not sure you picked the right pool for it; monthly variance on a pool as big as deepbit would be quite low anyway. If you would offer it for some of the smaller pools like bitminter, it might be a lot more interesting.

OP updated with more information, fewer charts and an IPO date - 28/6/12 at 0:00 UTC. The insurance bonds will be on sale for approximately one week until block height 187487, when the bond will start paying coupons.

This is not a mining bond. This is insurance for miners, equivalent to a PPS payout at a very low fee.

Please post or message me if you have any questions or would like to buy bonds in advance of the IPO.

I'm not completely certain that the public data from P2pool is accurate, and the long term longer than expected round lengths gives a very high probability that the expected round length is longer than average.

I wouldn't want to offer insurance to p2pool miners until I'm confident that the published pool data is accurate, and then the artificially longer average round length would have to be estimated. I'm keen to include p2pool, but only after any teething issues with the other insurance bonds have been fixed and I can spend some time on a p2pool specific insurance bond.

As a sidenote, I can only insure miners at pools that publish reliable and accurate statistics.

Interesting idea. Any plan of offering the insurance for p2pool?

Pool insurance or miner insurance?

hmm
can pools buy insurance too?
<.<
>.>

tl;dr: Low fee PPS for proportional DeepBit miners

This service provides earnings insurance for Bitcoin miners who want some of the benefits of Pay-Per-Share without changing pools or paying a large Pay-Per-Share fee.

Bitcoin miners on DeepBit who prefer a non Pay-Per-Share payment method often post on the forum with concerns about variance affecting their earnings - especially if they are new miners, mine with a low hashrate, or are dependant on their short term mining earnings in some way. Orphaned blocks are of course paid in a coupon claim.

GLBSE link: https://glbse.com/asset/view/MEI.DEEPBIT.A1  - active
Unfortunately some MEI.DEEPBIT.A bonds were not returned so I created a new ticker with a new contract. MEI.DEEPBIT.A1 is available now will start paying coupons from blockheight 191520 for 488 DeepBit rounds, at intervals of every 122 rounds, at 0.1016 per bond. It is also properly callable, with a return of one satoshi for each bond on expiry. 100 bonds will cover 1450 Ghps in 1:1 ratio.

If you wish to purchase any bonds after activation you will need to place a bid. If anyone wants to find out what I will accept for a bid at any point, please PM me.


GLBSE link: https://glbse.com/asset/view/MEI.DEEPBIT.B - inactive

Those people buying and selling MEI.DEEPBIT.A bonds and MEI.DEEPBIT.B bonds, please cease trading. This bond has expired and will pay no more coupons. Please send me your bonds so I can retire the security.


General information:
Ticker: MEI.DEEPBIT.A1 and MEI.DEEPBIT.B

GLBSE link: https://glbse.com/asset/view/MEI.DEEPBIT.A1
GLBSE link: https://glbse.com/asset/view/MEI.DEEPBIT.B

Company name: Miner Earnings Insurance

Short description: A bond insuring against miner earnings shortfalls on DeepBit.net. The holder of this bond will be paid 10 coupons at regular intervals for the lifetime of the bond.

The holder of this bond will be paid an amount divided into up to 10 coupons at regular intervals for the lifetime of the bond.

Expiration:
The number of rounds for which a bond will be active is based on a pool's previous week's average hashrate to provide an estimated two weeks of operation before expiry.

Depending on how close DeepBit's actual hashrate is to the estimated hashrate, the coupon may expire sooner or later than the fourteen day estimate. This does not affect the value of each coupon or the total coupon amount over the life of the bond.

DeepBit.net does not publish the total round shares for orphaned blocks, so this is estimated from the duration of the round and the average pool hashrate for the coupon cycle.

Expected value:

The expected value of each bond is 0.1, and the expected value of each coupon is the expected value of the bond divided by the number of coupons to be paid. For example, if 10 coupons are paid and the bond's expected value is 0.1, the expected value of each of the 10 coupons is 0.01 btc.

% Probability of lifetime bond values earning more than listed amount, by hashrate:



Coupons:
Coupons will be paid up to 10 times during the lifetime of the bond. The exact number will be published  at the forum link prior to sale. The nth coupon of k coupons will be paid after the pool completes n/k * lifetime rounds. The expected value of a coupon is 0.1/k .

% Probability of coupons earning more than listed amount, by pool hashrate:




IPO Price:
The initial price of the bond varies by pool hashrate. The lower the pool hashrate, the greater the bond price. This is to help manage the greater variance expected in miner payments over lower number of rounds.


Calling:
This bond is callable, in which case I will pay:
If called after expiry, I will pay one satoshi per bond.

Please note that this is not approved by or executed in conjunction with DeepBit.net, and data for calculating the coupon amount is obtained from publicly available information only.

I welcome any comments, suggestions or requests.