The Blockchain Revolution Gets Endorsement in Wall Street Survey | Hacker News Reader

The Blockchain Revolution Gets Endorsement in Wall Street Survey

bloomberg.com

41 pointsherendin31 comments

The actual Bitcoin network (7 transactions per second maximum worldwide) is insufficiently fast enough to conduct settlement of one player of a stock market simulation trading with 5 bots in a tutorial level. Reports of Wall Street adopting it for settlement purposes seem slightly optimistic.

You are correct regarding the low tps and the incompatibility with stock market transactions, but the rest of wall street dwarfs the stock market in market capitalization. The bank debt market they discussed in the article is very low transaction volume but high value and it would most definitely benefit from bitcoin or 'blockchain' technology. Recording and transferring ownership of securities is really just the tip of the iceberg. If interest payments could be made through the ledger then default status would just be a query into the public database. With smart contracts you could also produce automated structured securities where the waterfall rules are encoded and executed in the network. The main difficulty is the border where the network interfaces with the real world. The network cannot enforce collateral transfers, foreclosures, seizures or similar transactions that occur on physical assets so you still need a system of courts and sheriff's.

The bank debt market [...] would most definitely benefit from bitcoin or 'blockchain' technology.

Why?

In the primary market, when debt is first issued through a syndicate of banks, there is not really much of a problem. There is a really great chapter about this process I think in the "The new Market Wizards" book by Schwager, though it might be in the first book, "Market Wizards." The banks call all of their customers and get commitments for dollar amounts towards the deal. Then they put together the paperwork and issue securities. It's kind of like a bond, but it's technically a loan. Each of those investors receive some sort of documentation that they own some piece of that loan and the bank (or some administrator) distributes interest payments to the holders.

The problem arises in the secondary market where initial holders want to sell their piece prior to maturity. There is a long process by which they have to prove ownership and transfer ownership. It's kind of like the process of closing on a home where there is a title search and deed transfer. During this period the buyer and seller are both open to shenanigans from the other party trying to back out of the trade or change the terms. It's also very labor intensive and archaic. If title was stored and transfered through a public trusted distributed ledger system then it could be done instantaneously with minimal effort.

As I said above, there is much more potential here, which makes bitcoin way more promising than just a system for recording and transfering ownership of assets.

Why does it have to be public and distributed? Why not a central authority?

To add to this, note that the central authority does not need to be trusted beyond Bitcoin's "incentive-compatible" assumption. Consider e.g. a public Merkle tree (I believe Cryptolog runs one of those): a dishonest authority could refuse to accept your messages (and your money), but changing the past would be easily detected. (Consider storing a signature from the authority on "as of $TODAY, the root of the Merkle tree is $HASH".)

You need something like the blockchain if you demand a fully distributed solution, but it's easy enough to build a system which does not need a trusted central party.

>Reports of Wall Street adopting it for settlement purposes seem slightly optimistic

Here is just one example: (and I'm aware of many more pilot programs)

http://www.nasdaq.com/press-release/nasdaq-launches-enterpri...

the 1MB blocksize limits are rapidly being tested. However with some of the recent proposals, blocksize would double every 2 years, so something like this:

  2016 8MB 0.1Mb/s 26Tx/s
  2018 16MB 0.2Mb/s 53Tx/s
  2020 32MB 0.4Mb/s 106Tx/s
  2022 64MB 0.9Mb/s 213Tx/s
  2024 128MB 1.7Mb/s 426Tx/s

Right now there is a theoretical ceiling of 604,800 bitcoin transactions that could be undertaken every day. By 2016 that could be 2.2 million, by 2018 4.57 million, 2020 9 million , and 2024, 36 million.

There's a lot of space to grow and a lot of time to pass in which numerous innovations could come to existance, sidechains, lightning networks are a few 'transaction offloading schemes we've seen recently.

In addition, the blockchain proper only needs to be used as a periodic settlement platform, where tx's are farmed together into blocks. and not as a trading engine. (non PoW-based permissioned ledgers are a better fit for that)

You can batch any number of signed transfers of digital assets in to a single hash that is embedded in the blockchain in a single transaction every 10 minutes.

> 7 transactions per second maximum worldwide

What is the reason for this? Source?

The Bitcoin "protocol" (for reasoning on the scare quotes search HN for me and that word) allows blocks to get up to 1 MB in size, currently. A block is mined once every ten minutes. The smallest transaction legal in the protocol is ~224 bytes. You can do the math from here.

There's a contentious proposal in the community right now to raise the block size and hence the implicit maximum TPS of the network. Why is that contentious? Oh boy, long story.

I would be interested in hearing that story. Care to describe it, or point towards some reading material? :).

Start here: http://davidsterry.com/blog/2015/06/arguments-in-the-bitcoin...

Also: https://tradeblock.com/blog/bitcoin-network-capacity-analysi... https://tradeblock.com/blog/bitcoin-network-capacity-analysi... https://tradeblock.com/blog/bitcoin-network-capacity-analysi... https://tradeblock.com/blog/bitcoin-network-capacity-analysi... https://tradeblock.com/blog/bitcoin-network-capacity-analysi...

Hit me up if you need more reading.

Thanks for the links!

The blockchain generates one new block roughly every 10 minutes.

Each block is currently limited to around 1MB (but for lots of reasons, the blocks might be smaller)

Each transaction in a block takes up some amount of space. Depending on the number of inputs and outputs, a transaction can vary in size from around 200 bytes to several kilobytes.

Assuming a 1024kb block and (say) 256 bytes per transaction, that's 4096 transactions every 600 seconds, or roughly 7 a second.

https://imgur.com/SbOzKdJ

I'm not sure I understand the point of using a blockchain in securities trading? I mean, it's not like it would actually have any of the advantages of Bitcoin[1]. You'll still need to settle accounts periodically (transfer money/stock certificates/etc). Those things aren't going to go away.

Once the parties need to trust each other for that, they might as well cut costs by pooling money and buying a centralized order-matching system. That is, make exactly the thing they already have today.

[1]-I don't think those advantages are all that great either, but that's beside the point. The point is they don't even apply here!

https://en.wikipedia.org/wiki/Depository_Trust_%26_Clearing_...

Once you look in to how we're currently able to clear and settle say stock trades, I think you'll agree that there is some room for improvement not only technically but from a trust perspective.

Given that most Bitcoin mining is done behind the Great Firewall[1], what happens if a midlevel bureaucrat partitions the network? (inadvertently or not) Seems there'd be a greater partion behind the firewall, and a lesser partition outside the firewall.

How does the protocol deal with that? Assuming it's partitioned for days, weeks, or months?

[1] Chun Wang comment from http://comments.gmane.org/gmane.comp.bitcoin.devel/8001

"Ignorant. You seem do not understand the current situation. We suffered from orphans a lot when we started in 2013. It is now your turn. If Western miners do not find a China-based VPN into China, or if Western pools do not manage to improve their connectivity to China, or run a node in China, it would be them to have higher orphans, not us. Because we have 50%+."

Assuming a complete net split? Then you have chain B in China and B' outside it, each a valid ledger of a different set of transactions. When they're reunited all the transactions only on B' get briefly invalidated as B' nodes reorganize to support the valid chain. (B is more valid than B' because it is presumably longer.).

This will likely expose the network to double-spend attacks, where e.g. you take an output spent 2 weeks ago on B' and get a transaction to a B node which moves it internally in your wallet (or to a confederate). This will cause Bitcoin to forget the B' transaction and every subsequent transaction tainted by it. Your goal as the attacker is to have 1+ of these transactions give you economic value prior to your counterparty recognizing the reversal.

Good news, though: a net split forking China off Bitcoin would probably result in Bitcoin's central authorities strongly recommending everyone stop processing transactions until they came back.

All it would take is a SINGLE bitcoin user in all of China to figure out a way to sneak 50mb of data a day through the wall to prevent such a network partition.

Indeed, information wants to be free.

There is already talk of launching satellites to disseminate block headers (though initially that's more to protect against Sybil attacks than to connect miners) https://groups.google.com/forum/#!forum/bitsat-project

This is one of the reasons why the Bitcoin Blockchain is not necessarily the best choice as a platform for non-Bitcoin purposes. That's why most of the interest from banks is in blockchain technology - i.e. they're looking at creating their own blockchains.

This is obviously a trivial point, but the word blockchain seems to have lost its definite article.

eg (from the linked article): "Last month, Symbiont, which plans to use blockchain to make ..."

"Other firms investigating finance-related uses of blockchain ..."

exactly. Wall Street is enamored with distributed databases, i.e. "shared replicated ledgers". Not the specific bitcoin blockchain or cryptocurrency.

Blockchain as a term has become a bit like Kleenex. It's just a generic term to describe a distributed database

> Eighty-four percent of respondents said blockchain could reduce the risk a trade won’t settle and the time that process takes, while 74 percent said it could alleviate the chance your counterparty to a trade won’t make good on the deal.

Karma to anyone who can explain this to me. How does the blockchain reduce counterparty risk. The counterparty can either produce the required shares or they can't.

And to be honest, currently cash equities settle on a T + 3 days basis and options settle on a T + 1 day basis. Most people view that as a feature not a bug.

If the markets wanted instantaneous settlement of cash equities they could do it, no one does.

Any asset can be created on top of the bitcoin with meta-layers such as CounterParty [1] (which is what Symbiont is basing their solution off of) Colored coins [2] or a hybrid approach [3] recently demoed by Deloitte

[1] http://counterparty.io/ [2] http://coloredcoins.org/ [3] http://rubixbydeloitte.com/

(These platforms have received recent backing by NYSE and NASDAQ, respectively)

Assets created through Colored coins must use a gateway in a quasi-centralized fashion, however Assets created via counterparty can be traded p2p without a middleman, as they can be natively escrowed by a protocol which enforces atomic swaps and a fair deterministic order matching engine.

DTCC's status quo is T+3(days), if there are trades that can be settled on a blockchain, that can be reduced to T+1(hr) or less.

IMO Bitcoin is a horribly inefficient trading platform, (aside from the 'advantage' of HFT bots being on equal footing as retail investors) but an ideal settlement/netting/clearing layer. As a minimum there are huge transparency gains to be gleaned.

The introduction of Smart contracts potentially opens up the field for a whole new paradigm of smart securities, as well. There are a whole host of established entities researching on blockchain securitization and/or smart contracts at the moment, a subset of them are below:

  CBW Bank
  ANZ
  Westpac
  Commonwealth Bank of Australia
  BNY MELLON
  LHV Bank
  Barclays
  UBS
  Goldman Sachs
  ABN Amro
  ING
  RoboBank
  SWIFT
  Santander
  Standard chartered
  DBS
  USAA
  BBVA
  KPMG
  InfoSys Finacle
  CitiBank
  DTCC
  Deutsche Borse
  Markit
  EuroCCP
  CME

I appreciate the effort but I can't see how this in anyway answer's the question I asked below.

> How does the blockchain reduce counterparty risk. The counterparty can either produce the required shares or they can't.

Let's simplify. I buy 100 MSFT from you. How does the blockchain help settle this trade. At some point I still need to deliver 100 shares to you.

In a simplified way you can use the Bitcoin scripting system to make binding bids and offers without needing a central exchange. And you can achieve person to person atomic swaps of currency for securities – delivery versus payment – without needing a custodian.

Let's assume MSFT shares are represented as specially encoded satoshis.

Below you can find is examples of (fictional) MSFT shares [1] [2] being represented on the blockchain in that way. I'll use an example of CounterParty, which is a meta-layer on the blockchain. The Counterparty protocol acts as an escrow service, and thereby eliminates counterparty risk from the exchange of assets.

[1] http://blockscan.com/assetInfo/MSFT [2] https://coindaddy.io/search?network=xcp&q=MSFT

Now that MSFT asset is stored on the bitcoin blockchain. Actually it is encoded as an OP_RETURN script in the transaction, an breakdown of such an asset you can see here:

  434e545250525459|00000000|000000000004fadf000000174876e800000000000000000000000000
       |             |          |
       |             |          └── All of this is the operation data (maximum 28 bytes)
       |             └──────────────── This is the transaction type identifier (4 bytes)
       └───────────────────────────────── The string CNTRPRTY in hexadecimal (8 bytes)

A breakdown of the sending of such an asset on the bitcoin blockchain:

  000000000004fadf|000000174876e800
       |               |
       |               └───── quantity (8 bytes)
       └────────────────── asset name (8 bytes)

There's some further reading in the links below in more depth than I could hope to go into here:

http://counterparty.io/docs/counterparty_features/#overview

http://counterparty.io/docs/counterparty_features/#protocol-...

Tl;DR There can be no single point of failure in the delivery of securities nor the payment stage. Identical information is relayed to all market participants at all times, consolidated on a single global ledger

(although the option to jump between private 'permissioned' ledgers is possible, and being researched by certain institutions as more suitable for their needs)

Saying you think something is possible is not an endorsement. When Wall Street professionals actually try it, that is an endorse.

Speed trading automated by smart contracts

Reply on news.ycombinator.com