Topic: What do you want in a new Bitcoin Exchange? - page 2. (Read 5796 times)

... although you have to provide a way to take delivery.

All fiat/BTC exchanges provide the ability to take delivery of the fiat via the banking system.  In order to have a credible (i.e. not-a-sham) metals/BTC exchange you need a way to deliver the metal.

+1

Thank you, anu, for striking yet another blow in the war to beat back this silliness that keeps popping up.

A big part of what exchanges (both currency and stock) do is interface with existing banks/courts/governments.  That role will never be "distributed".  But by playing this role, exchanges enable the distributification (?) of "everything else".

An API call to withdraw BTC.

Currently only MtGox and CampBX have this.

Tradehill claims to have this but it doesn't actually work (or at least their tech support staff doesn't think it works).

I think you miss my point which is that a Bitcoin exchange must be a gateway between the proprietary, centralized world of financial institutions and the free, decentralized world of Bitcoin or other XCoin. As those institutions assume that the counter party is also a financial institution registered as a company, and possibly with a banking license, an exchange has to conform to those expectations.

"Whatever" transfers go between two peers, private entities, members of the "whatever" system. From the point of view of the "whatever" system another decentralized system even doesn't exist.

This happens at TradeHill and is really annoying. CampBX has a great way to solve this: a "Spend X" trade. You just need to type in the amount you want to spend and the maximum price you are willing to pay and it calculates how many BTC you can buy (including the fee).

The behavior I have described has been in place for at least 8 months (including on the previous trading engine).  I guess in the big scheme it may be "new".

Simple 'bank -> excange', like Intersango, or better still would be to make it even easier.
Select a leverage degree per trade.
Input either in USD\local currency (GBP for me) or BTC and the other is calculated automatically.
Intelegent UI, for example if there is a "market" vs "custom" price option not then also giving "price (ignore if market)" after that while "market" is selected.

If you have a chart please make it the bitcoins charts chart as there are various different requirements here, for example I like high res but not candlesticks, and all the indicators and other things are helpfull.

Not really. It could be additional to the pure text based interface just to let you easily see where you have your orders. A personal market depth should not be complicated for a trader that also somehow has to understand the global market depth. Allowing to manipulate the own market depth is only logical.

My main criticism was the case where I want to change an order, so you stick to delete one order, enter a quanity, enter a price and submit.

I'm pretty sure I ran exactly into this problem. They improved a lot since then.

Splitting is definitely new.

Holy crap, that's complicated.  I'll stick to price, quantity, submit thank you.  BTW, on MtGox, your fist example would be activated just fine (buy 50BTC@$2).  The fee comes out after the trade, so you would end up with 0.994*50 BTC if your order was filled.  Even if you only had $98, the order would automatically split into an active order of 49BTC@$2 and an "insufficient funds" order of 1BTC@$2.

Finally I would love to see the market depth in real time, too like on mtgoxlive within the same graph.

I want an easy put and change orders interface. Of course this would not target those API users but rather the small people who want to toy around with BTC.

I don't trade much but as far as I know on Gox it works like this:
I have 100$.
I put in an order 50BTC@2$.
It doesn't get active cause I forgot about the fee.
I do a second order of 49BTC@2$.
Later I place a third order of 30BTC@3$.
It doesn't get active as I don't have money left so I have to cancel the other offer first.
Putting, cancelling, watching the market ... involves different tools. I use my calculator to somehow put in all my coins/$$ etc. (MtGox got better with that respect I guess).

I would rather want to have some drag and drop interface like this:

Red are my buy orders and blue my sell orders.
I can drag these in from outside and out to a recycle bin.
I am allowed to save invalid/inactive orders (C and D).
If I charge my account, these become active (and visible in the order book).
E and F are in contradiction as they sum up to more than the 20BTC that I have (blue BTC line) so the last of these would be inactive, too.
When I add/move a red/blue order, the lower red/blue line ($/BTC remaining) moves accordingly.

While dragging and at Mouse-Over I would see the exact price and amount.
Mouse-Wheel would allow me to zoom in and further adjust the Satoshis/ct.

I can switch my interface to BTC or $. The above image shows the $-view, so my 20BTC offer @ 4.2something is shown as a 84$ offer.

Finally I can set the interface to instant confirmation or to ask me for confirmation when saving my changes (default).

I doubt I could trust such a system, but it could run entirely on bitcoin deposits much like Bitcoinica denominates all positions in +/- USD. Consider one of numerous decentralized bots, each with an account at an exchange. Users pay into a bot pool with bitcoin and thereafter trade amongst the other users of the bot account while the bot balances all trades through its exchange account(s). Could a bot register and manage accounts without any single human knowing the details and would anyone trust such a bot? I don't know, but it's interesting, and theoretically possible.

But how do you want to do a SWIFT/SEPA/VISA/PayPal/whatever transfer to a decentralized system? I don't see how a decentralized system will get hooked up to the Swift network and a clearinghouse. These providers make assumptions about the nature of the entities that connect and a P2P network does not qualify.

Is that slick enough?

I did not look into that really, because I don't see any important alt-coins. But that's my IMHO.


Deposit problems are solved using green addresses standard proposed by DeepBit. Then you can deposit from trusted sources instantly.

Downloadable app - yes, it's a must really for fast trading, and we're doing that in ICBIT.

Not really necessary. Private transactions over centralized systems, would be too difficult to control. I again refer to torrent model which is nicely works over centralized connection to internet providers.

- Alt-coins

- Fixed deposit addresses, so we can deposit directly in from a pool

- Slick interface but not sluggish

- PERHAPS a downloadable app that you can exchange on your desktop?  is this a good idea?

- PERHAPS a pool built into an exchange?

if I can think of other ideas i'll post them here.

Damn, that is quite the dangling carrot. Someone get on it!

I have been doing automated trading on some of the bitcoin exchanges (I am doing a significant portion of Tradehill's volume). I think the exchanges we have are decent considering how fast they were thrown together, but there is much room for improvement. I like most of zhoutong's suggestions and I'll throw a couple cents in here:

 - try even lower fees -- yes, I'm surprised nobody has done this yet. Or you could do what some real-world exchange (I think it was BATS?) did, and actually give equity away to the highest volume traders in your first X (3?) months of operation, proportional to their volume. This gets people interested in putting volume on your platform and aligns your interest with the high-volume traders down the road.

 - make your exchange FAST. I have speed issues with all three of mtgox, tradehill, and campbx. And when trading heats up, the exchanges often get so slow that trading intelligently is nearly impossible.
Being fast means getting away from sql-based matching engines, and using an in-memory tree-based structure.

 - there are a number of things that each of the major exchanges could do to improve their REST apis. There are features missing and some design issues. And yes, eventually the awesomest exchange will have connection-based APIs.

 - Don't be broken. It's unnerving when deposits get held up for days because of a software glitch. Or when the book is regularly crossed due to a matching engine glitch. Have a really good test environment and top-notch software people deeply involved.

Tech and feature-wise, I like bitfloor.com best. It is a solid and lightning-fast platform. It's built by somebody who knows what he's doing. Obviously, it has very little liquidity though, which makes it pretty useless until it gets some.

I think there's room for a good gold and silver bitcoin trading platform, but I think mixing it with a usd/btc exchange is probably a mistake. I would focus on one or the other, lest you do neither well. (Shameless plug: if anyone wants to team up on the gold/silver thing, send me a PM; especially if you've experience in the bullion market -- I am looking to put together a team for this).
 
If I were to start a new exchange, I would team up with the bitfloor guy and basically find a way to bring that platform more volume. Or barring that, I would at least use his matching engine (it's been open sourced), or look to it for a design to base mine on.

Thanks


I think I can hear the developers typing to implement everything ASAP