[UPDATED] MTGox and dark pool

[theymos]

People will do it anyway with bots, so I don't see any reason to remove it.

[1714924275]

1714924275

[1714924275]

1714924275

[1714924275]

1714924275

[imanikin]

There is also the unfair fact that 1 big trader is more valuable to the market than 1 small one since a big trader provides more liquidity. So an exchange has more incentive to keep the big guys happy.
 
I'm actually a bit curious what would happen if the order book wasn't shown at all. I suppose it would make trading less interesting.  

The point is also that dark pools distort price discovery and the commodity's market price, at any given moment...

I think one of the main reasons Bitcoin and its successors have a bright future, is because in the existing financial system, zillions of small guys are really sick of being bent over by all types of entities “to keep the big guys happy.” There are far more small guys than big guys...  

If the books were completely closed, my bet is that eventually an insider trading gang will form around whoever does have access to the books, and will be able to distort and manipulate the price even better than now...

I get that the free-market cheaters and hypocrites will continue to devise ways to cheat the fundamental market forces and be hypocritical about it, because it's profitable to them, and well-worth the complex, technical rationalizations and self-delusions. I am just not willing to partake in such places anymore. I am sure the happy big guys won't miss my small money... So, have fun, jed!

[BitterTea]

lol, this again...

If there was no dark pool, big traders would most likely still not publish their buy/sell intentions in the order book, they would use trading bots. There is absolutely nothing wrong with this. I am not obligated to announce my economic intentions to the world, just because I disagree with state intervention in markets.

[NghtRppr]

I have no opinion on whether dark pools are "good" or "bad", whatever that means, but as the operator of a market, my number one goal is transparency. When people place an ask or bid, I want them to be confident that they have all the information that is currently available. Whatever asks or bids are placed later are a different story. I can't control that. The bots are welcome to do their thing but out in the open.

[S3052]

People will do it anyway with bots, so I don't see any reason to remove it.

thanks for your perspective!

Great

[Psychoactive]

lol, this again...

If there was no dark pool, big traders would most likely still not publish their buy/sell intentions in the order book, they would use trading bots. There is absolutely nothing wrong with this. I am not obligated to announce my economic intentions to the world, just because I disagree with state intervention in markets.

Then I shouldn't be forced to publish my economic intentions to the world, even if I have less than 1000BTC. Dark pool should be open to everyone or not at all.

The argument that bots can simulate dark pool applies to small traders : they can use a bot to simulate dark pool behaviour, so we may just as well allow dark pools for them too. There is no reason why dark pool should concern only big orders.

No dark pool or dark pool available to everyone.

As someone else imagined, we could also have only dark pools and make completely blind bids, with no information on others' bids. Anyhow, I'm not sure we'd like that, dark pool bidders included, because their bids are based on bids in the normal pool. The idea of the market is precisely to give you information on general economic intentions. This is why I think the dark pool should be removed. Even if bots are able to simulate it, at least you can analyse their behaviour and try to recognize bids placed by bots and other bids. And at least you'll have to work  to have it. + 'dark' sounds terrifying

[killerstorm]

The point is also that dark pools distort price discovery and the commodity's market price, at any given moment...

Price discovery? In last 48h price went from 0.72 to 0.80 then back to 0.73 then up to 0.76.
Is this a price discovery?

What fundamentals changed in last 48h?

I think this is a volatility caused by large orders. Check volume and price charts alongside, looks like it is.

There are simply not enough large players on the market -- market-makers, arbitragers, speculators -- who could smooth out effect of these large orders moving the market. You can see that eventually price recovers, so this is clearly a problem of depth.

In can be shown that dark pool orders work towards stabilizing current price. Thus it is good for the market because volatility is the biggest problem.

I think one of the main reasons Bitcoin and its successors have a bright future, is because in the existing financial system, zillions of small guys are really sick of being bent over by all types of entities “to keep the big guys happy.” There are far more small guys than big guys...  

If the books were completely closed, my bet is that eventually an insider trading gang will form around whoever does have access to the books, and will be able to distort and manipulate the price even better than now...

I get that the free-market cheaters and hypocrites will continue to devise ways to cheat the fundamental market forces and be hypocritical about it, because it's profitable to them, and well-worth the complex, technical rationalizations and self-delusions. I am just not willing to partake in such places anymore. I am sure the happy big guys won't miss my small money... So, have fun, jed!

This is just a piece of FUD. You cannot explain how dark pools are bad for you and so you resort to emotions about "big guys" and "bending over".

If you will compare situation with dark pools and without, dark pools will be strictly superior as long as dark trading volume is a small fraction of total trading volume.

Note that in dark-only case people could just trade OTC and just using mtgox price as a reference. It is not your business to tell other people where and how they should trade.

[S3052]

The point is also that dark pools distort price discovery and the commodity's market price, at any given moment...

Price discovery? In last 48h price went from 0.72 to 0.80 then back to 0.73 then up to 0.76.
Is this a price discovery?

What fundamentals changed in last 48h?

I think this is a volatility caused by large orders. Check volume and price charts alongside, looks like it is.

There are simply not enough large players on the market -- market-makers, arbitragers, speculators -- who could smooth out effect of these large orders moving the market. You can see that eventually price recovers, so this is clearly a problem of depth.

In can be shown that dark pool orders work towards stabilizing current price. Thus it is good for the market because volatility is the biggest problem.

I think one of the main reasons Bitcoin and its successors have a bright future, is because in the existing financial system, zillions of small guys are really sick of being bent over by all types of entities “to keep the big guys happy.” There are far more small guys than big guys...  

If the books were completely closed, my bet is that eventually an insider trading gang will form around whoever does have access to the books, and will be able to distort and manipulate the price even better than now...

I get that the free-market cheaters and hypocrites will continue to devise ways to cheat the fundamental market forces and be hypocritical about it, because it's profitable to them, and well-worth the complex, technical rationalizations and self-delusions. I am just not willing to partake in such places anymore. I am sure the happy big guys won't miss my small money... So, have fun, jed!

This is just a piece of FUD. You cannot explain how dark pools are bad for you and so you resort to emotions about "big guys" and "bending over".

If you will compare situation with dark pools and without, dark pools will be strictly superior as long as dark trading volume is a small fraction of total trading volume.

Note that in dark-only case people could just trade OTC and just using mtgox price as a reference. It is not your business to tell other people where and how they should trade.

Agree with you.

Also, the poll should only be open to registered users of MtGox. It is about satisfying current customers, rather than given people a right to vot who has absolutely no clue about trading.

[imanikin]

In can be shown that dark pool orders work towards stabilizing current price. Thus it is good for the market because volatility is the biggest problem.

This is just a piece of FUD. You cannot explain how dark pools are bad for you and so you resort to emotions about "big guys" and "bending over".

As i mentioned, i've already explained elsewhere why dark pools are not for me, and detrimental to Bitcoin imho...

No need to take my word for it though. Those on both sides of the ocean, who “have a clue about trading”, don't seem to agree with you or S3052 or the rest of the dark-pool dwellers regarding dark pools and their effect on price discovery, volatility, or liquidity:

http://cachef.ft.com/cms/s/0/f2f7e53c-e343-11de-8d36-00144feab49a.html

It comes from this CFA Institute report:

http://www.cfapubs.org/toc/ccb/2009/2009/13

Here is another useful snippet:

http://www.ft.com/cms/s/0/d5f8b10c-0f70-11de-ba10-0000779fd2ac.html

All you have to do is google “dark pools currency exchange distortion” to  find out more about the way "happy big guys" are riding the rest of us... 

[S3052]

I know these sites, but dont agree with their view.

Having observed the bitcoin market for 6 months on a daily basis, my only point is that the bitcoin trading has become smoother since the introduction of dark pools.

anyway I am not the decision maker, and accept either decision.

[nanotube]

Here's my take on the issue of dark pools.

There are two parties that need to be considered when thinking about the 'fairness' of dark pools. One party is the prospective buyer or seller coming in and placing a market order. For this party, the existence of dark pools presents no disadvantage, since the at-market transaction only stands to be executed at a price that is strictly better than the bid/ask spread posted in the open order book. Thus far, this is the only party that has been considered in the discussion.

The second party, is the people who place limit orders - namely the outstanding open bids and asks. When a party comes in, looks at the open orders, and places a limit order accordingly, it has expectation that say, given that a trade in amount X comes in on the buy or sell side, his order is going to be filled. If there is a hidden dark pool, however, he is operating on incomplete information when placing his order, and stands to get his order filled slower, or not at all. This is a clear disadvantage to the limit order placing parties.

Thus, the dark pools in their current implementation, essentially take from the limit order placers and give to the market order placers and the dark pool order placers. This amounts to a 'taking' from one party to give to another, and is in my opinion not a good, or fair, thing to do.
   
I think the proper way to do dark pool orders is to use "iceberg orders", wherein a dark pool order entry gets two inputs, the total order quantity, and a display quantity. This is how it is currently done on exchanges that support dark pool liquidity in addition to open order book. I quote from the wikipedia article on dark pools, http://en.wikipedia.org/wiki/Dark_liquidity

Iceberg orders generally specify an additional display quantity, smaller than the overall order quantity. The order is queued along with other orders but only the display quantity is printed to the market depth. When the order reaches the front of its price queue, only the display quantity is filled before the order is automatically put at the back of the queue and must wait for its next chance to get a fill. Such orders will therefore get filled less quickly than the fully public equivalent, and they often carry an explicit cost penalty in the form of a larger execution cost charged by the market.

Note the defining feature of this style of dark pool: the dark pool order user gets the same priority as the other limit order placers, and once his display quantity is filled, he goes to the back of the queue. This way, there is no taking from the limit order placers. The fact that this type of order would fill slower than a fully-open order, would encourage people to put larger parts of their order into the open (as would the extra charge by the exchange, if it is instituted.)

One final note: the current implementation of dark pools is NOT equivalent to what one could do with a bot, contrary to has been stated a couple times in this thread. When using a bot to continually maintain a smaller display order, one's whole order quantity is NOT prioritized over all outstanding limit orders. Rather, before the bot can react to replace the display quantity, any market orders coming through would be filled by other outstanding orders. In fact, you may notice that the activity that is achievable by a bot is exactly equal to the "iceberg orders" described above. Thus, those who would like to achieve equivalence to what could be done via botting, should vote to remove the current dark pool implementation and replace it with the "iceberg" order type.
   
These comments only apply to the dark/light order type, since the full-dark order is effectively completely separated from the rest of the market, I have no opinion on whether to keep it or do away with it, I think it is irrelevant.
   
Therefore, following from the reasoning above, I vote for the removal of the current style of dark/light order type, and its replacement by a proper iceberg-style order type, which does not take from the limit orders and give to the dark and market orders.
   
Given the available poll options, I chose the second option, namely, "keep it, but remove the dark/normal option". If there were an option to say "change dark/normal to a standard iceberg order" I would have voted for it, but there isn't.
   
I ask all participants in this poll to consider the above, and vote for either option 2 (remove dark/light, keep full-dark), or option 3 (remove both), if you agree.

[Keefe]

That's an interesting concept. I hadn't heard of iceberg orders before. It seems like the functionality is exactly like a bot with no latency. I see it as a way to bring such bot functionality to everyone, possibly for a higher fee, instead of only to those skilled at coding and with access to a reliable server.

I'm not sure it will satisfy all current dark+normal order users though, since they could miss out on large market orders that are placed in one piece. But that's the cost of hiding one's size. It seems fair enough.

[toffoo]

First off, let me thank you for posting this interesting and important topic here for the community to discuss.  I'm glad to see you thinking deeply about making any change to the dark pool option and using this forum as the mechanism for ideas and feedback.

When I first saw this topic posted, I thought about it pretty deeply but couldn't think of an ideal solution so I voted "leave it as it is".

Now that I've read about nanotube's iceberg order idea, I think that it is the best compromise and if it were one of the offered options I would vote for that.  

Let me raise a few points here that I didn't see posted previously...

First, Mt.Gox is by far the best and most liquid BTC/USD exchange in existence at the moment and provides a crucial function to the community and an increasingly important role in the future growth of Bitcoin.  We should all be cognizant of how it got to this position and should not take the decision lightly to change it.  To some extent, "if it ain't broke, don't fix it".

Second, the question of yes/no/how to dark pools has already pretty much been answered by existing electronic financial markets in the world.  We should learn from them, what they have done and how they have implemented it.  The bottom line is that pretty much any electronic market in the world has a dark pool of some sort, and if they don't, third-party off-exchange dealers quickly offer them.

Third, thinking for a moment what is best for Mt.Gox the exchange, rather than for the community of buyers, sellers, and traders, if we were to shut down the dark pool, what's to stop somebody else from opening another one?  Given that Mt.Gox's complete order book and trade history are totally public information in real-time, what would stop me or anybody else from opening a 'darkmtgox.com' to offer dark pool trading based around Mt.Gox's publicly posted prices?  As long as some traders feel dark pool trading is useful, it's almost a certainty this would happen.  I don't think this would be good for Mt.Gox or the community in general to fragment the most liquid market we have.

Fourth, I think the most pressing issue with the BTC/USD market at the moment is the volatility of the exchange rate and lack of liquidity for large trades.  If Bitcoin is going to make it big, we cannot continue to have weeks where the value of a Bitcoin drops (or jumps) by 50% or where orders of 10k BTC take out all the posted bids on the exchange.  If this is the most pressing issue, why are we considering removing something that clearly increases liquidity and decreases volatility?

As mentioned earlier in this thread, the way to increase liquidity and dampen volatility of any electronic exchange is to encourage the participation of more market-makers, arbitrageurs, and speculators.  These are generally limit-order traders and, as rightly pointed out by nanotube, are really the only group who are hurt by the existence of dark pool orders.

With these points in mind, I think the only sensible change to make is to look into nanotube's iceberg order idea.  It is a good comprise that balances the interests of both large traders and liquidity providers, while avoiding any disruption or damage to Mt.Gox's current success.

If we make any change at all, it should be that, and the next question becomes the details of the implementation...

[MagicalTux]

At this point, votes are:

Keep it as is    19 (29.7%)
Keep it, but remove "dark+normal" option    7 (10.9%)
Remove it    35 (54.7%)
Limit it to specific users    3 (4.7%)

I'm resetting the votes and adding an option for nanotube's "iceberg order".

[imanikin]

Will the iceberg option "carry an explicit cost penalty in the form of a larger execution cost charged by the market"?

[imanikin]

Please correct me if i misunderstood: if the iceberg option is implemented, the dark-pool-only option will still exist. That means that the MtGox Bitcoin price will still be distorted, because some of the largest funds will be able to skirt the fundamental forces of supply and demand in the price discovery at Bitcoin's main exchange.

Then my vote is still to remove the dark pool, and keep things simple in Bitcoin's natural style. One of the main reasons Bitcoin was invented was to give people an alternative to the corrupt manipulated monetary and exchange system that exists - not to become just another fancy extension of it, with all of its underhanded, hypocritical tricks and double-standards...

From the Main-Street perspective, imho, the biggest concern and impediment to widespread adoption is not the Bitcoin market volatility; rather it's a question of trust and suspicion that Bitcoin's market value is being distorted and manipulated by the same traditionally crooked entities we all know so well...

As already stated, dark pool artists and other large cheaters have plenty of resources to open dark exchanges, create bots, and find innumerable ways to cheat the markets and the rest of us. (Bernie Madoff – one of the founders of NASDAQ – proved that for all times...)  Let them do it in their own dark and shady places...

Let the Bitcoin price discovered at MtGox – Bitcoin's flagship exchange – be free of dark pool's distortion, beyond suspicion and reproach. 

[BitterTea]

Wow, spoken like a true politician. Lots of fear mongering, no facts.

[Psychoactive]

At this point, votes are:

Keep it as is    19 (29.7%)
Keep it, but remove "dark+normal" option    7 (10.9%)
Remove it    35 (54.7%)
Limit it to specific users    3 (4.7%)

I'm resetting the votes and adding an option for nanotube's "iceberg order".

MagicalTux, to whom will the iceberg order be open ?

[Cablesaurus]

There is no reason to remove something people can just accomplish with bots. That simply would further the gap between the extremely tech savvy and the "general public."

The Bitcoin community should be fast realizing that the relationship with technically inexperienced business owners, investors, and consumers should be nurtured if the currency is to prosper.

[toffoo]

Well, now we can get into the discussion of the details of the implementation of iceberg orders...

MagicTux, I'm glad to see that you've reset the poll and added the iceberg order option, but it seems you've assumed that it should just replace the "dark+normal" option and presumably the "dark only" option would remain as-is if this option wins?  Would people be happy with that?  Should the "dark only" option remain independent of the iceberg order option or should it be done away with entirely so that all orders must show at least "the tip of the iceberg" to the public order book?

One point to make that could potentially explain the initial blast and ongoing popularity of the "Remove it entirely" option in the original poll is that, as pointed out by nanotube, it is limit-order traders who are the only ones really hurt by the "dark+normal" option.  As potentially more frequent and more sophisticated traders, they are probably more likely to be following this thread carefully and voting, than the casual, occasional trader.  Also, the very casual, perhaps politically-inclided voter, who maybe has very little understanding of what a dark pool is and how it works, is probably also most likely to have voted for the "Remove it entirely" option.

In the original poll, I disliked the "remove dark+normal, but leave dark only" option for two reasons.  First, if public orders cannot interact with the dark pool only orders, the possibility exists for the two markets to become decoupled from each other.  For example, the best bid/offer in the public market might be 0.71-0.73 and the dark pool could be 0.75-0.80.  Since trades couldn't take place between the two, this situation could remain for some time, and most of the market wouldn't even be aware of it, because the crossed market is "in the dark".  Second, from the perspective of the large dark-only order placer, it would probably be pretty annoying to see trades printing on the public book thru your dark limit price, potentially thousands of BTC, and not participate in any of that volume because your order was left as "dark".

How often are "dark only" orders currently used?  Apart from MagicTux we cannot really be sure, but to put some data to this discussion, I ran a screen of the publicly available Mt.Gox trade reports.  Assuming that for a "dark only" order to get done, a trade report of at least 1000 BTC will have to be printed.  (Of course public, non-dark trades of 1000+ BTC and "dark+normal" trades could also occur, so these results are only potentially "dark only" trades.)  For the first three months of this year, I found about 1000 reported Mt.Gox trades of 1000 or more BTC, with many reports of exactly 1000 BTC and the largest being about 17,000 BTC.  So, I think it's fair to say that the "dark only" option is probably currently being used and is responsible for a reasonable portion of volume.

I personally think that it would probably be okay to leave the "dark only" option along side the iceberg option.  Neither casual market-order traders nor sophisticated limit-order traders would be helped nor hurt by the existence of the "dark only" orders.  Large traders would have the option to use "dark only" orders, with their potential downsides mentioned above, or iceberg orders.  Likewise, as long as iceberg orders are an option, I wouldn't be too bothered if "dark only" orders were eliminated entirely.


Now, as for the implementation details of iceberg orders, I have a few ideas:

First, for them to work properly I think we first need to fix the "minimum tick size" for Mt.Gox orders.  In fact, at the risk of getting side-tracked, if I was in charge of Mt.Gox development, I think this would be the order of my current priority list:

1. continue improving/automating bank transfers, PayPal, money movement, etc.
2. fix the floating point issues with USD and BTC balances and implement a minimum tick-size for orders
3. re-implement the WebSockets (or other real-time push) interface for bid/offer/last and market depth
4. figure out what to do with the dark pool
5. margin trading, options, etc.

As the wikipedia description points out, an important aspect of the iceberg order is this concept of the "queue" and price/time order priority.  In the current state of Mt.Gox, I do not really see various limit-orders queueing up in time order at fixed prices.  Because the minimum order tick size is so small, or undefined (is it 0.0001 or 0.000001?), limit orders currently stack in front of each other almost exclusively in price order.  If iceberg orders were implemented without changing this, it would still allow iceberg traders to step in front of limit-order traders and "penny" (or "teenie", if you're really ol' skool) them just like with "dark+normal" orders.

I propose a minimum tick size of 0.001.  This would allow ten spots for orders to queue up between the standard "big figures" of say 0.72-0.73 that we are used to seeing.  Implementing this would actually solve two problems.  It would allow for iceberg orders to function properly and also prevent the problem of bots or other speed traders from "pennying" limit-orders.

Will the iceberg option "carry an explicit cost penalty in the form of a larger execution cost charged by the market"?

What this means is that you'll be less likely to get your entire order done at a given price if you select the iceberg option because other public limit-orders will get a chance to participate at that given price (if a minimum tick size is implemented, and they are queued up there).  This encourages large order traders to leave a greater portion of their order "in public" and prevents them from "robbing" (entirely) limit-order traders.


Second, I propose that anyone can use iceberg orders, the commission structure is the same (no additional fees), and the minimum order size for an iceberg order is 500 BTC.  (i.e. the first 500 BTC "tip of the iceberg" will be shown on the public order book and anything beyond 500 BTC would be hidden)  Why 500 BTC?  In my experience this is about the size that begins to "move" or get a reaction out of the public order book.  And it would allow more traders to use the iceberg option than the current 1000 BTC minimum dark pool option.