Topic: GLBSE better, harder, stronger, faster, cheaper now with MAKER/TAKER - page 2. (Read 5540 times)

Meni: I just wanted to say, thank you for bringing mathematical clarity to everything that pops up on bitcointalk.  I'm rather embarrassed that I fell for some of these illusions... the brain works in mysterious ways :\

Hide by default but provide an ability to include them if the user chooses.

I'm using "dishonest" in a somewhat broad sense, maybe it's not the best term. Anyway, I applied it to an either/or situation - either one thinks this creates an improvement in a rational market (in which case I believe he is wrong), or he is using it to create an illusion of improvement, in which case he is dishonest.


Anyway, just in case my arguments for "it doesn't matter" are still unconvincing, think instead of a separation between buyer and seller fee, as compared to symmetric fees. Will an increased seller fee encourage more people to buy? No, it will only mean that sellers increase their offered listed price to compensate, so the trades and prices paid remain the same.

There are several wrong things with the chart. One of them is the the form of the graph makes it look like areas correspond to volume, when instead it's height. So if there are many trades with similar prices, they will look like a small total volume. The other is what you mentioned. bitcoincharts does this right - one bar chart with heights, one area graph of the cumulative volume, both of them showing only the relevant price range.

completely agree. I was wondering how to solve that problem myself. those silly bids are on a multitude of the assets. Id say the best method would be to hid anything more the 50%(?) off the average stock price.

request for some way to strip out the silly bids/asks from the charts, like the 2,000 bonds issued PPT series have 10,000 quantity bid @ 0.00000001 which just messes up the chart & makes the relevant part of the information to be not discernible, like just chart from 0.5 to 1.5 say - apart from this I like the new look & features, thanks!

https://glbse.com/asset/view/PPT.B

Meni is right. Right that it doesn't matter I mean. I don't think it's really dishonest though. The fee schedule is right there. Maybe it is dishonest like labeling a candy bar .99 and refusing to trade it for 99 cents is dishonest (demanding to collect a tax fee that wasn't included), actually that's pretty bad I guess. GLBSE is way less bad than that, the fee is explicitly mentioned on site unlike at 7-11.

+1

I'll try another way to explain it.

Let us define for every standing order two values: Maker execution price (MEP) and taker execution price(TEP).

The MEP is how much the person who placed the order pays (net, after deducting any fees) if it gets executed (this is positive for a bid, negative for an ask); The TEP is how much the person who executed the order pays (net. This is positive for an ask, negative for a bid).

Do you agree that the MEP and TEP are all that is to know about the order? It tells us exactly what will happen if the order is executed - how much the maker gets/pays, how much the taker gets/pays, and how much fee the exchange collects (their sum).

Now, the maker can choose the price. Whatever price he chooses, the MEP and TEP will be derived from it based on the specific fee structure used. Put another way, the maker can choose the MEP and TEP, with a constraint - their ratio is determined by the total fee collected by the exchange. It doesn't matter if the fee is 0.5% for both makers and takers, or 1% just from takers - the maker can choose any MEP and TEP as long their ratio is 101% (neglecting second order terms. More precisely, the fees are added multiplicatively and can be slightly different for bids and asks).

Now, once the MEP and TEP are determined, the number which the maker put in as the order price - and the number displayed in the order book - is completely irrelevant. The MEP and TEP contain all the information about what happens if the order is executed.

Now, a person coming to the exchange has two choices - execute an existing order with its MEP and TEP, or place an order of his own with MEP and TEP of his choice, with a constraint on the ratio. His choice does not at all depend on the internal workings of how the MEP and TEP are derived from the listed price - it all depends on how attractive is the current TEP, how much he wants it now, and how much he is willing to try getting instead a better MEP of his own, knowing that the probability of success depends on how attractive is the resulting TEP to future traders which will face the same choice, and his competition with the TEP of other orders.

The conclusion is that the fee structure doesn't matter. If everyone is rational, and the total fee is fixed, orders with the same MEP and TEP will be placed and executed.

In this language, when there is a maker/taker separation, the spread between the listed prices of lowest ask and highest bid will be small. But what really matters is the spread between the TEPs of the lowest ask and highest bid (since those are the prices people can actually execute), and these won't change.

So if it's all equivalent, what is the problem? It is mainly for traders who know how much they want to pay/get. Instead of just writing the amount, they will need to first figure out if there is already a matching order, and price it accordingly, since making will do a different translation to the price they actually get from taking. This again causes no functional difference - a software adapter could do the translation - but it just adds confusion.

You didn't prove anything you said they were both functionally the same.

That is hardly "broken" and hardly a proof when you assumed the 2nd buyer will always be accepting a market order.  The fact that there is no fee on adding liquidity will encourage at least some % of buyers to NOT accept the bid and instead place a new ask lower than the existing ask.

So you are right in that the total fees paid are the same you are wrong in your belief it won't have an effect on spreads.

Still you seem to be rather emotionally charged (usual for you) on something that is a standard across every major exchange in the world.  Exchanges want increases liquidity therefore rewarding those who increase liquidity .... increases liquidity. 

I mean do you think every single exchange in the world is confused and using a "broken model".  What would make you think that  liquidity would be improved if they disincentivized the act of adding to liquidity?

Like I said that was rhetorical.  I am going to step out now.

I've proven the claim, and you did not point out any problem with the proof.

Everything has a reason, it's just not always a good reason.

No, it does not. If the effective price of the ask (adjusting for fees) works for the buyer, he executes it. If not, he places a bid which will be at the same effective price no matter what is the fee structure.

This would all have been great if this fee structure actually encouraged placing standing orders. It does not. It's completely equivalent, it just changes the numbers displayed without changing what people actually pay.

Now you're just being insulting. Of course I'm not talking about changing the total fee, and I'm pretty sure in all my examples I kept the total fee invariant. In my last example, 1% taker fee + 0% maker fee = 0.5% trade fee + 0.5% trade fee.


Here's an exercise for you. Let's say the exchange starts with no orders. Make up rational players and describe what they want in life, and at what time. Describe what orders they place and execute, in either fee structure, like I did for John and Beth. Your challenge: Make it so the trades being executed (determined by how much people actually get and pay) is different in the two scenarios. I claim that you will not find such a scenario.

Well of course if the total fee if the same the total fee is the same HOWEVER even if the total fee is the same rewarding liquidity results in more competitive markets.  Tighter spreads, more standing orders, and deeper market depth.

Your insistence that MAKER/TAKER is broken is kinda silly.  There is a reason every single major exchange in the world uses that model.  It encourages people to place more competitive bids rather than just accept the bid or ask.  Each trader has the option.  Choose instant transaction and pay a fee (remove liquidity from the market) OR avoid a fee and risk non-execution by adding liquidity to the market.

If even a small % more traders chose the later (which rewarding liquidity encourages) you end up with tighter spreads and everyone wins.  Even those who just pay fee end up winning because the product is priced closer to true value.

Large market spreads reduce liquidity.  If you could only buy BTC at $5 and sell them for $4 you are much less likely to buy or sell.  Liquidity is reduced.  The market suffers and is less efficient.  While that is an extreme example anything that ADDS LIQUIDITY should be encouraged and fees based on increasing or reducing liquidity are a way to encourage more liquidity.  The whole market is improved.  The amount of the total fee is a total seperate topic.

1% round trip fee vs 0.5%  round trip fee vs 2.0% round trip fee.  You seem to be conflating the two things.  If GLBSE wanted to they could go to a MAKER/TAKER model and INCREASE fees.  One isn't mutually exclusive of the other.

As I explained, the tight spread is illusory. A 0 spread with 1% taker fee is equivalent to 1% spread (0.5% each way) with 0.5% trade fee.

It's possible that there's something about the details of their structure that makes this sensible. And it's also possible they're trying to trick people who are looking at some arbitrary metrics rather than how much they are actually paying. Either way, it doesn't belong here.

As long as you mention min tick rules, I think it would be great to have them - about 0.1% difference between allowed trade prices.

The maker/taker fee model is broken (at least this variation of it; I don't know if the idea is applicable in other contexts. Also, it's possible I'm wrong, but I'll need a very convincing case for it). Under assumption of rationality, it does not incentivize being a maker, and it does not improve liquidity. Of course, people aren't rational so it may have all sorts of effects in practice, but it's dishonest to do anything that capitalizes on people's naivety.

I've commented about this here, and the ensuing discussion may be of some interest. I'll repeat my main point here.

Case 1: The fee is a symmetric 0.25%.

John wants to sell 10 shares, as long as he gets at least 0.9975 BTC for each. He puts a sell order @ 1 BTC.
Beth wants to buy 10 shares, as long as she pays at most 1.0025 BTC for each.
John's order is just right for her - if it was any higher she wouldn't take it.
Beth executes this order, paying 1 BTC per share plus 0.25% fee.
John gets 9.975 BTC, Beth pays 10.025 BTC, GLBSE gets 0.05 BTC.

Case 2: The maker pays 0%, the taker pays 0.5%.

John wants to sell 10 shares, as long as he gets at least 0.9975 BTC for each. He puts a sell order @ 0.9975 BTC.
Beth wants to buy 10 shares, as long as she pays at most 1.0025 BTC for each.
John's order is just right for her - if it was any higher she wouldn't take it.
Beth executes this order, paying 0.9975 BTC per share plus 0.5% fee.
John gets 9.975 BTC, Beth pays 10.025 BTC, GLBSE gets 0.05 BTC.
(I have ignored second-order terms, 0.5% of 0.9975 BTC is really 0.0049875 BTC.)

Functional difference between the two cases: None.

It doesn't really matter if the taker just wants to buy/sell at whatever the current market price is. The maker is in competition with other makers; If makers have less fees, this will push all of them to lower the price (or increase when buying), so this will not effect what they actually gain from the trade. And it has no effect on what the taker eventually pays. Thus the maker/taker split doesn't increase making profits, and doesn't incentivize making. And the lower spread that is obtained is illusory, since the extra fee that will be paid by whoever wants to trade at the listed price negates it.

What it does do is create confusion and chaos. With symmetric fee, if I want to sell and get 0.9975 BTC per share, I just put an order at 1 BTC and am done with it. If there happens to be an existing order against which it can be executed, great. But with a maker/taker fee, I can't do that. I need to first look whether there already is an order to buy at 1.0025 BTC. If so, I need to sell at 1.0025 BTC to execute it, and I end up with 0.9975 BTC after the taker fee. If not, I need to sell at 0.9975 BTC, and this is what I'll get if the order is executed. If I mistakenly sell at 0.9975 BTC even when there is an existing buy order, we end up with two people who want to trade with each other but can't because of the artificial maker gap.


PS: A fee structure that might improve liquidity is volume discounts, like Mtgox. Market makers are likely to have large volumes, so this could help incentivize them.

This achieves one thing really, more market liquidity.

Liquidity providers can take advantage of the 0% trading fee.

People who want to trade get more liquidity at a better price and in theory the spread should close.

In the end I guess it can be summed up with "LURK MOAR!", right? As soon as you actually buy something, you get penalized, if you bid just 1 Satoshi off the current bid or buy, you have high chances that someone else will take the bait.

What problem shall a fee structure like this solve? It seems to me that it will bind more money to unfilled offers than necessary. Especially since offers on GLBSE are always backed by some actual amount of BTC in an account (I cannot put up a "Buy 100 of ABCD @ 1BTC" offer if I don't have 100 BTC not frozen in my account) this might give some huge order books but fewer trades imho.

We've made some changes over at GLBSE.com, which have significantly sped up the site.

On top of that we've added maker/taker for fees (we we're going to have both buyer/seller pay fees but figured this is better for all).

Maker/Taker means if you make an order and it's put on the orderbook (not taken up with another order) you pay 0%, if you take an order (i.e. you're order is matched up with another) you pay 0.5%.

Enjoy.