Suppose you're making the following assumption: Coca and Pepsi prices should roughly move together, because they essentially do the same thing. For simplicity, suppose they're both normally 100 USD each. Once in a while, for various reasons, Coke goes to 105 and Pepsi goes to 95. So, guessing that the long run price should return, you sell Coke and buy Pepsi. That's a statistical arb. (Not a pure arb, which would be sth like sell in London to buy in NY at the same time). Anyway, if you're and arb trader, this is the kind of thing you're doing all day long. You see Pepsi at 98 and coke at 102, you do the opposite to what I just said. And get out for a profit, hopefully. Of course the spread might move against you, that's a risk. But that's different from if both Coke and Pepsi moved up, to say 110, because you're long one and short the other.
Now this guy was on an ETF desk. ETFs are essentially the same as a stock, but they follow the index. So if the basket of stocks is under the index price, you could buy all the stocks and sell the ETF, and liquidate for a profit when the prices matched again.
The thing is, your guy at UBS has to have the flexibility to buy some ETFs sometimes, and to sell the individual shares. So if he just buys a load of index, who is gonna know if he "forgets" to sell some shares? In that case, he's just got a long position in the index, with nothing offsetting. Much bigger risk.
(My examples are simplified, actual arb trading has a load of complications.)
Profits are quite variable at a big bank in turmoil.
I can understand your surprise, but he's trading financial products. Most of them are simply numbers on a screen. There's no extra effort involved in buying 100M bucks worth of something over say 100K. It's not like if he bought and sold widgets which actually have to manufactured and delivered. And for the same reason, the bosses can't see him doing it either.
And if you wonder why it can be done at all, it's because you have to move a lot of money around to make money in this business. Suppose you're doing an arb of 100M of eg Coca Cola vs 100M of Pepsi, your risk is pretty low. (One cancels the other). But if you were to do just one of the two, you're suddenly taking a huge risk. But who is gonna be able to tell from casual observation?
Big banks in particular have a bunch of stuff going on, hirings and firings, new systems, that type of thing.
They made you PAY for losing a bag? Seriously? That's a friggin risk of the business. What do they do if you mess up an F/A 18?
I've seen people screw up in trading, and quite simply nothing happens to them. When I was starting out the bosses were always telling us "mistakes cost money" but in the end, they just ate the losses. Part of the business is that people will forget to repo, they'll forget to roll their overnight currency positions, and they'll sometimes accidentally trade millions of dollars by clicking in the wrong place. It happens.
First of all, he'd worked in the back office, so he'd know both people and procedures.
Second of all, anyone who's ever worked in finance can tell you big banks are chaotic. It's not really that strange that he can go about his business undetected for a while, because there's loads of traders with loads of portfolios. And most people on one desk are not going to be experts on the business of another. UBS has had a number of restructurings since the financial crisis. People are moved on, some desks are closed, some are merged, it's gonna be a mess. Makes it easier to hide.
Third, risk officers are not what you think. They are not the internal police, vigilantly keeping an eye out for every possible transgression. They look at the positions, calculate the risk (big can of worms, don't ask), and when someone is over their limit, they show up at the trader's desk and are told to fuck off.
Finally, it is not at all clear that technology played an important role in this fraud. Yes, some HFT market makers trade ETFs, but it's not clear his desk was. That doesn't mean a software error caused it, or that the fraud could not have occurred without whatever system he was using. From efinancial (which you need a subscription to read) the latest rumour is that he messed up a hedge in EURCHF, and his attempt to fix it made it worse.
Apparently it's standard for big firms to strip each layer and SEM it. Sounds immensely complicated to me. And even when you have the design, how the heck do you work out what it can do? Instruction set and all that? This might be particularly hard to do if it wasn't something that had a manual, like a missile.
Here's a technical thing I always wondered: if you capture a silicon chip from someone (missiles would have some of those, right?), how do you go about figuring out how it works, and what it's meant to do? Is there a way to figure out the layout of the dopants, or is it a matter of looking at the external connections and making an educated guess? You won't know what voltage it's meant be used at, whether there's a clock, etc...
But this is a very general fact about any business that's new: nobody knows whether it will provide anything useful, and it's very possible that only a few benefit.
The high frequency of trading has a lot to do with the business of market making. You're obliged by the exchange to provide prices, so if you can't move them often, you're more likely to get run over. If you're more likely to get run over, you make wider prices. That's not good for the end customer.
Now, suppose you do have an investor who's done the analysis. Those guys tend to be funds that trade in big size. When one of them comes in with a trade, what's the market maker supposed to do? The answer is that the MM doesn't do his own fundamental analysis and stick to that price. That's not sensible when you're obliged to make a price. What if you're wrong by a dollar? You'll be wiped out. The MM will sound out the market and try to disperse the large order to the rest of the market, including other MMs. This information transfer needs to be fast, or the risk of holding a position will grow. Bear in mind the MM knows the client finds it profitable to cross the spread.
Well, it's hard to do anything other than speculate when you don't have the research. My guess is that commodity markets, on a mechanical level, work in much the same way as stocks: there's a computer that takes in all the orders and sends our the orderbooks and fills. So a number of firms involved in stocks will have chosen to also do commodities, yielding a similar result.
Fees are high for retail people, of course. It's the same effort, but for less commission. How's that different from wholesaling any other product? You show up asking for repeat large orders of (widget), and you'll be able to negotiate a deal.
Rollbacks are fishy, true, but consider that there's two guys being rolled back. One would have won, one would have lost. Sometimes it's the HFT guy that would have won, sometimes he would have lost. It's more a question of keeping an orderly market than helping certain people. Keep in mind also that not all HFT guys are the same way on every single trade.
This is not possible in all markets, and is definitely fishy. But also, because most markets do not allow this, it is not the main business of HFT traders.
Actually, what you are describing is called market making. That mechanism does not depend on algorithms. It would exist even if there were no computers involved.
And actually, the MM IS doing something useful. If you show up on day 1 and want to sell, but there's no MM, the guy who shows up on day 2 and wants to buy will also leave having not traded. The MM is carrying over the risk between periods of high supply and periods of high demand.
Well, one could argue that the real problem with the having stock market is decimation of financial interest. Why would a shareholder care about how the managers of BigCorp behave if 1) They own a very small share of it 2) Their share in it is a very small proportion of their total wealth?
That is possibly the best fair criticism I've read about HFT yet. But it's only incidental: You say that market liquidity is bad for business, because it causes short-termism. And HFT makes the market better at what is does, so HFT is incrementally damaging for business.
Well, you may be right about short-termism, but I believe the discussion mainly centered on whether HFT was good for the stock market. I agree for some firms quarterly announcements create an incentive that isn't so great for them, but those firms are free to stay private. In fact, there's a whole industry of people who take public firms private in order to run them better. (Interestingly they often unload by going public again).
Slashdot Mirror
Suppose you're making the following assumption: Coca and Pepsi prices should roughly move together, because they essentially do the same thing. For simplicity, suppose they're both normally 100 USD each. Once in a while, for various reasons, Coke goes to 105 and Pepsi goes to 95. So, guessing that the long run price should return, you sell Coke and buy Pepsi. That's a statistical arb. (Not a pure arb, which would be sth like sell in London to buy in NY at the same time). Anyway, if you're and arb trader, this is the kind of thing you're doing all day long. You see Pepsi at 98 and coke at 102, you do the opposite to what I just said. And get out for a profit, hopefully. Of course the spread might move against you, that's a risk. But that's different from if both Coke and Pepsi moved up, to say 110, because you're long one and short the other.
Now this guy was on an ETF desk. ETFs are essentially the same as a stock, but they follow the index. So if the basket of stocks is under the index price, you could buy all the stocks and sell the ETF, and liquidate for a profit when the prices matched again.
The thing is, your guy at UBS has to have the flexibility to buy some ETFs sometimes, and to sell the individual shares. So if he just buys a load of index, who is gonna know if he "forgets" to sell some shares? In that case, he's just got a long position in the index, with nothing offsetting. Much bigger risk.
(My examples are simplified, actual arb trading has a load of complications.)
Profits are quite variable at a big bank in turmoil.
I can understand your surprise, but he's trading financial products. Most of them are simply numbers on a screen. There's no extra effort involved in buying 100M bucks worth of something over say 100K. It's not like if he bought and sold widgets which actually have to manufactured and delivered. And for the same reason, the bosses can't see him doing it either.
And if you wonder why it can be done at all, it's because you have to move a lot of money around to make money in this business. Suppose you're doing an arb of 100M of eg Coca Cola vs 100M of Pepsi, your risk is pretty low. (One cancels the other). But if you were to do just one of the two, you're suddenly taking a huge risk. But who is gonna be able to tell from casual observation?
Big banks in particular have a bunch of stuff going on, hirings and firings, new systems, that type of thing.
They made you PAY for losing a bag? Seriously? That's a friggin risk of the business. What do they do if you mess up an F/A 18?
I've seen people screw up in trading, and quite simply nothing happens to them. When I was starting out the bosses were always telling us "mistakes cost money" but in the end, they just ate the losses. Part of the business is that people will forget to repo, they'll forget to roll their overnight currency positions, and they'll sometimes accidentally trade millions of dollars by clicking in the wrong place. It happens.
Pretty simple theory: you see someone with similar skills as you making much more money. How do you feel?
To compare, people here tend not to whine when sports stars make a buttload. You can't really be jealous of someone you don't compare yourself to.
Citation needed?
They've been having board-level meetings about improving their risk management. I'd expect head to roll.
First of all, he'd worked in the back office, so he'd know both people and procedures.
Second of all, anyone who's ever worked in finance can tell you big banks are chaotic. It's not really that strange that he can go about his business undetected for a while, because there's loads of traders with loads of portfolios. And most people on one desk are not going to be experts on the business of another. UBS has had a number of restructurings since the financial crisis. People are moved on, some desks are closed, some are merged, it's gonna be a mess. Makes it easier to hide.
Third, risk officers are not what you think. They are not the internal police, vigilantly keeping an eye out for every possible transgression. They look at the positions, calculate the risk (big can of worms, don't ask), and when someone is over their limit, they show up at the trader's desk and are told to fuck off.
Finally, it is not at all clear that technology played an important role in this fraud. Yes, some HFT market makers trade ETFs, but it's not clear his desk was. That doesn't mean a software error caused it, or that the fraud could not have occurred without whatever system he was using. From efinancial (which you need a subscription to read) the latest rumour is that he messed up a hedge in EURCHF, and his attempt to fix it made it worse.
Thanks dude. That's pretty cool. I also found this link:
http://www.edn.com/article/457681-IC_reverse_engineering_a_design_team_perspective.php
Apparently it's standard for big firms to strip each layer and SEM it. Sounds immensely complicated to me. And even when you have the design, how the heck do you work out what it can do? Instruction set and all that? This might be particularly hard to do if it wasn't something that had a manual, like a missile.
Here's a technical thing I always wondered: if you capture a silicon chip from someone (missiles would have some of those, right?), how do you go about figuring out how it works, and what it's meant to do? Is there a way to figure out the layout of the dopants, or is it a matter of looking at the external connections and making an educated guess? You won't know what voltage it's meant be used at, whether there's a clock, etc...
'free to fuck everyone over as soon as you get your hands on some privilege'
You mean no different from any system of governance?
But this is a very general fact about any business that's new: nobody knows whether it will provide anything useful, and it's very possible that only a few benefit.
The high frequency of trading has a lot to do with the business of market making. You're obliged by the exchange to provide prices, so if you can't move them often, you're more likely to get run over. If you're more likely to get run over, you make wider prices. That's not good for the end customer.
Now, suppose you do have an investor who's done the analysis. Those guys tend to be funds that trade in big size. When one of them comes in with a trade, what's the market maker supposed to do? The answer is that the MM doesn't do his own fundamental analysis and stick to that price. That's not sensible when you're obliged to make a price. What if you're wrong by a dollar? You'll be wiped out. The MM will sound out the market and try to disperse the large order to the rest of the market, including other MMs. This information transfer needs to be fast, or the risk of holding a position will grow. Bear in mind the MM knows the client finds it profitable to cross the spread.
Well, it's hard to do anything other than speculate when you don't have the research. My guess is that commodity markets, on a mechanical level, work in much the same way as stocks: there's a computer that takes in all the orders and sends our the orderbooks and fills. So a number of firms involved in stocks will have chosen to also do commodities, yielding a similar result.
Ok, have you got such a paper?
Wait a minute. If you have some predefined determinant of value, why have a market at all? You can't force people to value things the way you want.
Fees are high for retail people, of course. It's the same effort, but for less commission. How's that different from wholesaling any other product? You show up asking for repeat large orders of (widget), and you'll be able to negotiate a deal.
Rollbacks are fishy, true, but consider that there's two guys being rolled back. One would have won, one would have lost. Sometimes it's the HFT guy that would have won, sometimes he would have lost. It's more a question of keeping an orderly market than helping certain people. Keep in mind also that not all HFT guys are the same way on every single trade.
This is not possible in all markets, and is definitely fishy. But also, because most markets do not allow this, it is not the main business of HFT traders.
Actually, what you are describing is called market making. That mechanism does not depend on algorithms. It would exist even if there were no computers involved.
And actually, the MM IS doing something useful. If you show up on day 1 and want to sell, but there's no MM, the guy who shows up on day 2 and wants to buy will also leave having not traded. The MM is carrying over the risk between periods of high supply and periods of high demand.
You're talking about flash orders? I find it fishy too, but FWIH it's on it's way out.
Well, one could argue that the real problem with the having stock market is decimation of financial interest. Why would a shareholder care about how the managers of BigCorp behave if 1) They own a very small share of it 2) Their share in it is a very small proportion of their total wealth?
Sure, I can see how some of those things are worrisome. Especially the bit about education costing a lot.
But I don't really get what your opinion is? We should regulate all these things, or we should sit down and have a think about them?
And why should we end something that we don't think is wrong?
Also, why does gambling still exist? Surely something is not right.
That depends on which speculator you mean, of course.
That is possibly the best fair criticism I've read about HFT yet. But it's only incidental: You say that market liquidity is bad for business, because it causes short-termism. And HFT makes the market better at what is does, so HFT is incrementally damaging for business.
Well, you may be right about short-termism, but I believe the discussion mainly centered on whether HFT was good for the stock market. I agree for some firms quarterly announcements create an incentive that isn't so great for them, but those firms are free to stay private. In fact, there's a whole industry of people who take public firms private in order to run them better. (Interestingly they often unload by going public again).
And gambling is wrong?