6 ms·

When you combine this article with others, like the fact that JP Morgan had zero days of trading losses over an entire quarter, you realize that the market is c

by jstalin 13y ago

When you combine this article with others, like the fact that JP Morgan had zero days of trading losses over an entire quarter, you realize that the market is clearly rigged by the big guys: http://www.zerohedge.com/news/2013-05-08/jp-morgan-has-zero-trading-losses-first-quarter http://www.zerohedge.com/news/2013-05-08/jp-morgan-has-zero-...

pfortuny 13y ago
Well, that is like statistically impossible. I mean either you are a magician or there is something wrong with you and the market. Assume the probability of a losing day is just .02 (i.e. 2%, which is rather good). A quarter has 60 trading days. Then .98^60 = .297+ which means it happens just once every 3 times. But that is a magician doing business. If the probability of a single losing day is .1 (one out of ten, good for humans, I guess), then the odds are .9^60 = .00179+ (once every 500 times) =(less than once in a century) So, really really spooky. Of course "once in a lifetime" events happen quite easily in the Stock Markets, but they tend to be negative (Long-term Capital Management, today's crisis...) Sorry (lots of edits because term=60 days)
- yetanotherphd 13y ago
  You are assuming that profits on successive days are uncorrelated. And in any case, all you have proven is that we can reject the null that the distribution of JP morgans daily returns this year, was different to the distribution of their daily returns in previous years. Let's leave guilty-by-rejecting-the-null-in-an-artificial-model to the courts and not let their logic infect HN too. All that's been shown is that JP Morgan have had 0 days with a trading loss. No one has actually shown they did anything wrong, or even explained what they might have done wrong.
- cperciva 13y ago
  Having zero days of trading losses in a quarter is easy -- just arrange your trades so that your risks are highly asymmetric. For example, you could sell billions of dollars worth of options tied to the fed rate; you'll make a small amount of money every day until the fed rate changes... and then you'll lose lots and lots of money all at once. Odds are that you'll have several quarters of consistent daily profits first though.
  - pfortuny 13y ago
    Yes, but we are speaking of real-life traders in banks, so their aim should not be "zero days of trading losses" but something different. So, as can be seen in http://www.zerohedge.com/news/2013-05-08/jp-morgan-has-zero-trading-losses-first-quarter http://www.zerohedge.com/news/2013-05-08/jp-morgan-has-zero-... quoted above by parent, the usual frequency is around .8-.9... I guess.
- sseveran 13y ago
  The probability of a losing day is actually far lower since most of the trades a bank has on are non-directional and net out.
  - pfortuny 13y ago
    Turns out (see http://www.zerohedge.com/news/2013-05-08/jp-morgan-has-zero-trading-losses-first-quarter http://www.zerohedge.com/news/2013-05-08/jp-morgan-has-zero-... above) that it is not that much lower than 0.1. IIRC, obviously.
    - sseveran 13y ago
      You (and the article) are assuming that all trades are created equal. At 95% accuracy you can easily never have a losing trading day. Plus given the massive amount of derivatives you can "mark to market" when convenient and end up with no losing trading days.
- roblev 13y ago
  My local supermarket has had zero days loss in trading in the last quarter. What does it do? It buys things for a cheap price and sells them at a higher price. Traders do exactly this, it is the service they provide. It is quite unlikely that all the traders in a big bank are going to lose money on the same day.
  - droidist2 13y ago
    That's what I was thinking. My bookie has a whole lot of good betting days. (I don't really have a bookie)
- JumpCrisscross 13y ago
  Couple factors that make your 2% i.i.d. loss probability unlikely: (1) Banks carry hedged books. You expect to win more frequently than 49 out of 50 times if you're instantaneously buying and selling FX in different regions, or delta hedging an option book. (2) The riskier assets also tend to be illiquid. This delays loss recognition, not necessarily out of deviousness, but because nobody realises the loan in question is a dud until someone tries to market it. (3) Market returns are not i.i.d. They are correlated. This makes streaks more likely.
teamgb 13y ago
Maybe the traders are hiding their losses from their bosses, shoving them into mysterious accounts, and hoping to turn them around at a later date. This is what Nick Leeson did which brought down Barings Bank, and what Jerome Kerviel did to SocGen. I'm sure it happens all the time.
ChuckMcM 13y ago
I've read a lot of articles like the OP and the zerohedge one. What rises out of noise is not so much 'good' or 'evil', 'correct' or 'incorrect', or even 'legal' or 'illegal' but an continual sense of anger that person A is getting away with cheating and I want to cheat too but can't because of condition Y. I suppose at some level its "fair" only if everyone can cheat. I watched a very strange film, a comedy, called Kung Fu Mahjong [1] in which the central theme are ladies that play Mahjong. And what struck me, perhaps as a westerner, or perhaps it was intentional, was that everyone in the movie cheats in some way or another, and everyone knows it. It reminded me of the kinds of things you would read about in the HFT blogs and forums. You could make a 'Kung Fu HFT' movie with the same theme. The bottom line is that I have very little sympathy for these guys. As the 'early release' scandal broke and then was documented it felt like more crap to shovel out of the door and not like any sort of progress toward a more durable system. [1] http://en.wikipedia.org/wiki/Kung_Fu_Mahjong http://en.wikipedia.org/wiki/Kung_Fu_Mahjong
- mcphilip 13y ago
  Market microstructure is such a massively complicated topic that few insiders, much less the SEC, thoroughly understand it. For instance, after NASDAQ went down for 3 hours on August 22, the SEC asked the NYSE and NASDAQ for a detailed timeline of events. Each exchange blamed the other. The SEC's response? It told the exchanges to cooperate and please fix the problem, declining to make any further public comment [1]. Trading at high speeds is like the wild west right now with little to no threat of law enforcement. [1]http://uk.reuters.com/article/2013/08/27/uk-nasdaq-halt-nyse-idUKBRE97Q04M20130827 http://uk.reuters.com/article/2013/08/27/uk-nasdaq-halt-nyse...
  - minimax 13y ago
    Market microstructure is such a massively complicated topic that few insiders, much less the SEC, thoroughly understand it. It's actually not super complicated. A limit order book is a straightforward thing. If you can do FizzBuzz, you can probably whip up a toy implementation of a matching engine in an afternoon. People think it's complicated because of all the FUD (c.f. the comments on this story), but the basic mechanics of how a modern exchange operates are remarkably simple.
gcb0 13y ago
That said, you still put your money there. Or on banks that put it there for you, for a fee no less. So, what is your point?

by jstalin 13y ago

pfortuny 13y ago

Well, that is like statistically impossible. I mean either you are a magician or there is something wrong with you and the market. Assume the probability of a losing day is just .02 (i.e. 2%, which is rather good). A quarter has 60 trading days. Then .98^60 = .297+ which means it happens just once every 3 times. But that is a magician doing business. If the probability of a single losing day is .1 (one out of ten, good for humans, I guess), then the odds are .9^60 = .00179+ (once every 500 times) =(less than once in a century) So, really really spooky. Of course "once in a lifetime" events happen quite easily in the Stock Markets, but they tend to be negative (Long-term Capital Management, today's crisis...) Sorry (lots of edits because term=60 days)

yetanotherphd 13y ago
You are assuming that profits on successive days are uncorrelated. And in any case, all you have proven is that we can reject the null that the distribution of JP morgans daily returns this year, was different to the distribution of their daily returns in previous years. Let's leave guilty-by-rejecting-the-null-in-an-artificial-model to the courts and not let their logic infect HN too. All that's been shown is that JP Morgan have had 0 days with a trading loss. No one has actually shown they did anything wrong, or even explained what they might have done wrong.
cperciva 13y ago
Having zero days of trading losses in a quarter is easy -- just arrange your trades so that your risks are highly asymmetric. For example, you could sell billions of dollars worth of options tied to the fed rate; you'll make a small amount of money every day until the fed rate changes... and then you'll lose lots and lots of money all at once. Odds are that you'll have several quarters of consistent daily profits first though.
- pfortuny 13y ago
  Yes, but we are speaking of real-life traders in banks, so their aim should not be "zero days of trading losses" but something different. So, as can be seen in http://www.zerohedge.com/news/2013-05-08/jp-morgan-has-zero-trading-losses-first-quarter http://www.zerohedge.com/news/2013-05-08/jp-morgan-has-zero-... quoted above by parent, the usual frequency is around .8-.9... I guess.
sseveran 13y ago
The probability of a losing day is actually far lower since most of the trades a bank has on are non-directional and net out.
- pfortuny 13y ago
  Turns out (see http://www.zerohedge.com/news/2013-05-08/jp-morgan-has-zero-trading-losses-first-quarter http://www.zerohedge.com/news/2013-05-08/jp-morgan-has-zero-... above) that it is not that much lower than 0.1. IIRC, obviously.
  - sseveran 13y ago
    You (and the article) are assuming that all trades are created equal. At 95% accuracy you can easily never have a losing trading day. Plus given the massive amount of derivatives you can "mark to market" when convenient and end up with no losing trading days.
roblev 13y ago
My local supermarket has had zero days loss in trading in the last quarter. What does it do? It buys things for a cheap price and sells them at a higher price. Traders do exactly this, it is the service they provide. It is quite unlikely that all the traders in a big bank are going to lose money on the same day.
- droidist2 13y ago
  That's what I was thinking. My bookie has a whole lot of good betting days. (I don't really have a bookie)
JumpCrisscross 13y ago
Couple factors that make your 2% i.i.d. loss probability unlikely: (1) Banks carry hedged books. You expect to win more frequently than 49 out of 50 times if you're instantaneously buying and selling FX in different regions, or delta hedging an option book. (2) The riskier assets also tend to be illiquid. This delays loss recognition, not necessarily out of deviousness, but because nobody realises the loan in question is a dud until someone tries to market it. (3) Market returns are not i.i.d. They are correlated. This makes streaks more likely.

teamgb 13y ago

Maybe the traders are hiding their losses from their bosses, shoving them into mysterious accounts, and hoping to turn them around at a later date. This is what Nick Leeson did which brought down Barings Bank, and what Jerome Kerviel did to SocGen. I'm sure it happens all the time.

ChuckMcM 13y ago

I've read a lot of articles like the OP and the zerohedge one. What rises out of noise is not so much 'good' or 'evil', 'correct' or 'incorrect', or even 'legal' or 'illegal' but an continual sense of anger that person A is getting away with cheating and I want to cheat too but can't because of condition Y. I suppose at some level its "fair" only if everyone can cheat. I watched a very strange film, a comedy, called Kung Fu Mahjong [1] in which the central theme are ladies that play Mahjong. And what struck me, perhaps as a westerner, or perhaps it was intentional, was that everyone in the movie cheats in some way or another, and everyone knows it. It reminded me of the kinds of things you would read about in the HFT blogs and forums. You could make a 'Kung Fu HFT' movie with the same theme. The bottom line is that I have very little sympathy for these guys. As the 'early release' scandal broke and then was documented it felt like more crap to shovel out of the door and not like any sort of progress toward a more durable system. [1] http://en.wikipedia.org/wiki/Kung_Fu_Mahjong http://en.wikipedia.org/wiki/Kung_Fu_Mahjong

mcphilip 13y ago
Market microstructure is such a massively complicated topic that few insiders, much less the SEC, thoroughly understand it. For instance, after NASDAQ went down for 3 hours on August 22, the SEC asked the NYSE and NASDAQ for a detailed timeline of events. Each exchange blamed the other. The SEC's response? It told the exchanges to cooperate and please fix the problem, declining to make any further public comment [1]. Trading at high speeds is like the wild west right now with little to no threat of law enforcement. [1]http://uk.reuters.com/article/2013/08/27/uk-nasdaq-halt-nyse-idUKBRE97Q04M20130827 http://uk.reuters.com/article/2013/08/27/uk-nasdaq-halt-nyse...
- minimax 13y ago
  Market microstructure is such a massively complicated topic that few insiders, much less the SEC, thoroughly understand it. It's actually not super complicated. A limit order book is a straightforward thing. If you can do FizzBuzz, you can probably whip up a toy implementation of a matching engine in an afternoon. People think it's complicated because of all the FUD (c.f. the comments on this story), but the basic mechanics of how a modern exchange operates are remarkably simple.

gcb0 13y ago

That said, you still put your money there. Or on banks that put it there for you, for a fee no less. So, what is your point?