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...)
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.
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-... quoted above by parent, the usual frequency is around .8-.9...
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.
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.
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.
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.
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)