Talk:List of trading losses

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

Credit Suisse 2007/2008: 1 billion USD lost[edit]

Credit Suisse Group discovered pricing errors on bonds that will cut first-quarter profit by about $1 billion, prompting the biggest share decline in more than five years.

Switzerland's second-largest bank took $2.85 billion of writedowns on asset-backed securities after an internal review found mismarkings by a group of traders and credit markets worsened. The Zurich-based bank said in a statement today that it's assessing whether 2007 earnings were also affected. [1]

References


Wacky sorting[edit]

I noticed that the column USD Amount lost was sorting strangely in my client. Entries with an extra digit of precision were randomly inserted in the sequence out of order, producing the same BS whether sorted up or down. The pattern is weird, and unlike anything I've seen in three decades of computer science. It's not ASCII sort in either forward or reverse alignment. It's not numeric sort for floating point, and it's not integer sort after ignoring the digits preceding the decimal point. It doesn't appear to be a secondary key injecting nonsense.

!data-sort-type="currency"|2007 USD Amount Lost — better, but still got 0.296 out of order with 0.30
!data-sort-type="number"|2007 USD Amount Lost — I forget now, but not correct 
!data-sort-type="text"|2007 USD Amount Lost — mostly the same as before 

One time I implemented a binary tree thinking that strcmp(a,b) returned Boolean values corresponding to either a<b or a≤b. This was very wrong of me, the kind of mistake one makes in the first month of using a new language when the IBM PC was still a shiny new thing. My tree worked in adding and locating elements, so I thought it was good. I was implementing a DSL within something like a #define symbol table, where deletes were fairly uncommon. But I tried a few deletes anyway, and suddenly I couldn't find my elements any longer: because my misuse of strcmp() did not implement a consistent order relation, the internal tree rotation on delete did not preserve findability. I pulled my hair out for long hours, and then when I finally found the reason between keyboard and chair, I grew up fast on reading K&R a lot more carefully. That's about how weird what I'm seeing looks to me now.

I'll admit that my Firefox on this system is a little behind the times. Maybe someone else can check this out. — MaxEnt 23:40, 22 August 2020 (UTC)[reply]

  ~ Tom.Reding (talkdgaf)  14:27, 8 February 2021 (UTC)[reply]