AP Twitter Hack

[DGuller]

I'm no expert on the subject, but I don't think we're talking about the old program trading that caused a 20% one day drop in the market back in the 80s.

Just take a look at this: people thought there was a bombing of the White House, the market dropped 1% and instantly recovered.  Hardly a good data point for fragilizing the market.

One of the problems with algorithm trading is that it causes liquidity to drop to near zero any time something uncertain happens;  that is, when you really don't want liquidity to drop off.  Algorithms simply trip into shutdown mode to avoid trading during an uncertain time.  When you have no bid or ask offers worth talking about, prices can get to pretty much any level.  That in turn can lead to all sorts of margin calls based on phantom losses and the usual chain reactions associated with them.

Then there are just unpredictable dynamic things that can happen when a couple of algorithms interact with each other and create a vortex of insanity.  When Amazon algorithm prices a used book about flies at $20,000,000, because it's programmed to beat another algorithm that priced the book at $18,000,000, it's amusing.  When such thing happens in the stock market, it can hurt real people who actually do something productive rather than figure out how to grab more of the bid/ask spread for themselves.

[DGuller]

How about the 2010 "Flash Crash"? That was over a 1000-point drop IIRC.

[DGuller]

Here's a fun reading about algorithm trading:  http://www.michaeleisen.org/blog/?p=358.

[Warspite]

When Amazon algorithm prices a used book about flies at $20,000,000, because it's programmed to beat another algorithm that priced the book at $18,000,000, it's amusing.  When such thing happens in the stock market, it can hurt real people who actually do something productive rather than figure out how to grab more of the bid/ask spread for themselves.

Interesting - I always wondered why certain used books had absolutely insane valuations (like £1800 for a used edition of some random old political science textbook).

[Valmy]

What should stock markets be all about Squeeze?

Raising capital for various business ventures?  I mean that was sorta why it was invented in the first place.

[grumbler]

Algorithmic (computer) trading models are no different from any other kind of trading model, except that they react more quickly.  Stupid people lose money faster by depending on algorithms than they do by depending on Aunt Jessie's bone aches, but I don't see why that is bad - the sooner the fools lose money, the faster the market is rid of their influence.  There will always be some fools in the market, though, so the market will never be completely rid of their influence.

The Flash Crash was an example of fools losing money.  For every dollar the fools lost selling stocks in a panic for five minutes, the smart people made a dollar buying those stocks in the five minutes and the twenty that followed.  The Flash Crash was interesting for the speed with which the crash occurred and the speed with which the recovery occurred, but other than the timeline, it wasn't all that interesting.  The stock market had fallen and risen in the past, even in the time period before computers were invented.

[The Brain]

I fail to be outraged by computer trading.