[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: spread indicator as ohlc?



PureBytes Links

Trading Reference Links

At 11:29 AM 10/13/2004, Gary Fritz wrote:

>But what if you have two slow issues, like the next-out contracts 
>of SP and ND?  There might not be any time when the two of them 
>update at nearly the same time.  So you can never calculate a 
>truly accurate spread.

Prof. Andrew Lo of MIT (with others) wrote a book a while back called the 
"The Econometrics of Financial Markets" that quantifies a lot of this with 
some pretty heavy mathematics.

The basic idea is that every market has a value vs. time and the listed 
trades are simply samples of the value from time to time (with added noise 
due to differences in reporting delays). You can extrapolate between samples 
to guess the correct value using correlation with some frequently traded 
market. For example, you could probably estimate the second-to-second value 
of the SP from the ES.

The attached picture shows the tracking between the NDX and QQQ. Notice that 
they track well within one percent except for the 4:15 PM (Eastern) value. 
This is because NDX stops at 4:00PM and QQQ trades until 4:15PM and a lot 
can happen in 15 minutes. 

So that if you compare the closing price of the NDX with the closing price 
of QQQ, you are building in a systematic error. It would be better to 
compare the closing price of the NDX with the 4:00PM price of QQQ.

And a stock might have a last traded at 3:05PM and that trade will be listed 
as it's closing price. The true value at the market close at 4:00PM would 
obviously depend upon what happened to the market between 3:05PM and 4:00PM.

As they say, "The devil is in the details."

Bob Fulks





Attachment: Omega.01.gif
Description: GIF image