|
PureBytes Links
Trading Reference Links
|
There are issues with timing and tick resolution that are causing
different performance results. I think the culprit is the spread.
The problem is that many people focus on the trade ticks and it's my
opinion that trade ticks give a misleading characterization of
moment-by-moment liquidity and fails to give you an accurate picture
on when you can beat the spread and when you must pay for it.
Because datafeeds disseminate information at slightly different times
and resolutions, that last tick at the close of a tick bar or time bar
will either fall on the bid or ask, or even another tier during a fast
moving market. Add all this up and you start to
get different manifestations of the same system using different
datafeeds and software products.
I don't think people are paying much attention on this slippage issue
partly because of the limitations of the software/datafeeds they use.
At the very least, their trading systems should have access to ES bid/ask, bid volume,
ask volume, and their order numbers. The CME disseminates this
information, and your systems should therefore have access to it.
This information beyond trade ticks can be useful. But it all
assumes, of course, that the CME stands by their promise that
orders are executed first-come-first served.
Wednesday, July 16, 2003, 10:38:30 AM, you wrote:
PK> Attain Capital did something somewhat like this... They compared how
PK> systems traded in TS2000, TS6, and TS7, and revealed the differing
PK> results people got in trend following systems based on how continuous
PK> contract data from various sources are constructed. I don't know if
PK> they looked at the tick level as well, but it might be worth a call to
PK> them. http://www.attaincapital.com/
PK> -peter
PK> -----Original Message-----
PK> From: robert pisani [mailto:pisani@xxxxxxx]
PK> Sent: Wednesday, July 16, 2003 1:22 PM
PK> To: omega-list@xxxxxxxxxx
PK> Subject: Re: support for 2000i
PK> I have DTN + TS 2000i. I've also been trying TS 7.3. I am running both
PK> concurrently. I see instances in the ES contract where TS 2000i is
PK> ahead
PK> of TS 7.3, and other instances where TS 7.3 is ahead of TS 2000i Do you
PK> know of any studies comparing the two?
PK> Bob
|