Frequency of cancellation intraday?

[mizhael]

Hi all,

I am looking at some intra-day level 2 data. Supposedly the data are order arrivals/updates. The documentation wasn't clear about what are the cancellation orders, so I am doing some experiments to test several possibilities of identification of cancellation orders.

Does anybody roughly have an idea about the number/frequency of cancellation orders intraday, from premarket to aftermarket in one day, and the proportion of cancellation orders to non-cancellation orders? For example, for QQQQ?

Here is the result from my data file, based on my current assumption about the identification of cancellation orders:

Total Number of Order Arrivals: 31688481
Number of Cancellation Order Arrivals: 9148736

Is it possible for Nasdaq level 2 order arrival data to have about 1/3 of its total arrivals/updates to be cancellation orders?

 

[gasy]

you are going about it the wrong way

looking for specific edge will lead to long term heartache, been down that road

instead .......focus on nature of the market

its basic nature, followed by nature of specific instrument

 

[R. Raskolnikov]

Excellent post, can't believe I am complimenting you but I am

Quote from gasy:

you are going about it the wrong way

looking for specific edge will lead to long term heartache, been down that road

instead .......focus on nature of the market

its basic nature, followed by nature of specific instrument

 

[auspiv]

how are you identifying the canceled orders? as far as i know, there isn't a way to tell...

 

[maxpi]

The exchanges set limits on the ratio of cancellations to completed orders. If a Broker can't stay below the maximum they get fined.. so brokers will have a ratio that they allow. Interactive has a forty to one limit [if memory servers] on some instruments, I forget which though...

 

[mizhael]

Quote from maxpi:

The exchanges set limits on the ratio of cancellations to completed orders. If a Broker can't stay below the maximum they get fined.. so brokers will have a ratio that they allow. Interactive has a forty to one limit [if memory servers] on some instruments, I forget which though...

Great piece of information. Could you please give me more verified details? Where to find documents for this?

It's basically saying that my current identification of cancellation doesn't make sense. It cannot be 1/3 of total order arrivals/updates.

 

[propseeker]

i've never heard of cancellation fees for nasdaq, i don't think they have them. a friend of mine really banged on their servers hard for almost a year, and at some point they just told him to stop it, but never any fees. it IS common for retail brokers to implement these limits, as execution bandwidth is usually shared.

regardless, that info is irrelevant to your research. 1/3 cancel ratio on a major ETF... i wouldn't be suprised at all. just double check your cancel assumption algo, and see if it makes sense.

 

[rosy2]

Quote from mizhael:



Is it possible for Nasdaq level 2 order arrival data to have about 1/3 of its total arrivals/updates to be cancellation orders?

from an orc software article...

Using Order-Trade Ratios growth on the London Stock Exchange (LSE) as proxy for the major European equity markets, JWG-IT, a think tank for EU-driven IT change in financial services, predic¬ts that an average of 25 orders will be needed to process a single trade (25:1) by the end of 2009.

 

[mizhael]

Quote from rosy2:

from an orc software article...

Using Order-Trade Ratios growth on the London Stock Exchange (LSE) as proxy for the major European equity markets, JWG-IT, a think tank for EU-driven IT change in financial services, predic�ts that an average of 25 orders will be needed to process a single trade (25:1) by the end of 2009.

What's the implied cancellation ratio (to other quotes)?

 

[promagma]

I know nasdaq has fees for excessive use of DOTI routing.
And options trading is subject to cancellation fees.

Any others?