0
$\begingroup$

I have Level 1 data that has already been aggregated into 0.5s buckets by the exchange.

I'd like to further aggregate the data into hourly and daily buckets. I plan to do this by simply taking a snapshot on each time interval i.e. at market open then market open plus one hour etc.

Q1.) Is there anything wrong in principle with this simple sampling strategy?

As always the devil is in the detail with these things. Some of the less liquid contacts may not trade for up to a couple of hours at a time what should I so in this situation? There will be several other issues I'm sure.

Q2.) Is there a set of benchmarks/best practice that someone can point me to for bucketing data where some points may be missing.

Q3.) Does it ever make sense to average some of the values (say over the last hour to estimate the daily close value? Is so are there any rules of thumb I can use to select the appropriate averaging period?

Thanks

$\endgroup$
1
$\begingroup$

Q1.) Is there anything wrong in principle with this simple sampling strategy?

I mean sampling is a valid strategy, it just may not be the best. WOuld a VWAP style price be better? Would just an average be better?

Typically when no trade has happened you can model the price as the last, average of the bid/ask spread, etc. The price you want depends on what you plan on using it for.

Q2.) Is there a set of benchmarks/best practice that someone can point me to for bucketing data where some points may be missing.

Again this can't be answered until we know what you want to use the data for.

Q3.) Does it ever make sense to average some of the values (say over the last hour to estimate the daily close value? Is so are there any rules of thumb I can use to select the appropriate averaging period?

I don't understand this question. You already have a closing price included with your data, why try to estimate it?

$\endgroup$
  • $\begingroup$ Response to comments to Q1.) Right now I am trying to aggregate the 0.5s tick data into a lower frequency such as hourly or daily to look for co-integrating relationships at these lower frequencies. In which case which of your suggestions above would be most applicable? The problem of missing data is pertinent! Would using average of bad/ask spread be appropriate when looking at co integration relationships? ore generally I want to be able to bucket my data in whatever frequency I desire and be confident that it it a correct a representation of the data at the chosen frequency that it can be. $\endgroup$ – Bazman Jan 25 '16 at 20:51
  • $\begingroup$ Response to Q3.) What I meant was that when estimating the daily closing price from tick data I could average over the last hour something like that. I was just trying to ascertain whether averaging would be heedful and if so some feel for the kind of averaging periods that might be appropriate (i.e 1hrs worth of tick data when estimating the close I mentioned earlier)? $\endgroup$ – Bazman Jan 25 '16 at 20:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.