Take the 2-minute tour ×
Quantitative Finance Stack Exchange is a question and answer site for finance professionals and academics. It's 100% free, no registration required.

I am new to this. I am confused on what consists of a tick data.

I have a trading platform in which I could collect data of exchange traded product like futures and stocks. While I am intending to use the platform for trading, I think I could collect the data directly from platform. The data consist of:

  1. Every traded quantity + time of transaction at exchange
  2. Every changes in the bid quantity, bid price, ask quantity, ask price + time at which these data reach my computer

While referring to some websites and literature regarding tick data, I am not sure:

  1. Is tick data consist of 1 or 2 or the combination of 1 and 2? Which data should I build my model on?
  2. Also, if I am using the combination of 1 and 2, any ideas on how to combine the two series as the timestamp of 1 and 2 is different (there is time lag between the traded time and the changes to bid and ask QUANTITY reflected on my comp)

Thanks

share|improve this question
    
Here's some useful links from google: trade-ideas.com/Glossary/Tick_Data.html investopedia.com/terms/t/tick.asp#axzz1xdsUTCJQ . Your other problem can be solved by just taking a little time to think out a solution. Maybe write a simple function in the language of your choice to match everything up based on the timestamps. –  Ockham Jun 13 '12 at 4:05
add comment

2 Answers

The phrase tick data can be a bit ambiguous (so double-check what you are getting when buying historical data). I would assume that "tick data" only means trade ticks, number 1 in your list. Number 2 would be called "bid/ask data" or "bid/ask ticks". Number 3, mentioned by Freddy, I hear called the order book, or "market depth data".

For your second question, record the time you receive the trade tick notification, and use that time in your model. Then the combined trade/bid/ask data is consistent. (Or use the difference to estimate latency, and subtract that from your bid/ask data timestamps - that has the advantage that it should match the official exchange data, if you need to patch data later.)

Going back to your first question, I'd suggest you build your model on just trade tick data initially. Then when you later add in bid/ask data it will take more CPU (much more data to deal with), but you may get better results. Or you may not. (There is an element of bluff in bid/ask prices, and even more so in market depth data, but a trade tick is real as it means someone has put their money where their mouth is).

share|improve this answer
add comment

It completely depends on your specific strategy model.

Tick data are generally not just trades but changes on the bid and offer as well. You can go a step further and define tick data for stocks as any change in the bid and offer in the whole order book, not just best bid/offer.

Regarding your second question, you can treat trades as simply a field for traded price and traded volume, thats it. Bid/Offer changes you use other fields such as bid price change/offer price change, bid volume change, offer volume change. If you manage the whole order book you would need to manage simply more fields if several order book internals change at the same time stamp.

Generally I advise you to always work off deltas, meaning you just look at changes from previous prices/volume or some use changes from market bid and offer prices. This will shrink the volume you need to deal with significantly. CPU resources are always cheaper than I/O resources.

share|improve this answer
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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