Analyze raw tick data

Question

I'd like to work with raw tick data and naturally this data is unevenly spaced (for example, a couple of quotes are at the same second etc.)

For example

10:12:35 - 14.44
10:12:35 - 14.45
10:12:35 - 14.47
10:12:36 - 14.46
10:12:36 - 14.49
10:12:37 - 14.50

My question is regarding how to set this data to "fixed" intervals for valid math calculation. Though I read here couple of suggestions on what should be done, I'm not sure its clear to me:

1. Do I have to manipulate the raw data to have "clean" timestamps as the x-axis to work with common technical indicators?

2. Or can I refer to the index as the x-axis (and just ignore the timestamp)? Can I look at it as a "stream" of ticks to anaylze?

Answer 1

If you just want to run some simplistic technical analysis on quotes, then select the last quote for each unique timestamp. That will ensure that you don't have duplicate timestamps. If you must have it evenly spaced (i.e. no gaps from one second to another), then you can reuse the previous quote to fill-in the missing value.

Answer 2

To help you understand why you need to follow recipes (like chrisaycock's) just have a look at your tick data. You will find ticks clustered at some points in time while they seem scarce at others.

If you proceed with your recipe 2, you will lose those clusters of activity and stretch them out. In periods of low activity you will condense the market.

Most indicators you mentioned will expect to work over a window of time, simply because the results are so much more meaningful. In theory, you could apply the same algorithms to stretched or condensed indexed ticks, but the results will also be valid for those periods.

That means for a cluster of activity that makes the price jump from X to X+10 within 1000 ticks but only one second of time your indicator might tell you to sell on the 400th tick and buy on the 800th. But to implement this you would have to execute a round-trip within less than a second.

Now also this indicator's results are less comparable. Because in a period of low market activity the indicator might give you the same result (sell on 400, buy on 800), but now it's stretched over, say, hours, and it's much more likely that you can realise this trade.

Answer 3

It really depends on what you're trying to do, a solution as simple as agging the data to x-sec OHLC bars may suffice and from the sounds of things that's what you need. Now if you need to work with order book dynamics then tick data is fairly crucial, then what I'd do for analysis is reconstruct the orderbook from the ticks then just take snapshots of the book (again depends on your requirements).