The one thing I never understood...

Quote from bluelou:

tommaso,
I understand now that you're looking at the number of ticks per unit of time, "tick rate", for liquidity purposes. Transaction rate is the arrival time b/t ticks. ACD is meant as a tool to analyze whether changes in transaction rate contain information. The evidence suggests that they do.

I disagree w/your comments on academic research. If you know what you're looking for you can separate the good from the bad research in a systematic way.

-Lou
More...


oops look like I probably did a gaffe. I hope not. Didn't I? :-)

[ Actually you might even be an author! :-)) ] In any case absolutely no offense was intended. I was including myself too in that general remark.

I just wanted to say that sometimes for theoretical models we can afford to do or propose models or assumptions which we would never use if we had to put our money on that.

It's like when you are just paper trading. You may do things that you would never do in real trading :-)


By the way, my initial doubt was how do you relate bid/ask ticks with transactions?

Does anybody have more information about the actual mechanism behind those bid/ask ticks ?

T


PS
wow take a look at DECK today ! This happens when i say that trend does not exist... :-)
 
Quote from bluelou:

tommaso,
I understand now that you're looking at the number of ticks per unit of time, "tick rate", for liquidity purposes. -Lou
More...

No, not at all. I may not have expressed myself clearly.

Simply, since your future actions are based on past ticks, it's clear that a lower tick rate would imply less data in a given time. And therefore your decisions may be less reliable if based on a smaller amount of data.

Also, spikes and sudden direction changes are more common in presence of low tick rate.

I do not see a sure connection (or proportionality) between tick rate and liquidity.

T
 
Tomasso,
This might be part of an answer to your question - but from a futures perspective. Depending on the data provider CME data will count bid/ask quotes as ticks. I spoke to the data group at CME to try to figure out what was going on. Though the answer still wasn't clear to me after several attempts this is what I gathered:

If a trade was filled outside of the BBBO then I think you'll get either get a bid/ask quote or that tick and the bid or asked quote closest to that tick. It wasn't quite clear to me.

In the case of the CME, their raw feed does not screen out bid/ask quotes recorded as ticks. However, real-time data providers do have the option of screening out these quotes. Again, this is what I was told by the CME.

For instance, IQFeed provides ticks only, no bid/ask. But TradeStation and TT give you the raw feed which includes bid/asked quotes. With the raw feed I was getting at least 15% more ticks than with IQFeed's trade-only ticks.

I suggest you call the exchange in question or test different data providers.

Personally, I'm using IQFeed for real-time data since I know that the ticks are trades-only - for futures at least. For historical tick data I use CQG Data Factory. When I download CQG's data and select "trades-only" this ends up being a near perfect match to IQFeed's tick data.

I just started using both providers for equities and I haven't seen anything obvious that would indicate that quotes are mixed in the tick data but I haven't taken a real close look either. I'll get to it in the next few days.

I hope this helps.

-Lou
 
How about this method? :

Use a very simply trend following system and observe the equity curve it produces. When it is moving up, you are in trend mode and should trade it.

Do the same for a choppy system, maybe even use the same strategy only inversed.

Set up some sort of a condition such as, when the current system has a draw down of 15%, switch systems... or something like that. This may not make a ton of money in one market, but it could do OK when traded against a number of markets.
 
Quote from bluelou:

Tommaso,
....
I just started using both providers for equities and I haven't seen anything obvious that would indicate that quotes are mixed in the tick data but I haven't taken a real close look either. I'll get to it in the next few days.

I hope this helps.

-Lou
More...

I can provide an example of the price ticks that are arriving. (this is a record of the very first chunk of price ticks for a stock when you just connect. On each line the first 4 numbers are a timestamp h/m/s/ms, the last 2 a ticktype code and price) :

9 28 44 796 4 166.95
9 28 44 796 9 165.15
9 28 44 843 1 166.87
9 28 44 843 2 166.90
9 28 44 906 16 192.24
9 28 44 906 15 158.09
9 28 44 906 18 192.24
9 28 44 906 17 115.44
9 28 44 921 20 202.96
9 28 44 921 19 111.62
9 28 45 390 4 166.90
9 28 45 453 1 166.89
9 28 45 453 2 166.93
9 28 46 656 1 166.91
9 28 47 906 2 166.96
9 28 48 203 2 166.98
9 28 48 484 4 166.96
9 28 50 406 1 166.89
9 28 50 421 2 166.94
9 28 53 765 4 166.89
9 28 56 531 1 166.86
9 28 56 531 2 166.93
9 29 0 546 1 166.87
9 29 4 468 1 166.86
9 29 10 437 1 166.85
9 29 10 437 2 166.92
9 29 13 437 1 166.86
9 29 14 62 4 166.90
9 29 14 62 1 166.90
9 29 14 78 2 166.93
9 29 18 984 1 166.86
9 29 19 265 4 166.93
9 29 19 281 1 166.88
9 29 19 296 2 166.94
9 29 21 203 4 166.88
9 29 21 218 1 166.87
9 29 21 218 2 166.93
9 29 23 218 2 166.90
9 29 26 421 4 166.90

...

1 bid 2 ask 4 last, etc.


T
 
Quote from Corey:

Instead of trying to write the holy grail indicator, why not spend your time identifying the current market condition? What does that mean? It means, write a system that identifies whether the market is favoring a trend following system, a mean reversion system, a fading system ... et cetera.
More...

This is what I do. I trade 9 systems concurrently, of three distinct types that are stastically non correlated. This provides a consistent risk adjusted return and lower drawdown. Since the systems are non correlated, all can be traded from the same float as each utilises captial at different times.

If you looked at my performance to date, you would be saying "what credit crisis?".

A couple of hints:
Track more than the instrument you are intending to trade - that is track the index or advance decline line etc, and the stock.
Work on improving the performance of all the systems as they work together, not on individual systems - a system that has an average performance can significantly improving the group performance.
If you take this approach, trade all the systems all the time or don't trade at all. Currently i can't take short trades (gov ban), thus I don't trade any systems. Hello holidays and beers at market open!

pneuma
 
I've been a bit busy, but catching up on these posts has been nice. Definitely a good discussion going on ... and no trolls yet (crosses fingers).

pneuma, I would be interested in hearing how your nine systems vary. Obviously, that might be too much information to give away, so I understand if you don't want to speak up about it. I assume that each system, given that they are 'non-correlated', trades a different strategy. Without getting into complex arbitrage, I could only count on one hand the types of strategies I would employ ... they perhaps they are too generic. If I say trend, reversion, and fade ... you might have "trend breakdown", "trend following", "fade exponential", ... et cetera...
 
Quote from Corey:

I've been a bit busy, but catching up on these posts has been nice. Definitely a good discussion going on ... and no trolls yet (crosses fingers).

pneuma, I would be interested in hearing how your nine systems vary. Obviously, that might be too much information to give away, so I understand if you don't want to speak up about it. I assume that each system, given that they are 'non-correlated', trades a different strategy. Without getting into complex arbitrage, I could only count on one hand the types of strategies I would employ ... they perhaps they are too generic. If I say trend, reversion, and fade ... you might have "trend breakdown", "trend following", "fade exponential", ... et cetera...
More...

You have miss read, 9 systems over 3 strategies and there aren't that many styles of strategies as you have just mentioned. Thus systems within a style are semi-correlated - ie vary in time frame and general approach (but long trend following is still long trend following), and strategies are uncorrelated - ie long, short, trend, reversion etc.

Everything you need to put it together is in this thread already from the various posters. All i can say is keep it really simple - every system i use could be explained on a serviette.

good luck, pneuma
 
Quote from tommaso:

No, not at all. I may not have expressed myself clearly.

Simply, since your future actions are based on past ticks, it's clear that a lower tick rate would imply less data in a given time. And therefore your decisions may be less reliable if based on a smaller amount of data.

Also, spikes and sudden direction changes are more common in presence of low tick rate.

I do not see a sure connection (or proportionality) between tick rate and liquidity.


T
More...


Interesting discussion here. While they are not directly addressing "tick rate," per se., there is a related argument that large price fluctuations (instantaneous volatility) are not so much determined by large orders or volume/tick rate in and of itself, but rather by available liquidity or lack thereof, and its continuity on the order books. "Price fluctuations are driven almost entirely by liquidity fluctuations."

http://arxiv.org/PS_cache/cond-mat/pdf/0312/0312703v2.pdf



There are quite a few papers and studies (even entire simulated exchanges) addressing some of the concepts discussed here.
 
You must log in or register to reply here.
Back
Top