TA - Algo, TCA, and execution tactic Identification

[kmiklas]

Dear Peeps:

I'm not much into TA, so please forgive my ignorance, but I had a thought...

Do you study Transaction Cost Analysis as a driver of movement into your TA? Specifically, signature patterns of algorithms, and execution tactics?

VWAP: Typically if a big investor wants to take a big position in something, they won't dump a buy order for 100000000 shares--they'll move the price against themselves! Instead, they'll run some sophisticated version of VWAP. Can you identify this in TA? Are you able to see that Big Money has an algo running on an instrument?

TACTICS: Can you identify when said investor is using an execution strat? Like Iceberg, latching, etc?

TCA: Generally, is TA advanced enough to see the tactics of your target? To understand how patterns move based on typical transaction cost analysis (TCA), and the tactics employed to minimize TC?

Basically, reverse-engineer the strategies resulting from TCA.

EXAMPLE:

JPM wants to buy 10M shares of IBM. They run a modified VWAP for a week, with stop-sweeps and psych moves when the book is thin. They pick up the shares at a 2% transaction cost, Moving the stock only 5.2% during the week, and saving 1% over a simple VWAP, and 3% over TWAP. Your TCA identifies the pending, and you piggyback the entire trade for a nice fat 5.2% in a week.

Dark pools change everything of course, this assumes open market trades.

I'm thinking of writing an algo to do this. It will be hard.

I really like this form of Technical trading because the concept centers on the NOW. Traditional TA will look at _past_ performance, like moving averages, slopes, etc--and we all know what they say about past performance! They this form of TA focuses on _present_ performance.

Thx, Keith XD
"Trading is War"

 

[Girija]

Problem will be timing. You can run heuristics after execution. But why take a awful lot complex route. On depth of market if I see size at support zone i assume someone knows something I don't know. In other words piggy back the size. Equities like AAPl there can be thousands of reasons why someone is trading.
There are hundreds of ways large order is broken.
Watch level 2 for ES futures and you can follow the pattern. Dxfeed sells software that marks what you want may be you can see the tool and then follow the data for a clue....

 

[qlai]

I think you are mixing up timeframes here. You may be able to identify on a particular day that there’s algo running, even though it’s unlikely because they know how to hide their activities. But then you start talking about weeks and any news will change price so the algo will just adjust itself to current price, volume, and volatility. You don’t know what the range of prices they are willing to buy/sell at and what their tolerance for pain is.
Let’s say you stick with intraday, you would also need really good full depth level2 market data to figure out what they are doing, which is crazy expensive.

 

[easymon1]

signature patterns of algorithms

What would be three signature patterns of algorithms?

 

[kmiklas]

What would be three signature patterns of algorithms?

Here's one: a very basic VWAP. [1] I'm wondering if I can identify this algo being run by an institution with some form of technical analysis. Perhaps write a "VWAP detector algo."

Note that in practice, it's quite a bit more complex... and subtle! You won't see a clean line, "head-and-shoulders," or any other obvious pattern. They are being deceptive, and the clues will be tiny little anomalies, only seen by sharp-eyed observers.

1. Johnson, Barry. 2011. "Algorithmic Trading and DMA"

 

[easymon1]

Two more?

 

[kmiklas]

Two more?

#2: If I can perceive the plan of
the enemy; in this case, to take a position with VWAP, I have won! I know present, and the future. Just get on the trolley and ca$h in.

Attached from Sun-Tzu, "The Art of War"

 

[kmiklas]

#3: Hiding... [3]

What do you think? Is there hope for technical analysis based on perception and identification of in-flight transaction cost minimization algo/strats from institutional traders and market movers, or am I an idiot?

3. ibid Barry Johnson

 

[kmiklas]

"""
...
4. Thus, what enables the wise sovereign and the good general to strike and conquer, and achieve things beyond the reach of ordinary men, is foreknowledge.

5. Now this foreknowledge cannot be elicited from spirits; it cannot be obtained inductively from experience, nor by any deductive calculation.

6. Knowledge of the enemy's dispositions can only be obtained from other men.

7. Hence the use of spies, of whom there are five classes: (1) Local spies; (2) inward spies; (3) converted spies; (4) doomed spies; (5) surviving spies.
...
"""
sun-tzu art-of-war ch13

This was written before the Information Age. Today, knowledge can also be obtained from data feeds.

I'm working on a sequel; namely, "The Art of Technical Trading" XD

 

[Girija]

Here's one: a very basic VWAP. [1] I'm wondering if I can identify this algo being run by an institution with some form of technical analysis. Perhaps write a "VWAP detector algo."

Note that in practice, it's quite a bit more complex... and subtle! You won't see a clean line, "head-and-shoulders," or any other obvious pattern. They are being deceptive, and the clues will be tiny little anomalies, only seen by sharp-eyed observers.

1. Johnson, Barry. 2011. "Algorithmic Trading and DMA"

Play around with fourier transform and find the noise on arbitrary data.