How long before you were fully automated?

[travis]

Quote from BlackMage:

No, this is a bit too simplistic view. Banks runs fully automated market making strategies just to mention one example. It can also be difficult to completely distinguish between a program for efficient execution and a program for generation of profit ("strategy trading"). Its a complex world.

Regarding latency, take for example a look at this thread for a discussion on this (if you have not done it all ready):

http://www.elitetrader.com/vb/showthread.php?threadid=161595&perpage=6&pagenumber=1

Latency becomes an increasing issue when increasing the frequency of your trading and for some business models it is essential. But don't make the mistake to think that people who cares about latency are in any way stupid.

Who are "they" with the low returns that you refer to?

Who are they? Fund managers. Everywhere I read that if a fund makes a 20% return every year it is considered to be a very good fund. Well, I would consider that a low return, since, as you can also see from those real money trading tournaments, we can achieve 100% per month.

 

[travis]

Quote from monty21:

What language are you guys using to build your systems? I assume C++?

Does anyone just use Excel and VBA?

Can you guys mention what language you think is the best for programming an automated system?

Me just excel and vba, but that's all I could learn because I don't have a programming background. I find it great, but I can't compare it with anything else.

 

[travis]

Quote from TSGannGalt:

I think most of the people lose focus and only think about "sending" orders. But you have to understand the flip-side of having a fast network. It's to get your fills quickly / efficiently. I think everyone's gone through the frustration of delayed fills from your broker.

A lot (not all) models and systems start managing trades after the acks are received from the brokers. Receiving acks are equally important during fast markets, and when you run a High Frick model, any bulk order can tip the model over.

Thanks for the info, it all makes sense to me. Personally, I never noticed any delays from IB, but I only buy 1 contract at a time, and by "delay" I am speaking of seconds, not milliseconds (and so). I mean a delay that someone who's not expert in this "latency field" could notice.

 

[PointOne]

Quote from jack hershey:

Programming coding is most functional when what goes to the execution platform is something, that on a given fractal, extracts the market's offer. what you see here DOES NOT USE PRICE IN ANY WAY TO achieve this basic level of operation. The code is less than three standard pages long. It is derved, in my opinion, deductively and uses 1950's type stuff. If you can see this chart (made on Earth Day)and its worth, then you are capable and you can bridge across from being capable to being expert in taking the market's offer. This is a representation of "making money" and making money most of the time and to the extent offered.

Jack
how did this automatic system suppress the short signal which may be generated by the fast stoch and volume (see circled area in the attachment)? I note that the slow stoch had made it all the way across and the MACD had crossed over above zero.

 

[TSGannGalt]

Quote from PointOne:

Jack
how did this automatic system suppress the short signal which may be generated by the fast stoch and volume (see circled area in the attachment)? I note that the slow stoch had made it all the way across and the MACD had crossed over above zero.

Here's another question.

I'd like to see a pre-system study and observational result that proves that his "deductive theories" are true.

I don't care much about his STC or other models he has. I just need a a sound observational (meaning tested proof) that his "theories" are what he claims to be. Obviously, a situational "so-and-so" is making money by trading my way is not going to work, due to it being inductive.

PS. I'd like to note that I am not bashing on jack hershey. I'm merely providing him a place to make intellectual responses. Also, I'm not going to bother following his posts around like he says about another thread. The discussion proposed stays within this thread. And going to keep it that way.

 

[TSGannGalt]

Quote from travis:

Who are they? Fund managers. Everywhere I read that if a fund makes a 20% return every year it is considered to be a very good fund. Well, I would consider that a low return, since, as you can also see from those real money trading tournaments, we can achieve 100% per month.

OK... here's a few things...

1. Algorithmic trading (execution models which broker offers clients like VWAP, TWAP, Pounce... etc. etc. ) are usually used by funds that trade a long time frame, like asset management companies and etc. Before Algo. trading, asset management firms would call the broker using "Worked Orders" which allowed the sales trader to execute the orders using their discretion.

So an example would be, "Buy 1,000,000 shares of XXXX, Worked by today". The sales trader would receive the order and use their discretion to execute the 1M share order by the end of the day, at the lowest price as much as possible. These days, models took over that role by sending split order within the day to provide the client with the best possible execution.

2. Most of the Hedge Funds use institutional DMA on outright trades or trade through Stat. Arb specific Swap accounts. The difference is the leverage of the account is allowed. Outright trades' margin is the value of the single trade. For stat. arb / swap accounts, it's the cumulative value.

3. There are other types of accounts, catered for the clients' need. Some firms may have a separate account for holding stocks that they want to go short, instead of paying and looking for one on the broker's short list....

Anyways... there are a lot of things institutions do that provides the funds some edge that retail firms don't....

It's a very different environment.

 

[jack hershey]

Quote from TSGannGalt:

OK... here's a few things...

1. Algorithmic trading (execution models which broker offers clients like VWAP, TWAP, Pounce... etc. etc. ) are usually used by funds that trade a long time frame, like asset management companies and etc. Before Algo. trading, asset management firms would call the broker using "Worked Orders" which allowed the sales trader to execute the orders using their discretion.

So an example would be, "Buy 1,000,000 shares of XXXX, Worked by today". The sales trader would receive the order and use their discretion to execute the 1M share order by the end of the day, at the lowest price as much as possible. These days, models took over that role by sending split order within the day to provide the client with the best possible execution.

This a very familiar thing to me and is the basis of the PVT trading. As you see in the illustration it comes from a volume signal that leads price.

The bottom chart shows the basis of the tabular look up table which isssues the signals.

Besides this, there is the matter of which universe of stocks the illustration applies to. In the late 50's I eyeballed the YTD data for H-L and chose what became known as high Beta stocks years later. Today it is automatic using script and IBD data. You push a button and the list appears.

From the 30 minute volume bar chart at the bottom, you see that the bars add up to 100%.

The look up table has columns that are self explanitory. DU is a volume that is 0.25 times the 65 day average given on most stocks. 45 days would be better and that is being worked out with a platform provider that over a year's work has finally, successfully debugged his fundamental data supply.

The First Rising Volume (FRV) opens the window for trading. You have an hour to 1-1/2 hours to make an entry BEFORE price begins to move. This column is based on 0.65 of the average daily volume.

Hold ends when volume is no longer sustained. Peakiung volume is greater than the 65 day average.

The ATS was created by using volume as a leading indicator of Price and the trigger values of volume are shown as color changes on the columns.

I use the same hypothesis set for all trading: intraday, interday and interweek. These are trading fractals. Within in each one there are three fractals as well: one on each side of the trading fractal up until expert level money velocity.

A typical example for me can only be one tenth of TSGannGalt's prowess. I limit myself to 100,000 shares or in the 2 million dollar range for interday position trading using PVT. Common prfits are 10% a turn and uncommon is 30% a turn and maximum has been 17 points on 100,000 shares in the 30 dollar range.

When the SEC was struggling with using computers to "catch" and site people, I was "caught" and they found out they were poor programmers. so they got an education by me, deductively, going through their erroneous reasoning. Government employees are not too swift even today.

How stocks cycle is important to know. I use a scoring technique that has three variables so I get 8 combinations which flow in a cyclic order for a selected universe based on fundamental analysis. The three variables are P, V, and A/D. The frequncy ratios are 1:2:4, respectively. Raw score is binary vectors (time rate of change, the parametric measure dictated by Theory). Going long is when all variables change simultaneously (as expected). In decimal it is 0>>>7 and 4>>>3.

2. Most of the Hedge Funds use institutional DMA on outright trades or trade through Stat. Arb specific Swap accounts. The difference is the leverage of the account is allowed. Outright trades' margin is the value of the single trade. For stat. arb / swap accounts, it's the cumulative value.

3. There are other types of accounts, catered for the clients' need. Some firms may have a separate account for holding stocks that they want to go short, instead of paying and looking for one on the broker's short list....

Anyways... there are a lot of things institutions do that provides the funds some edge that retail firms don't....

It's a very different environment.

 

[jack hershey]

Quote from PointOne:

Jack
how did this automatic system suppress the short signal which may be generated by the fast stoch and volume (see circled area in the attachment)? I note that the slow stoch had made it all the way across and the MACD had crossed over above zero.

Per your Q, a chart is posted for the interval of your circle.

Four signals appeared and they related to other than the "channel" level of the "basic" Cash Cow ATS.

As shown signals do not an ATS make. Raw Information comes into an ATS and several themes result: The carrier, bells and whistles to create parallel degrees of freedon, the filters to steer on the non stationary window main event under way, gates to narrow the steering to a minimum number of degrees for sufficiency to give certainty, and decisions to go to output actions or prevent any outputs to the execution platform.

We have over the five bars, 19 status information elements and four signals. And a completed ATS routine that Brought specific closure and a new status within each bar.

Your conclusion about what happened over the interval is RIGHT ON AND VERY VERY EXCELLENT!!!!!

HOW is the query. Status is used in two ways. And status is like a wall on two levels. One is an inner wall that keeps the profit segment rolling along. It is interrupted by signals relative to end effects of profit segments. And it is known that, deductively, an end of one segment is the beginning of another segment.

The outer second part is the housekeeping with regard to being in the trading fractal and NOT doing anything messy like jumping fractals.

All of induction is not fractal sensitive since it deals in non theoretical logic that is devoted to statisticulating with raw data and Craying it to death as fast as possble and sprurting in as low a latency manner as possible.

Our illustration is making 50 points a contract for up to non partial fills of 500 contracts. Soften it to 10 actions with 400 contracts and multiply by 50 and 50 to come up with the extraction for the day. The number is 100 and four zeros after 100. Two commas are required.

HOW is a million dollar question and you asked it and two walls achieved your conclusion of what. As you see there are no loose ends or uncertainty since the status (19 info status shots) is determined robustly as they say. The four lousy signals did not apply to trading on the channel fractal, so they were walled out.

Cash Cow Basic will go to Cash Cow Basic Supreme then cash cow intermediate (traverse trading) and finally cash cow expert (tape trading).

Basic is around 2 to 3 times ATR as you see,

Supreme will take the flutter out One in staed of threse rversals at channel overlap beginning 10 becomes 6 which is within Basic's 4 to 7 trades.

Intermediate goes up to 15 trades and expert is 20 to 40 as expected.

 

[jack hershey]

Quote from TSGannGalt:

Some of the things you mention do not make much sense to me.

snip ...

Seriously, I've never heard of old school traders in Chicago or NYC going to Traders Expo as a catch up event. And whenever you go out drinking on Friday/Sat. nights, you see the same people over and over again. Maybe it's a Chicago or NYC thing, they crammed up all the industry people in a small box...

snip

MY CHI or NYC my recollections are of after market get togethers during the week that focus on two things: the market is one of them and the ratios are good for the other. One guy who may be reading this said to a third party in CHI one such evening: "WTF, you just got a card and it took me a year to get one." It was a very funny moment for all of us......

Lets move on. Old school traders don't do much of anything as I see it. An ex President of the big SP contract floor expressed to me a couple of comments that went like this " I stayed up all night thinking after that session yesterday" also " You two guys are red Masserattis' just off the line and I am a "59 Buick"...

Lets go to the Vegas Expo....... you see from left to right: Netto, Anderson, an ex GS and a guy who went through the 9/11. four in a row. They ALL have their programmers sitting somewhere and Netto's is standing in the back. I know Anderson's car from Italy is running four months late and it is CUSTOM all caps. The moderator (Owner of Expo) is only going to do Q's for fifteen minutes and they ALL shut him OUT.

Netto is laughing at the GS guy and going at a mile a minute explaining why...

Business is done Afterwards and it involves whales and other experts.

Go to the NYC Expo........ same panel some new players and some "regulars". Laying off 8 digit money is on the table afterwards. They (we) keep in touch by cell.

During live trading session I am sitting on the floor below screen and Netto is acing live trading against some "vendor CEO" who fucks up badly and Netto and I are doing hand signals..... At the end Expo staff swipe my name badge I take out of my posket and ask: "do you want to do this next year" The conversation is in 8 digit layoffs after the session.

The last digits of my cell are: 7, 6, 5, 4........ That's the scoring hold series for PVT trading.

Programming, I feel is not too swift, by the old schoolers. It is a very hot item in about 30 ways for the contemporary experts who are on point and cleaning up in the markets...

Pointone just asked a "How" question relative to "what" he saw. The "what" happened over and over on the chart. I'm sure he saw the logic flow sheet when it was originally posted and corrected.

I think pointone has the picture: the core is actually the Basic "channel" level where the big profit segments are the foundation from the Hypothesis SET

Then intermediate and expert break up the big chunks into 3 or more smaller MORE profitable segments as a fractal shift is done at each level of iterative refinement.

So pointone examined a chart. 10 actions.

10,000 extracted for 4 contracts

100,000 extraceted for 40 contracts

1,000,000 extracted for 400 contracts.

On Fridays you sweep the accounts and shift the profits down to a slower way of making money, all using the same hypothsis SET in different markets on different fractals.

Lets say Travis codes up the PVT for stock position trading.. Let's say Pointone codes up the indicator Cash Cow Basic.

let me get to the rest of your Q's and scepticism.

 

[jack hershey]

Quote from TSGannGalt:

Here's another question.

I'd like to see a pre-system study and observational result that proves that his "deductive theories" are true.

I depended on Dow Theory and Granville's P, V relationship and 4 th Ed of Magee up to page 7. then, of course, I kept alert for the other theories as they were proven as the years passed. for math George Bolle got the job done in the 1840' s and be fore that integral and differential calculus converted to statisitical stuff because of non continuous function limitations. Matrix theory gave me the migration concept one a made matricies of market variables.

For anyone stating now, I suggest the shortcuts.

Warm Up

1. Examining the Riddle of Induction

2. Making use of the elements in the Paradigm based solution to 1.

Begin with this group next:

3. Logical Theory --Carnap

4. Principle of Indifference

5. Bayesian Theory -- Rev T. Bayes and the "frequentists"

6. Invariance Theory

7. hypothseis sets -- Keynes

8. Classical Theory

9. Underlying Measure Invariance Theory

10. Paradigms

11. Kuhn's property tests for Paradigms and their Hypothesis Sets.

12 Symmetry

After this you have the ATS done

Some people like to be first. so you can figureout if you are for what you created.

Use the third Characterisitic of paradigms: defensibility.

most people cannot and do not follow the above path; instead they use frequency induction and get what they get.

Also Simplicity Favoring and its buddy curve fitting appear in lasser efforts.

Once you have deduced the "carrier" you can use theory on the bells and whistles as well.

I happened to use economically organized heirarchies. See sweep chart of nine table on three fractal levels. There are several bells and whistles too.

I don't care much about his STC or other models he has. I just need a a sound observational (meaning tested proof) that his "theories" are what he claims to be. Obviously, a situational "so-and-so" is making money by trading my way is not going to work, due to it being inductive.

A lot of people take themselves ther and do not know anything about anything and they build a set of assumptions all based on their reality and make very humorous conclusions. A few just go straight to OCD and skip making money at all.

PS. I'd like to note that I am not bashing on jack hershey. I'm merely providing him a place to make intellectual responses. Also, I'm not going to bother following his posts around like he says about another thread. The discussion proposed stays within this thread. And going to keep it that way.

sounds great lets skip the posted logic flow charts.