34 replies to this topic

[Cburg]

Dear Forum,

I am intereseted in using the buy and sell signals from a free website (americanbulls.com) as input to an auto trading program. Would someone be interested in helping with this? It could be a very profitable venture!

Thank you,

Sincerely,
Mike
:cool:

[WingedPanther]

Unless you're offering static pay, regardless of the program's results, this sounds like a great way to lose a lot of money. A poor algorithm could very easily cost you your shirt.

[Cburg]

I am certainly willing to pay for program development.
The AmericanBulls program is very well proven. It doesn't produce positive returns every time, but the overall track record is amazingly good. It is posted with each stock. Check it out: HLNT HIGHLINE TECHNICAL INNOVATIONS INC
:cool:

[WingedPanther]

Ah, so the goal is to monitor american bulls and just buy/sell as it indicates. That wasn't clear to me, at first.

[Upstream]

To be honest, I would not invest my time or money at this stage. I would like to know how much do you know about trading, finance and economy to start with?

Getting the data is not that hard at all…

[Cburg]

Hi Upstream,

Thank you for the reply,
This sounds encouraging:

Quote

Getting the data is not that hard at all…

So, you think it is possible?

I have been investing for some 25 years. I began with my retirement account and mutual funds. I then went to ETF's with my savings, and later moved to individual stocks. I have been a student of the markets for this long. I have investigated various trading strategies, and tried a number of auto trading programs; all have lost money.

AmericanBulls.com is amazing. The percentage of positive to negative returns over the two year history listed has got to be like 99%. Many are up 1000's of percent. Just enter a number of stock symbols and check it out. As best I can discern, they look at price action only, and use the Japenese Candlestick analysis to determine reversal points.

My problem is that I cannot disipline myself to follow the signals. I invariably think I know better, somehow. Auto trading would fix this for me. I would, of course test carefully before committing any substantial amount of funds, and then probably only auto trade 1/2 of any position in one stock.

There are a number of brokers who offer an auto trading interface. I have an account with ETrade. Interactive Brokers offers $1 trades.

There are a number of different languages used for auto trading. Many offer modules that perform order entry and execution management. I was thinking we could use one of these, either straight, or as a guide to programming.

I have looked at the source code for the web site, and the buy and sell signals do not occur on the same line each time. They are within a certain range, however. Do you think the program could search for the signals over a specified range of lines?

Thank you for your help!

Sincerely,
Mike

[WingedPanther]

My suggestion, before investing in such a program, would be to look at the performance around major, unexpected crashes. For example, you cited its performance over the last two years, which have had quite a bit of growth. The issue is what they did around the housing market crash.

[Cburg]

Dear WP,
Thank you for your concern. That data is not available, and I doubt the company would supply it. They are not very responsive. As you can see, however, in the data available; it does an amazingly good job of detecting down turns, and stays on the sideline until an up tick signal is detected. All of the performance data was achieved without going short, as well.
As you can see in the history data, however, it does not make 100% correct trades. To the contraty, some 25% of them are usually incorrect. Over the record, however, and with compounding, many stocks return 1000's of percent returns. Certainly, if it makes a bad call on a precipitus drop this will adversly affect the results, and can be devistating. To hedge against this, as in all investing, the thing to do is to diversify across sectors and markets.
Best regards,
Mike

Edited by Cburg, 24 May 2011 - 10:22 PM.

[Upstream]

Dear Mike,

Quote

To be honest, I would not invest my time or money at this stage. I would like to know how much do you know about trading, finance and economy to start with?

Getting the data is not that hard at all…

When posting previous I was expecting to unveil a lazy gold digger, not bothered by any knowledge of the financial markets, thinking he can do what no other has accomplished: reducing stock portfolio management to an simple program that does it all by it self.

Reading your reply I know this is not the case so I will invest more time. My first question is:
- Is it legal to grab/harvest the data from the target site?
- Is the data offered via an other commercial channel for use in 3rd party software?

I would guess it’s not permitted to grab the data as suggested. Even when permitted, using this method the data would be prone to corruption due to the inherently unreliable nature of the method itself.

The way I see it there is only one more sensible option to explore: is the data available to 3rd party developers? And what does it cost? And then in what format is it available?

I would first get answers to these questions before perusing this business venture any further.

I really like your general idea and would like to contribute if feasible but here is a final note from my side.

The software to develop needs to be quite sophisticated based on my understanding of your wishes, at least taking into regards:

- portfolio management (basic level)
- strategy implementation for trading at stock level (advanced) to take into account matters such as selling signal with intend to buy again at lower price etc
- data collection module (basic)
- interface with broker system for direct trading (? Intermediate to hard?)

If you are serious and if it’s doable I am interested in pursuing this venture furthermore!

Kind regards,

Upstream, BBA

[Upstream]

zzzzzzzzzz

[Upstream]

Cburg said:

I have looked at the source code for the web site, and the buy and sell signals do not occur on the same line each time. They are within a certain range, however. Do you think the program could search for the signals over a specified range of lines?

I took a quick look at the source and I think it is possible in 1 or 2 ways.
1. We take the stock list page as a starting point:
Americanbulls List of Stocks

We can read all data into destination database (I would copy entire file to program, then extract data). To extract we determine the begin/end of each block containing data for a single stock. I would say for now: each trade signal is an end/begin of a single stock block. We read the file top to bottom and stock 1 is everything before the first occurrence of any valid trade signal (buy/wait/ect) and the 2nd is everything after the first signal including the next signal found (needs tweaking for trimming 1st stock but still).

2. We again start at the url:
Americanbulls List of Stocks

But know we read only all the stock notations and get the signal values for each from the deep link page where each stock is displayed in more detail. The link we would craft to our needs is:
http://americanbulls...?CompanyTicker=[COM NAME]&MarketTicker=[EXCHANGE MARKET]&Typ=S

Replace the info between brackets and get:

Ford (notation derived from listing=F_WS)
F_WS FORD MOTOR CO WTS

FAA Claymore (FAA)
FAA CLAYMORE NYSE ARCA AIRLINE ETF

It’s as simple as that… I think

[WingedPanther]

Cburg said:

Dear WP,
Thank you for your concern. That data is not available, and I doubt the company would supply it. They are not very responsive. As you can see, however, in the data available; it does an amazingly good job of detecting down turns, and stays on the sideline until an up tick signal is detected. All of the performance data was achieved without going short, as well.

Given that there are a number of retirement investment experts who are predicting ANOTHER crash soon, you might want to get that data. If their 25% error rate falls on such a crash that hits multiple sectors, you could be in a world of trouble. Knowing how they did around a crash and whether they've adjusted their algorithms if they lost big on them would be vital knowledge.