Seeking Programmer for Hire for complex project

[ctrlbrk]

Dear programmer,

I want to hire a Ninja programmer that can deliver to me:

1) Fully functioning source
2) Function I can easily incorporate into my other strategies
3) Ability to connect to my private MySQL database
4) Store information into my DB regarding the success/failure of last trade
5) Retrieve information from my DB regarding past trades success/failure

The information recorded on the trade would be something like:
a) Date/time of entry/exit
b) Instrument
c) Entry price
d) MAE
e) MFE
f) Exit price
g) Stop
h) Target
i) Multiple indicator values at entry and exit, such as "indicator 1: value 1", "indicator 1: value 2", "indicator 2: value 1", "indicator 3: value 1"

It is also very likely the DB will need to store a lot of other information in the database, such as the open/high of each bar and the values of each indicator passed to it. This is so that later when a trade fails, the database will contain the information necessary to determine what the indicator values would need to have been in order to either not take the trade if the target was not possible before hitting the stop, or what the indicator values should have been to get an earlier/later/better entry in order to hit the target before hitting the stop.

My strategy will need to be able to use the global function to query the database for this information, so it can make changes to itself (self-adapting/self-learning) for future trades.

I can call an update function onbarupdate (close of bar) so database has access to ninja variables. It can also be called immediately upon execution of an Entry order, with predefined stops and targets. So it can use this info to determine if a trade was a success or failure, or maybe there is another way to query Ninja directly for this.

I am willing to take on partners that want to split the cost of development and receive the full code, you can let me know if interested.

Mike

[ctrlbrk]

Well, I've managed to accomplish the beginning of this on my own.

I now have a mySQL database storing trade transactions, and a whole slew of indicator values at the time of the trade, along with the profit target, stop limit, etc.

It is only a beginning, but I can now query the DB to see if the last trade was successful, and know what the values of certain indicators were if it was (or wasn't), and then Ninja could take a different action.

Below is a screen shot of mysql db table, along with the trades tab from ninja.

Woot!



Mike

[Elliott Wave]

I had a similar idea a while ago. It seems like it could be a great way to develop and improve trading strategies.

Good luck.

[pauly139]

Hi guys

I wish you good luck with your project - I spent 3 years as part of a team that did just this. We analysed 100s of millions of trades / indicator combinations, including combinations of indicators. We used Microsoft Analysis Server with a cluster of 10 pcs automatically collecting trade data and analysing over many 1000s of hours.

In our case it proved to be a dead end, all you get are the classic curve fitting results.

Rgds
Paul

[ctrlbrk]

Originally Posted by pauly139

Hi guys

I wish you good luck with your project - I spent 3 years as part of a team that did just this. We analysed 100s of millions of trades / indicator combinations, including combinations of indicators. We used Microsoft Analysis Server with a cluster of 10 pcs automatically collecting trade data and analysing over many 1000s of hours.

In our case it proved to be a dead end, all you get are the classic curve fitting results.

Rgds
Paul

Wow.

But surely you agree you can make better trade decisions (or the strategy can) by examining some of this information, yes?

Mike

[pauly139]

It depends upon how you approach the whole exercise. If you use such a system to try and find indicators that work, no. If you design your analysis to look at indicators that you know have a certain performance, using whatever metrics you decide are useful, and then compare that performance in different market situations, then some value can be found.

However this was just my experience - others may have had better success :-)

[ScottB]

I am late to this thread but my experience is much the same as pauly139's. I was with a hedge fund working on high frequency/algorithmic trading and in the end most of that stuff didn't help.

What does work is pure price action analysis around opening ranges and pivots (R1, S1, etc). On another forum, someone (not me) started what amounted to an adult (to use the term loosely) food fight by quoting studies that seemed to indicate that most indicators were not useful at all and the rest were only marginally useful.

Traders just don't want to hear that. When looking at a chart, your eyes will naturally be drawn to things that look predictive but try those same techniques against what is affectionally known as the hard right edge of the chart.