A Statistical Edge

Hello All.

My name is David and I was curious what software is commonly used to analyze and sort through trading data in order to try to find an edge. I have been thinking that I know enough to get around with the technical analysis (Price Actions, S/R, Trendlines), but I am lacking an edge. I believe that if I am able to focus my time on obtaining this edge, I will be able apply the current technicals I know more efficiently.

By analyze I mean:
Probability of nth candle being bearish after x amount bullish candles?
Probability of candle continuing in the same direction after engulfing the previous candle?
Probability that a break of a range with x amount of candles will continue in that desired direction?

These are just examples. I know these are the basics, but I need to start somewhere. If anyone can shine some light are there any books/resources that can help me with this?

I am planning to cut down to one instrument, selecting one time frame, and working an edge of that specific instrument and timeframe.

Thoughts?

Anyone?

http://www.mtprogramming.com/programming.html

they offer a service called statistical data modeling. Maybe they can tell you where to start or what books to read.

Regards

For a single pair modelling, your best bet might be to explore using neural nets. I know of a guy here who got a neural net based model done for gold (xauusd) and traded very successfully.

Caveats:

1) its a black-box approach - the neural net is trained to analyze cluster of candles to forecast when to go long or short, but you will never know the reason why - so there is kind of faith thing involved

2) can be expensive to get a good model done

3) since this is essentially behavior modelling, the model may need to be re-worked if the underlying behavior changes.

Re, the last point, for example the decoupling of chf had an impact on the model that this guy was using and he had it re-trained. His trading is the best/ most accurate I have seen - I know since I trade the same pair exclusively, but using traditional tech analysis, and his trades are by far more profitable and accurate.

If you want, I can ask him for contact info of the guys who did this for him - you can pm me for this

Cheers

​If I may, the statistical analysis you are seeking would be better applied to how YOU react to the markets, i.e. entering and exiting. Applying these mathematical models onto the markets, or even price action you will soon find is inconclusive at best. If you are looking to find a method to enter and/or exit your studies will eventually show a 50% split on probabilities - given a large enough data sample.

I have found it much more productive to study how order flow effects price and then study MY reaction to the market. Do I enter long when I should've entered short? What about the price action or order flow made me do that? Gather this kind of data and it will soon be come clear to you how to trade.

Unfortunately it is not easy and can be greatly time consuming but the information and probabilities you gain from this kind of study are far more valuable. And in the end you will understand the most important thing about trading. How YOU react to the markets AND how to leverage that into a winning strategy.