Hello there! And random walk model

[outspan]

Hi all ,

sorry for my poor English. I'm new to FX trading. I'm a telecom engineering student and while trying out a couple demo accounts, I tried to put my mathematical analysis / probability / signal theory notions to good use and elaborate a mathematical model which I could later use to create a reasonable strategy for a mini live account.

Given the huge amount of money traded (which should make it very hard for anyone or anything to influence the market directly), has it already been considered to model FX trends with a random walk, at least on a sufficiently wide time scale? Combined with the stddev indicator tool which I've seen on some trading platform, I think it could provide an excellent basis for a model.

I tried to build a very simple strategy built on this model, and I got a 95% success rate (although I'm calculating this just on the 20 trades I made so far after elaborating the model, which can't be statistically significant -- but it's still encouraging). The catch is that when I lose, I lose much more than what I profit on the average trade.

So anyhow, back to my question: do you think the random walk model is a valid one? What are other valid mathematical models?


Thanks,
Dario

[akeakamai]

I believe any method that incorporates trend-following, chart patterns (including candlesticks), or support&resistance levels, which takes into account good risk management has the potential to be profitable.

I do not believe however that a statistical model, that does not utilize one of the above trader psychology patterns, can have a true "edge" over the market. Even if you do find a mathematical pattern, I'd be willing to bet that it can be traced back to a psychological pattern of traders.

[JonnyFX]

Lot of newbies to trading; stocks, commodities, fx
have a tendency to over-think and over-analyze.

And don't get in the mistake of applying math, statistics,
or any level of statistical modeling to the concept
of trading. Please.

I myself only use a little bit of news from CNBC and
Heiken Ashi Smoothed trending indicator candlestick.
That's it.

Bottom line: trading the market with market makers,
professional smart money, and fellow traders is not
about math.

[outspan]

Thanks for your answers.

Of course I don't mean that such a model should be used blindly, but I have reason to believe that, in conjuction with trend-following and knowledge of the most common patterns to know where to open and close (rather than these techniques alone), it could yield fairly decent results.

Is there such things as a free tool that I could use to test different models and strategies on historical data (including stop and limit orders?)


Thanks,
Dario

[tonymand]

Originally Posted by outspan

Hi all ,

sorry for my poor English. I'm new to FX trading. I'm a telecom engineering student and while trying out a couple demo accounts, I tried to put my mathematical analysis / probability / signal theory notions to good use and elaborate a mathematical model which I could later use to create a reasonable strategy for a mini live account.

Given the huge amount of money traded (which should make it very hard for anyone or anything to influence the market directly), has it already been considered to model FX trends with a random walk, at least on a sufficiently wide time scale? Combined with the stddev indicator tool which I've seen on some trading platform, I think it could provide an excellent basis for a model.

I tried to build a very simple strategy built on this model, and I got a 95% success rate (although I'm calculating this just on the 20 trades I made so far after elaborating the model, which can't be statistically significant -- but it's still encouraging). The catch is that when I lose, I lose much more than what I profit on the average trade.

So anyhow, back to my question: do you think the random walk model is a valid one? What are other valid mathematical models?


Thanks,
Dario

Consider 2 things. First the vast computing skills and access to PhD mathematicians the banks have. Are you likely to outdo them? Second if they do find an edge how long do you think it takes the market to catch up and render their edge meaningless

[daydreamer65]

Originally Posted by outspan

Thanks for your answers.

Of course I don't mean that such a model should be used blindly, but I have reason to believe that, in conjuction with trend-following and knowledge of the most common patterns to know where to open and close (rather than these techniques alone), it could yield fairly decent results.

Is there such things as a free tool that I could use to test different models and strategies on historical data (including stop and limit orders?)


Thanks,
Dario

Search metatrader4 & mq4 on google.

[outspan]

Originally Posted by tonymand

Consider 2 things. First the vast computing skills and access to PhD mathematicians the banks have. Are you likely to outdo them? Second if they do find an edge how long do you think it takes the market to catch up and render their edge meaningless

Mathematicians don't worry me, they usually suck at modelling real-life situations -- that is what engineers are for!

But apart from jokes, maths is a tremendously powerful tool when you know how to interpret your results and error propagation. I don't think I have the multimillionaire strategy at hand, I've just jotted a couple ideas and I'll wait till the end of my finals to perfect and test them extensively on a demo account.

At least for me, knowing that I can estimate the probability of winning / losing a trade (which is trivial with a random walk model) and the risk/reward ratio would take away a lot of dangerous emotions out of my "real" trading. That can't be bad.

As for what you say, FX has got to be a zero-sum game. I don't have to aim at taking money out of banks, just out of people who don't use a valid and consistent strategy like I can only hope mine will be

[outspan]

Originally Posted by akeakamai

Even if you do find a mathematical pattern, I'd be willing to bet that it can be traced back to a psychological pattern of traders.

Finding a mathematical pattern would be way out of my league (and I definitely agree with you that patterns are mostly psychological, especially in day trading, even though I don't know much about FX at all), but finding a model is much much easier .

[akeakamai]

Originally Posted by outspan

Finding a mathematical pattern would be way out of my league (and I definitely agree with you that patterns are mostly psychological, especially in day trading, even though I don't know much about FX at all), but finding a model is much much easier .

Well I have no doubt a model could help define an edge in the market, but what exactly are you going to model? You have to have somewhere to start, I mean you can't just expect to "model" the workings of the entire foreign exchange market.

[outspan]

Originally Posted by akeakamai

Well I have no doubt a model could help define an edge in the market, but what exactly are you going to model? You have to have somewhere to start, I mean you can't just expect to "model" the workings of the entire foreign exchange market.

To put it simply: a random walk has a Gaussian distribution in which stop losses and take profits are vertical lines cutting the distrubution function, the first on the negative side, the second in the positive side (the X axis is the P/L).

The probability of incurring in a stop loss is the integral in X of the distribution function from -infinity to the stop loss value. The probability of incurring in a TP is the integral from there to infinity. Knowing the stddev of the distribution, you can calculate what are the SL and TP values that guarantees an expected value of profit that is > 0.

It's not actually that easy, but I hope I gave you the idea. Math rocks!