Forex Trading: Any Tips for Random Trading?

[iDimension]

Hello, I have a question and hope you can help. If you understand forex trading you might be able to give a more informative answer, or maybe it's just pure math, who knows.

I wanted to ask your opinion of 100% random trading. So on my platform I have 9 currency pairs selected and I use a random number generator ranging from 1 - 9. Whatever number it lands on that is the currency pair I will be trading. Then I will roll another random number and if it's 1 then I'll sell, if it's 2 then I'll buy.

I will run this 3 times producing 3 different trades. Then I'll set a stop loss of 50pips and a take profit of 150pips and just leave all 3 until the market takes them out. Wait 24 hours and repeat.

So with my understanding if all 3 trades lose then I will lose 150pips but even if 1 of them wins then I will make profit.

I mean at the end of the day it's completely random... I may get lucky and select 3 awesome trades that are doing amazing. I may select 3 trades that do terribly. Although over the long run should this not make money?

This is purely a thought I know it's probably a bad idea otherwise everyone would be doing it but can someone tell me where the flaw lies because I cannot see one.

The only thing I can think of is that -50 is 3 times closers to 0 than 150 is so you're 3 times more likely to lose... meaning over a very long time, you'd even out at 50% win 50% lose?

I mean at the end of the day a trade can only move up or down. So it's 50/50 whether you guess this correctly.

 

[OrangeDog]

That is a stupid idea and a good way to lose money. Stop being lazy, pick up some good books, and learn to trade. Youll make a $20 investment that will pay for itself if you put the work in. But hey, if you want to give me your money Ill make some investments for you.

 

[iDimension]

That is a stupid idea and a good way to lose money. Stop being lazy, pick up some good books, and learn to trade. Youll make a $20 investment that will pay for itself if you put the work in. But hey, if you want to give me your money Ill make some investments for you.

I'd prefer you told me why it's a terrible idea instead of simply stating it so.

 

[OrangeDog]

Why? Because you aren't researching the market, because you are blatantly being lazy, because if you have to ask a question about trading like that on a physics forum you probably shouldn't be trading anything at all. At the very least you don't even know the statistics of the system you are trading in, so the assumption that you have a 50/50 chance or whatever is bogus. And again, this question just comes off as you not wanting to put the effort into do something right. Seriously, Ill take your money and make money with it. Ill charge you $15 a trade.

 

[iDimension]

Why? Because you aren't researching the market, because you are blatantly being lazy, because if you have to ask a question about trading like that on a physics forum you probably shouldn't be trading anything at all. At the very least you don't even know the statistics of the system you are trading in, so the assumption that you have a 50/50 chance or whatever is bogus. And again, this question just comes off as you not wanting to put the effort into do something right. Seriously, Ill take your money and make money with it. Ill charge you $15 a trade.

For a start I am not even trading real money. I am simply messing around on a demo account just getting to understand the system and how it all works. I just had a thought off the top of my head and could not see why mathematically, it would not work. If someone can explain the mathematics as to why it won't work the of course I'll drop it and move on.

 

[OrangeDog]

There are many things wrong with your idea beyond the mathematics. In paragraph 2, your idea makes no sense. If you roll a random number, 1, you buy, if it is 2, you sell. Ok. How does that reflect how your trades will behave in real life? What is their probability distribution? If you know this you can calculate the expectation of winning vs losing and compare results. You don't need a simulation to do that.

Number 2, why are you selecting currency pairs randomly? This makes no sense. You select pairs based on the fact that if you trade them you will probably (or hope) make money. This brings me to another point, you cannot predict how the currency or whatever you are trading will do unless you make assumptions that are unrealistic. I think your best bet is to use old data and apply your strategy and see if you make money.

 

[iDimension]

There are many things wrong with your idea beyond the mathematics. In paragraph 2, your idea makes no sense. If you roll a random number, 1, you buy, if it is 2, you sell. Ok. How does that reflect how your trades will behave in real life? What is their probability distribution? If you know this you can calculate the expectation of winning vs losing and compare results. You don't need a simulation to do that.

Number 2, why are you selecting currency pairs randomly? This makes no sense. You select pairs based on the fact that if you trade them you will probably (or hope) make money. This brings me to another point, you cannot predict how the currency or whatever you are trading will do unless you make assumptions that are unrealistic. I think your best bet is to use old data and apply your strategy and see if you make money.

Firstly I will list currency pairs

EUR/USD
EUR/JPY
EUR/GBP
GBP/USD
GBP/JPY
USD/CAD
USD/JPY
AUD/USD
NZD/USD

So straight away we can say that there is a 100% probability of the price going up or down. So you have a 50/50 chance at guessing the correct shift, either up or down. I roll a 1, 7 and 9 so my pairs to trade are EUR/USD, USD/JPY and NZD/USD. Now I roll 3 more times to get the buy and sells. I roll and get 1, 1 and 2.

So my trades become

EUR/USD Sell
USD/JPY Sell
NZD/USD Buy

Now I simply set my stops at -50pips and my take profits at +150pips and wait. A few hours or days later the EUR/USD and the USD/JPY lost but the NZD/USD won. I've lost 100pips but I gained 150pips on the final trade giving my total profit of +50pips.

Rinse and repeat.

Now it's your turn to explain why this will not work. Just so you know I actually tested this today except I used -30pips and +90.. but it's the same ratio of 3:1

but this is just one piece of sample data which is why I wanted to ask here and check to see why the mathematics says it will fail. Also in that image when the price was right at the take profit I moved my stop loss right up and put my limit a little higher, just to see if I could get a few extra pips.

 

[russ_watters]

So straight away we can say that there is a 100% probability of the price going up or down. So you have a 50/50 chance at guessing the correct shift, either up or down.

The first part is true, the second part is false. The fact that the total adds up to 100% doesn't tell you anything about how much is ups and how much is downs. Indeed, if they were exactly 50/50, the market value would never change when smoothed over the long term.

People seem to believe that if you do enough trades (shrink the timing to maximize randomness), it gets close to 50/50 - and that that's a good thing: it does, but it's not: the long term growth is the thing you want and this method attempts to filter it out!

The second and much worse flaw is assuming equal probabilities of a 3x gain and 1x loss. I can't imagine what would make you believe such a thing. Maybe you are again guessing that making them small makes the probabilities equal. It doesnt.

If perfectly executed and commission free, this would yield exactly the long term gain/loss of the vehicle.

There are no easy shortcuts here, just like at the casino.

 

[OrangeDog]

People seem to believe that if you do enough trades (shrink the timing to maximize randomness), it gets close to 50/50 - and that that's a good thing: it does, but it's not: the long term growth is the thing you want and this method attempts to filter it out!

That is only true if the underlaying probability distribution follows. As I said before, it is assumed in the Black-Scholes model that the price of whatever they are pricing (I trade stocks not futures or options) follows Brownian motion. Essentially this means that the price varies with a normal distrubition centered on zero. A more accurate distribution might be a skewed normal or log-normal, but my area of study is fluid mechanics not stochastic calculus and financial mathematics.

 

[iDimension]

The first part is true, the second part is false. The fact that the total adds up to 100% doesn't tell you anything about how much is ups and how much is downs. Indeed, if they were exactly 50/50, the market value would never change when smoothed over the long term.

People seem to believe that if you do enough trades (shrink the timing to maximize randomness), it gets close to 50/50 - and that that's a good thing: it does, but it's not: the long term growth is the thing you want and this method attempts to filter it out!

The second and much worse flaw is assuming equal probabilities of a 3x gain and 1x loss. I can't imagine what would make you believe such a thing. Maybe you are again guessing that making them small makes the probabilities equal. It doesnt.

If perfectly executed and commission free, this would yield exactly the long term gain/loss of the vehicle.

There are no easy shortcuts here, just like at the casino.

Thank you for your reply Russ. Actually when trading the day chart the ups and downs is quite commonly referred to as "noise" because the market almost certainly does more both up and down by quite a large amount each day. It's over the long term that the trend takes place.

So there may be a 100% chance that the market doesn't rise, but when I randomise a number, it might be sell or buy. So I have a 50% chance of guessing the correct move. I don't particular care if there is 0 chance of the market going up, because there is a 50% chance that I actually chose to buy, and thus I am in the winning trade.

That is only true if the underlaying probability distribution follows. As I said before, it is assumed in the Black-Scholes model that the price of whatever they are pricing (I trade stocks not futures or options) follows Brownian motion. Essentially this means that the price varies with a normal distrubition centered on zero. A more accurate distribution might be a skewed normal or log-normal, but my area of study is fluid mechanics not stochastic calculus and financial mathematics.

I think forex is really quite different to stocks. Can you really apply the same methods and thinking to forex as you can to stocks?

 

[OrangeDog]

Im sure they are different. My point was that you should read a book on the subject to get a handle on the ins-and-outs before putting your money in the market.

 

[iDimension]

Im sure they are different. My point was that you should read a book on the subject to get a handle on the ins-and-outs before putting your money in the market.

Trust me I plan on doing that. I am already 50% through a free online course. This simply just crossed my mind and I wanted to share it with some people who might have a better understanding of statistics / probability or anything that can highlight why this system is certain to fail. If of course there is a reason.

 

[OrangeDog]

A rule of thumb is that you should pick books that have been around for a long time. In trading there are lots of fads that come and go. There is lots of hype on things that prove fruitless and there can be disdain for things that turn out quite profitable. This was basically one of the causes for the dot com bubble - speculation and over-enthusiasm for businesses that were crap. In buying a book that has stood the test of time you mitigate the risk of owning a text that follows the trends rather than tried and true methods. Futures I think is relatively new, so it might be more difficult to find quality literature.

 

[iDimension]

A rule of thumb is that you should pick books that have been around for a long time. In trading there are lots of fads that come and go. There is lots of hype on things that prove fruitless and there can be disdain for things that turn out quite profitable. This was basically one of the causes for the dot com bubble - speculation and over-enthusiasm for businesses that were crap. In buying a book that has stood the test of time you mitigate the risk of owning a text that follows the trends rather than tried and true methods. Futures I think is relatively new, so it might be more difficult to find quality literature.

Cheers for the tips. I'll try and find a good forex book. I plan on using the demo account for at least 6 months anyway and I want to try lots of different strategies. While reading will give you a great understanding, it's important not to shy aware from experimentation.

 

[OrangeDog]

I agree. I would recommend putting $500 dollars down off the bat, reading your book, and every month depositing a fixed amount into your account. Maybe give yourself a bench mark like 1 chapter a week. As you learn more youll adjust your trading strategy and you will have "skin in the game" so it will be that much more important to you to make the right call.

 

[russ_watters]

OrangeDog, it sounds like you are trying to disagree with me, but I don't see how any of what you said actually does! You seem to be saying that if centered on zero, you should see exactly 50% of next moves be up and 50% be down. I agree: That's what I said.

I only provided the caveat that the long term motion has to show up in there, even if it is small. For example, if a stock goes up 10% in a year and there is 1 trade per second, then the line the distribution is centered on goes up by 0.00006% per trade. That's pretty close to 50/50 centered on zero (the ratio will depend on that distribution you mention: how much it tends to go up or down per trade), but not quite: it matters over the long term because that's your potential profit/loss (otherwise, your profit/loss potential is exactly zero).

[RE:noise] So there may be a 100% chance that the market doesn't rise, but when I randomize a number, it might be sell or buy. So I have a 50% chance of guessing the correct move.

I missed that: you are trying to profit from both gains and losses. In that case, you are filtering out all of the long term motion and focusing entirely on the noise. That makes the long term profit/loss potential of this strategy exactly zero (if executed properly).

 

[iDimension]

I missed that: you are trying to profit from both gains and losses. In that case, you are filtering out all of the long term motion and focusing entirely on the noise. That makes the long term profit/loss potential of this strategy exactly zero (if executed properly).

Thanks Russ. Is this because over a significant length of time you have a 1/3 chance of tripling your money but it's also 3 times harder to get a winning trade?

So overall if you start with £1million, after a very long length of time, you'll walk away with £1million making your profit/loss exactly £0?

Or is it £0 profit/loss for some other reason?

Thanks for explaining.

 

[OrangeDog]

OrangeDog, it sounds like you are trying to disagree with me, but I don't see how any of what you said actually does! You seem to be saying that if centered on zero, you should see exactly 50% of next moves be up and 50% be down. I agree: That's what I said.

I only provided the caveat that the long term motion has to show up in there, even if it is small. For example, if a stock goes up 10% in a year and there is 1 trade per second, then the line the distribution is centered on goes up by 0.00006% per trade. That's pretty close to 50/50 centered on zero (the ratio will depend on that distribution you mention: how much it tends to go up or down per trade), but not quite: it matters over the long term because that's your potential profit/loss (otherwise, your profit/loss potential is exactly zero).

I missed that: you are trying to profit from both gains and losses. In that case, you are filtering out all of the long term motion and focusing entirely on the noise. That makes the long term profit/loss potential of this strategy exactly zero (if executed properly).

I was more like clarifying rather than agreeing or disagreeing.

 

[Vanadium 50]

iDimension, if this worked, it would work whether "1" is buy and "2" is sell or the reverse, right? That's what random means, right?

And if you'd get a net gain on one arrangement, wouldn't you get a net gain on the other? And if that's true, doesn't it make sense to do both? Have half your money under the "1" is buy and "2" is sell rubric, and the other half in the opposite one. In fact, now you can just trade with yourself and cut the commissions, making even more money.

 

[iDimension]

iDimension, if this worked, it would work whether "1" is buy and "2" is sell or the reverse, right? That's what random means, right?

And if you'd get a net gain on one arrangement, wouldn't you get a net gain on the other? And if that's true, doesn't it make sense to do both? Have half your money under the "1" is buy and "2" is sell rubric, and the other half in the opposite one. In fact, now you can just trade with yourself and cut the commissions, making even more money.

I'm not sure I understand what you mean.

If the EUR drops against the USD, that doesn't mean that the USD rise against other currencies.

Let's assume the following, which is typical for a typical trading day.

I randomly select to buy or sell the EUR against the USD. There is a 50% chance that I will select buy and a 50% chance I will select sell. Then I can say with absolute certainty that the market will either go up or down by at least 100pips as it almost always rises or falls 100+pips each day. This is normal in forex. So let's look at this random example that my machine spits out.

Buy EUR/USD and the price rises = win.
Buy EUR/USD and the price falls = lose.

Sell EUR/USD and the price rises = lose.
Sell EUR//USD and the price falls = win.

Now these two trades are totally random but ultimately, it's 50/50 whether I win or lose this trade. Agreed?

So the main question I'm thinking about is -50pips is easier to achieve than +150pips. Therefore I am thinking that even though you only need to win 1 in 3 trades to break even, there is only a 16.6% chance per trade of hitting 150pips and a 83.4% chance of hitting -50. Assuming that 150pips is 3 times harder to hit than -50pips.

*EDIT* OK I think I have it figured. Could someone verify or tell me if this is wrong.

I place 3 completely random trades that look like this
EUR/USD buy (83.4% chance to lose 50pips) (16.6% chance to win 150pips)
GBP/JPY sell (83.4% chance to lose 50pips) (16.6% chance to win 150pips)
AUS/USD sell (83.4% chance to lose 50pips) (16.6% chance to win 150pips)

So after 3 trades there is a
20% chance to win 450pips
80% chance to lose 150pips

So actually these odds are not in my favour? It means I lose 4/5ths of the time?

 

[DiracPool]

When I saw the title of this thread, I thought you were talking about something else. I thought to myself, NO, I'm NOT interested in trading these things. I don't think they're still around, but I used them a lot in the mid 80's before the AIDS crisis made us switch to latex . I'm not gay, btw, not that there's anything wrong with that

http://condoms.typepad.com/condomania/2003/08/forex-condoms.html

 

[russ_watters]

Thanks Russ. Is this because over a significant length of time you have a 1/3 chance of tripling your money but it's also 3 times harder to get a winning trade?

So overall if you start with £1million, after a very long length of time, you'll walk away with £1million making your profit/loss exactly £0?

Right.

 

[jw6661]

buy yinn/yang in the stock market.

buy em both. but 60% of one and 40% of the other.

buy nugt/dust. but 70% of one and 30% the other. gauge the market and figure it out.

as for the orig question. i am not sure. seems like brilliant experiment. and its just that an experiment. unbiased to me.