Random Price Movement | Page 7 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Is random really random? 231 replies

random price movement - reasons why your system was wrong 5 replies

My feeling on price movements - they are random 13 replies

GBP/USD daily pip movement vs. EUR/USD daily pip movement 3 replies

Price action pattern, and random thoughts 6 replies

Attachments: Random Price Movement
Exit Attachments

Random Price Movement

  • #121
  • Sep 3, 2006 5:21pm Sep 3, 2006 5:21pm
Quoting Turveyd
Unfortunately, its the casual observers which write the books which you people read, then treat as gospel.
You people? Is that referring to the population of us who clearly know much less than you about trading?
  • #122
  • Sep 3, 2006 5:29pm Sep 3, 2006 5:29pm
Thats who most of the books are written by about trading, if you can't trade profitably write a book about it



Only the people that think it's still random Phil, got a mate who read in some science mag thats its random and therefore has no interest as that must be right LOL
Nothing to it, but to do it!!! Stick to the plan FOOL!!!!
  • #123
  • Sep 3, 2006 6:20pm Sep 3, 2006 6:20pm
Sorry Turveyd,

It's tough to take your Market Commentary serious when I observe posts like these.
Quoting Turveyd
Unfortunately, its the casual observers which write the books which you people read, then treat as gospel.
Quoting Turveyd
Read Chaos Thoery, it better explains the markets than Random.
Quoting Turveyd
No, using SMA's I can tell you, exactly when a trend is changing direction
Quoting Turveyd
Markets move from Buying / Selling Pressure, if you can identify the market has selling pressure at that time, get on short, 80% of the time you'll profit.
20% of the time, the seller finishes and the buyers move it up again, or a bigger buyer comes in to take advantage of the lower prices.
Quoting Turveyd
Fairly common, when I place a trade for it to hover around the Zero area for hours, when it was moving nicely, tis very strange.
Quoting Turveyd
That and lets face it Likely only 5% of the regular posters on here, and truely long term profitable, sadly I'm not in the 5% yet but soon .
Quoting Turveyd
I'm logging off, first full week trading, it's a killer in more ways than 1.
All this and trade recommendations too...
  • #124
  • Sep 4, 2006 9:16pm Sep 4, 2006 9:16pm
Granted, this thread's digressed, but I do have an idea that *may* be interesting to some...

I'm 100% in on the idea that random coin flips can imitate very closely what the market does. My own studies have shown that there is a near 50/50 split in market movement, and thus, no surprise that the clever excel spreadsheet that kicked this whole thing off is capable of representing that movement.

(and just because I posted an earlier mesage about 'hey I like math and markets aren't random,' doesn't mean I didn't get the point)

SO... why not test trading systems against randomly generated data in addition to real market data? I'm not saying we should throw out the back-testing of real data - I'm asking is there something to be learned by back-testing random (or "pseudo-random," if you prefer) data? If we produce 1000 sets of random data and back-test a basic 50 SMA / 200 SMA cross system on all of them, would we get the same statistic results as testing them on real FX data? If we did, what would that mean? If we didn't, what would THAT mean?

I'm not advocating it... I just think its a curious question in light of the solid point that this thread (and the related resources) has made - that there is some correlation between a 50/50 probability and the way the market moves... regardless of whether that movement is based on sound reasoning, random movement, or the still-resounding repercussions of Elvis swinging his hips.

Sorry for another long post... what can I say - I'm verbose.

-miscon777
Et ecce equus pallidus...
  • #125
  • Sep 6, 2006 10:50am Sep 6, 2006 10:50am
Quoting miscon777
Granted, this thread's digressed, but I do have an idea that *may* be interesting to some...

I'm 100% in on the idea that random coin flips can imitate very closely what the market does. My own studies have shown that there is a near 50/50 split in market movement, and thus, no surprise that the clever excel spreadsheet that kicked this whole thing off is capable of representing that movement.

(and just because I posted an earlier mesage about 'hey I like math and markets aren't random,' doesn't mean I didn't get the point)

SO... why not test trading systems against randomly generated data in addition to real market data? I'm not saying we should throw out the back-testing of real data - I'm asking is there something to be learned by back-testing random (or "pseudo-random," if you prefer) data? If we produce 1000 sets of random data and back-test a basic 50 SMA / 200 SMA cross system on all of them, would we get the same statistic results as testing them on real FX data? If we did, what would that mean? If we didn't, what would THAT mean?

I'm not advocating it... I just think its a curious question in light of the solid point that this thread (and the related resources) has made - that there is some correlation between a 50/50 probability and the way the market moves... regardless of whether that movement is based on sound reasoning, random movement, or the still-resounding repercussions of Elvis swinging his hips.

Sorry for another long post... what can I say - I'm verbose.

-miscon777
miscon,

There are a variety of ways to do this. Some testing methods employ monte carlo techniques which jumble your trades to show the historical significance of the exact sequence of your trades. Some methods jumble the input data before it reaches the testing platform. Others, create new data out of existing data (http://www.breakoutfutures.com/Newsl...letter1205.htm). Given these possibilities it is completely logical that you would consider testing your system on a series of randomly generated data. A good system captures a piece of price action. It is not dependent upon an exploitable price inefficiency. One could argue that point if they paid a fraction of a pip spread but for the rest of us who routinely pay 3-4 pips that is not a viable option for us.

Phil
  • #126
  • Sep 6, 2006 1:03pm Sep 6, 2006 1:03pm
Miscon, stick to the real thing, you'd be purely learning how to trade a psuedo random number generator, nothing more, even if by a amazing coincidence it brings out trade like patterns in ways ( fluke entirely )

Random, won't and can't allow for news related moves, up trends and down trends.

Infact I think playing a entirely random market would be all about luck, nothing more, might teach good money management stills maybe though.
Nothing to it, but to do it!!! Stick to the plan FOOL!!!!
  • #127
  • Sep 6, 2006 1:48pm Sep 6, 2006 1:48pm
Quoting Turveyd
Miscon, stick to the real thing, you'd be purely learning how to trade a psuedo random number generator, nothing more, even if by a amazing coincidence it brings out trade like patterns in ways ( fluke entirely )

Random, won't and can't allow for news related moves, up trends and down trends.

Infact I think playing a entirely random market would be all about luck, nothing more, might teach good money management stills maybe though.
Dude, that's the point. If your system can capture market movement from entirely random moves then it is likely to catch movement in real markets where biases and inefficiencies exist. It's a testing methodology used by real traders.
  • #128
  • Sep 8, 2006 7:40am Sep 8, 2006 7:40am
Quoting miscon777
SO... why not test trading systems against randomly generated data in addition to real market data? I'm not saying we should throw out the back-testing of real data - I'm asking is there something to be learned by back-testing random (or "pseudo-random," if you prefer) data? If we produce 1000 sets of random data and back-test a basic 50 SMA / 200 SMA cross system on all of them, would we get the same statistic results as testing them on real FX data? If we did, what would that mean? If we didn't, what would THAT mean?

-miscon777
I have done that once for some random values and some currencypairs. The results are what you could expect. The average at the random values was around zero and the average at the pairs was slightly higher. You should expect that because the expectation of the random values is zero and the FX-market is trending a little bit.

I am not sure what happens when you do it for a very very long time. Because one of the inefficiencies in the forex-market on the very long term is that interest-rates are mean-reversing, not trending. That means that interest-rates tend to stay close to each other. If one rate is 20% and the other 5%, then you can expect that the chance that the difference will be smaller in the future will be higher than the opposite. And differences in the interest-rates are on the long term driving forces in the forexmarket.
  • #129
  • Sep 8, 2006 10:53am Sep 8, 2006 10:53am
Thanks for the feedback, folks.

Phil - I read about Monte Carlo analysis on adaptrade.com shortly after posting this, and managed to make the connection between what I had said / asked there and what MC analysis is all about. Naturally, I was pretty pleased to see that I wasn't alone in thinking that real data has value, but so does messing with it a bit to simulate and account for the inability to mathematically predict future movement.

No such thing as an original idea, right? Heheh.

JosTheelen, you are right on when you point out that a 50/50 random generation will hit 0 over time, and particularly so in large sets of data... and that the markets are only slightly better in terms of probability. Thanks for a good reminder.

-miscon777
Et ecce equus pallidus...
  • #130
  • Dec 15, 2006 7:21pm Dec 15, 2006 7:21pm
Here's my shot at reviving this thread. Some may have had enough of a certain someone . But discussions are what forums are all about.

From Larry Williams Long-Term Secrets to Short-Term Trading (Pages 14-15)

The Non-Random Market

Commodity Percent Time Close > Open
Pork Bellies 51
Cotton 53
Beans 51
Wheat 52
Brit Pound 56
Gold 52
Nikkei 55
Eurodollar 57
US Bonds 52
S&P 500 53
Average 53.2

Now the same thing except Close > Open percentages after previous day was a down close and previous 2 days were a down closes:

Pork Bellies 55 55
Cotton 53 55
Beans 56 56
Wheat 53 55
Brit Pound 57 56
Gold 58 55
Nikkei 56 60
Eurodollar 59 56
US Bonds 54 52
S&P 500 55 53
Average 55.8 55.2

"While price action does involve a large degree of randomness, it is far from a totally random game"

Any thoughts?
  • #131
  • Dec 16, 2006 5:37am Dec 16, 2006 5:37am
Quoting miscon777
The reality is that markets are mostly based on large institutions making decisions with large orders. There's nothing random about that - its a series of human beings making decisions for good reasons, and even if mathematical "randomness" can reproduce the resulting patterns, it doesn't change the core reasons that those patterns develop.


-miscon777
A few years ago I coded up the mathematical test for randomness - commonly known as the drunken stagger, and proved to myself that mkt movements were very very near random.

The test says that if one marks a series of steps of size delta both up and down from zero then if the process is random it will take n squared steps to traverse n steps away from zero regardless of direction (+ve or -ve from zero). Therefore for the mkt to move 2 stepsizes away from 0 it will require on average 4 steps to achieve, 9 moves to achieve 3 steps etc. Using several years of tick by tick currency data I experimented with many delta sizes in terms of pips across all currency pairs available and found the average to be unerringly close to n squared to move n steps, regardless of currency or stepsize.

By definition if a process is random it is unpredictable. Due to my experiences with the results quoted above, I have requested many times on FF for people to quote their results using whatever technical analysis approach they wish over a decent number of trades but I have yet to see a single result quoted. If anyone can quote results that can demonstrate non random behaviour then it would add greatly to this discussion. A good test would be to use say 2 years of data to tune parameters and the next 2 years for testing. The unseen testing results are the results to be quoted.

On another note - there have been a number of posts on this thread talking about how MM can help you beat randomness. I don't believe this to be true - it can only enhance your profit growth if your system has a positive expectation (i.e overall probabilty of winning greater than 50% - if wins and losses are not equal this needs to be factored in when calculating expectation), if not it will only expedite your path to ruin !
  • #132
  • Dec 16, 2006 11:22am Dec 16, 2006 11:22am
It could be
1) Coincidence - it is very very unprobable that number of c>o and o>c are exactly equal. Deviation of 1-3% is definitely not a sign of nonrandomness.
2) Result of inflation.
  • #133
  • Dec 16, 2006 11:38am Dec 16, 2006 11:38am
Quoting philmcgrew
I got a Head and Shoulders pattern out or a random coin flip:
In TA, you need volume to confirm a break out, which in forex is estimated from futures trading, but is volume just the same. This program does not have volume to confirm technical analysis moves, and volume is a part of technical analysis. So this program does not prove anything.
  • #134
  • Edited 11:59am Dec 16, 2006 11:43am | Edited 11:59am
Quoting richp
A few years ago I coded up the mathematical test for randomness - commonly known as the drunken stagger, and proved to myself that mkt movements were very very near random.

The test says that if one marks a series of steps of size delta both up and down from zero then if the process is random it will take n squared steps to traverse n steps away from zero regardless of direction (+ve or -ve from zero). Therefore for the mkt to move 2 stepsizes away from 0 it will require on average 4 steps to achieve, 9 moves to achieve 3 steps etc. Using several years of tick by tick currency data I experimented with many delta sizes in terms of pips across all currency pairs available and found the average to be unerringly close to n squared to move n steps, regardless of currency or stepsize.

By definition if a process is random it is unpredictable. Due to my experiences with the results quoted above, I have requested many times on FF for people to quote their results using whatever technical analysis approach they wish over a decent number of trades but I have yet to see a single result quoted. If anyone can quote results that can demonstrate non random behaviour then it would add greatly to this discussion. A good test would be to use say 2 years of data to tune parameters and the next 2 years for testing. The unseen testing results are the results to be quoted.

On another note - there have been a number of posts on this thread talking about how MM can help you beat randomness. I don't believe this to be true - it can only enhance your profit growth if your system has a positive expectation (i.e overall probabilty of winning greater than 50% - if wins and losses are not equal this needs to be factored in when calculating expectation), if not it will only expedite your path to ruin !
Exactly what I am saying all the time. One can not beat random market if it is completely random by any means.

However I disagree with your opinion on money management. Money management could mean a) Way of determining how much I gonna risk in the trade. In this case, you are right and moneymanagement can no way affect percentual output of the system. Moneymanagement could also mean b) way of determining when to exit a trade. It this case, it is basicly part of the system which should give you an adventage over the market. And it does not matter whether you exit on cross of certain set of indicators or absolute profit target. Both first both second method should give you the edge.

My opinion of trading strategies is this: The closer your system comes to psychology of market participants(S/R are very close, on the other hand, fibo, elliots and craps like that are far far away), the more accurately you will be able to predict future markets movements.

And another think which came into my mind: More participants in the market are speculating on price movement, the more random the market gets. Since big banks do speculate on interest rates, not on price movement, I considef FX to be perfectly OK.
  • #135
  • Dec 16, 2006 11:48am Dec 16, 2006 11:48am
Quoting raistin
My opinion of trading strategies is this: The closer your system comes to psychology of market participants(S/R are very close, on the other hand, fibo, elliots and craps like that are far far away), the more accurately you will be able to predict future markets movements.
Elliot Wave is based completely on psychology of market participants, as reflected in the waves, and it uses trendlines, S/R as you describe. I think you should learn about it before declaring it crap, w/o having a clue about what it is.
  • #136
  • Dec 16, 2006 11:56am Dec 16, 2006 11:56am
Quoting wwwin
Elliot Wave is based completely on psychology of market participants, as reflected in the waves, and it uses trendlines, S/R as you describe. I think you should learn about it before declaring it crap, w/o having a clue about what it is.
The fact that Eliot himself didn´t make single penny using this strategy is overwhelming proff. I´would rather invest my time in examining other strategies.
  • #137
  • Dec 16, 2006 12:08pm Dec 16, 2006 12:08pm
Quoting raistin
The fact that Eliot himself didn´t make single penny using this strategy is overwhelming proff. I´would rather invest my time in examining other strategies.
That is your choice, but calling it crap w/o having a clue about what is based on does not make any sense. In particular, when you state that you believe in market psychology and S/R, both integral parts of EW, that was the clue as to how clueless you are about Elliot Wave
  • #138
  • Dec 16, 2006 12:41pm Dec 16, 2006 12:41pm
As far as I know Elliot suppose, that maket is moving in waves and that every movement can be divided into 5 or 3 waves. To me, It does not seems to corespond to market psychology. Now lets cut the offtopic -)
  • #139
  • Edited 1:12pm Dec 16, 2006 12:59pm | Edited 1:12pm
Quoting raistin
As far as I know Elliot suppose, that maket is moving in waves and that every movement can be divided into 5 or 3 waves. To me, It does not seems to corespond to market psychology. Now lets cut the offtopic -)
OK, It is a very complex subject and takes months if not years to learn well. You cannot learn it well by reading about it in FF. A lot of people misuse it here. I guess my point is not to judge it by what predictions are being made on FF supposedly based on EW. It's not the theory's fault if it is not correctly employed, just like it is not your car's fault if you run a red light, unless of course the brakes failed. Cheers.
  • #140
  • Dec 16, 2006 6:05pm Dec 16, 2006 6:05pm
Simply put technical analysis works because we use it!

Yes it is something made up out of humans inherit need to understand and simplify things that can not be understood easily.

100 years ago trading the stock market technically would have been close to impossible. 30 years ago technical analysts were laughed and thought to be crazy. Today just about every trader in the world uses some form of analysis.

Yes humans to have a need to make a pattern out of everything and to simplify the most complex thing. Easy is one of the most powerful words in the English language. We want everything to be easy and technical analysis simplifies a complex market it brings it all down to a few lines.

Are those lines useful? Yes they are useful because of humans inherit need to simplify everything in life.

They are useful because however many years ago thousands of people bought into the idea of an easier way to trade forex. They are going to continue to grow more and more useful as thousands upon thousands of new traders come into this market every week all using technical analysis.

The fact is tech analysis is slowly becoming more powerful than fundamental analysis and soon fundamental analysis will become useless. When we get 100,000 new retail traders next year and 200,000 the year after most of which using tech analysis fundamentals will become obsolete. I see it myself the central banks are losing their grip on the market they do not have as much effect as they use to.

It's human nature we like to simplify things we had ovens we made microwaves we had bikes we made cars.

What it all comes down to is technical analysis if used right is the most effective way if predicting market movement.
1 trader viewing now