Randomness of the Market and Trends
Does anyone know how I can attach the worksheet so everyone on this forum can use it?
Now look at the graph closely. What do we see? Just about every thing you would see in a “real” price chart: trends,support and resistance, double tops, bounces of ma’s, etc.
This brings up the question. Is the real market just the result of random price action or is it a presentation of a pseudo-random price action?
There is an excellent (and inexpensive paperback) book on this subject which should be mandatory for all traders (especially FOREX traders): "Fooled by Randomness" by Nassim Taleb.
You might also consider "The (mis) Behaviour of Markets" by Benoit Mandelbrot. Again this is an inexpensive paperback.
There is an excellent (and inexpensive paperback) book on this subject which should be mandatory for all traders (especially FOREX traders): "Fooled by Randomness" by Nassim Taleb.
You might also consider "The (mis) Behaviour of Markets" by Benoit Mandelbrot. Again this is an inexpensive paperback.
I have read "The (mis) Behavior of Markets". Excellent book. Mandelbrot (the discoverer of fractals) is a brilliant mind and excellent writer. I will read "Fooled by Randomness".
I have zipped the worksheet and attached it. I hope it gets attached.
I wrote this spread sheet out of curiosity one day, wondering what the results would be if I used a random sequence of ticks. I was amazed at the results. It gave me a new perspective on the market. It also gave some fuel to Michael Covel's (Trend Trading) criticisms about so-called experts who analyze the market. It would appear that most of the analysis is pointless and meaningless. You can't predict the market anymore then you can predict a random sequence of events.
It also gives fuel to Ed Sekota's remark when he calls fundamentals, "funny-mentals".
Everything you need to know is on the chart, randomly generated or not.
Zoom the spreadsheet to 50 or 75 per cent and notice the tick action around the 8 and 20 ma's. More often than not they hug and bounce of these ma's as in the real charts, but remember what you see in the spreadsheet is all generated by a random sequence!
What is obvious, looking at the worksheet, is that trends emerge even when random ticks are generated. You can't escape trends! Hence, EA's are excellent tools to take advantage of this by using the crossover as an entry and exit signal for the beginning and ending of a trend. The trick is to use the right ma's and biases to capture the moves and filter out the false trends (short time duration movements.)
I've been hint-ing the issue of random market lately. But I've always hesitated in referring to it as 100% random. Had I did that, I'd have to imply that most scalpers take-profit and stop-loss levels puts them at odds much worse than typical casino games. At this point-in-time, I still hold the belief that the market is bias or non-random on a very low frequency. And most of this non-random event is obvious only on larger Time-Frames. Most retail traders Bankrolls cannot ride out large Time-Frames.
I started thinking the market is really random when observing Fundamental News releases and how prices react to them. Put bluntly, the market reaction to Good-News is just the same as it's reaction to Bad-News. Example: Good-News for Eur, you'd expect the Eur to rise at least more than 50% of the time creating a bias. Truth is on some of the best news, it'd head in the opposite direction or just stall. If this is not randomness at it's best then I don't know what is.
I've been hint-ing the issue of random market lately. But I've always hesitated in referring to it as 100% random. Had I did that, I'd have to imply that most scalpers take-profit and stop-loss levels puts them at odds much worse than typical casino games. At this point-in-time, I still hold the belief that the market is bias or non-random on a very low frequency. And most of this non-random event is obvious only on larger Time-Frames. Most retail traders Bankrolls cannot ride out large Time-Frames.
I started thinking the market is really random when observing Fundamental News releases and how prices react to them. Put bluntly, the market reaction to Good-News is just the same as it's reaction to Bad-News. Example: Good-News for Eur, you'd expect the Eur to rise at least more than 50% of the time creating a bias. Truth is on some of the best news, it'd head in the opposite direction or just stall. If this is not randomness at it's best then I don't know what is.
Good points. In spite of the worksheet results, I lean toward the conclusions of Mandelbroit who clearly demonstrated there is more to the market than pure randomness. The historical data of the market clearly contradicts the normal, continous models of orthodox finance theory. In the previous posts I just wanted to point out that trends exist even in random sequences of prices. But clearly we can see price moves that can be attributed to non-random events.
As an example, look at the USDJPY pair at around 10-30-2011. Here we see a horrendous spike up in the currency rate. This is not a random event and all random models would reject the probability of this happening by pure chance as astronomically impossible. But we know central bank intervention of the Bank of Japan can explain this spike. From time to time they buy up U.S. dollars with their yen to artificially strengthen the dollar against the yen, else their whole export economy with relation to the U.S. would be detrimentally effected. Also, look at the magnitude of change of the currency pair rate at the time the Japan earthquake-tsunami disaster happened. The change in price was "caused" by something extraneous to the market.
As a programmer of EA's using moving averages, one must protect against these unexpected spikes by putting in stop losses even though "normally" a trade would be exited by a moving average crossover. Such a catastrophic jump in price would outrace any change in the moving average so that it would take too long for a closing crossover to occur before severe damage is done.
I have zipped the worksheet and attached it. I hope it gets attached.
Just as a computational aside, I noticed that you calculated the B column as, for example
B4 = sum($A$1:A4)
Bn= sum($A$1:An)
It would be a lot less computationally expensive to calculate this as
B4 = B3 + A4
Bn = B(n-1) + An with B1 = A1 to initialize the sequence.
It also demonstrates very clearly that the market movement is incrementally random. In other words the next step moves from the last position.
Don't forget that "spike" events which occur more often than predicted by a Gaussian model can still be random. This is the "fat tails" phenomenon always cited by analysts.
Don't forget that "spike" events which occur more often than predicted by a Gaussian model can still be random. This is the "fat tails" phenomenon always cited by analysts.
True, but they are usually "well-behaved" in the Gaussian model if we assume random Brownian movement. But on page 21 of Mandelbroit's The (Mis)Behavior of Markets, his graphs of the standard deviation of the Dow index show something revealing. The 1987 crash showed a 22 sigma deviation. He stated: "The odds of that are something less than one in 10 (to the 50th power)!"
Clearly, the 1987 crash departed from the classical models. Its as if you were looking at a pollen grain under a microscope suspended in water and you observerved the jitter and noted that the average displacement over time stayed constrained within lets say 5 sigmas from a mean value. But all of a sudden if the pollen shot off the microscope slide and slammed into a wall twenty feet away, we would intuively conclude that something extrinsic to the pollen, such as a gust of wind, accounted for the pollen's motion, not the bombardment of water molecules as Einstein pointed out.
The Long Term Capital Management fiasco that happened years ago, happened because the fund's investment strategy was based on classical models on how the market behaved -- in predicatable, smooth, continous moves. However, when reality hit the wall, and unforseen variables emerged, such as the Russian default on their debt... well as Michael Covel put it, LTCM went through the cheese grater.
The EA's that I write are not based on any type of prediction of the market or any finance model of the market. My strategy is to follow the market because I know I can't predict what the market will do next. Moving averages are very powerful tools to follow the market, even though there is a lag associated with them. But this lag may be good if we view it as confirmation.
By the way if you haven't I recommend watching a fine documentary on YouTube entitled: Quants: The Alchemists of Wall Streets. https://www.mql5.com/go?link=https://www.youtube.com/watch?v=ed2FWNWwE3I It features Paul Wilmott, a well-known Wall Street guru in quantitative finance and some of the warnings he made about the too much reliance on mathematical predictive models.
Incidently, if you do a Fibonacci analysis of the the worksheet graph's black line, you will see near coincidences to the Fibonacci levels, usually not exact coincidences. However, if you looked at a real tick chart and did a Fibonacci analysis, you would find that rarely do the ticks there hit the Fibonacci levels exactly too.
True, but they are usually "well-behaved" in the Gaussian model if we assume random Brownian movement. But on page 21 of Mandelbroit's The (Mis)Behavior of Markets, his graphs of the standard deviation of the Dow index show something revealing. The 1987 crash showed a 22 sigma deviation. He stated: "The odds of that are something less than one in 10 (to the 50th power)!"
That's not fair! My 2008 UK paperback version seems quite different to your copy. Chapter 1 of Part One ranges from page 3 to page 24 and is entitled Risk, Ruin and Reward. The only graphics are two sets of 4 charts, asking which is fake.
Thanks for the link to Quants: The Alchemists of Wall Streets. I will take a look. EDIT: I have seen it before but it was worth watching again :-)
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
Hello everyone. I have attached a worksheet that might enlighten and benefit everyone on this forum. This graph displays a chart of 3 moving averages: the 8, the 20, and the 100 ma (blue, red, and green lines). It also has a black line which is a composite of market tick prices.
What is interesting about this graph is that all of it was generated by a random function, a random sequence of 1’s, 0’s, and -1’s. If you maximize the spreadsheet to the 100 percent zoom level and look under column A, you will see the random function used.
The results from the random function go like this: A “1” value means an up tick, a “0” means no change, and a “-1” means a down tick. This sequence of 1’s,0’s and -1’s is then cumulatively summed and the results displayed in the graph with the ma’s calculated superimposed.
You can generate a new sequence by hitting cntrl c. If you set the zoom factor to 25 percent you will see a bird’s eye view of the graph. 75 percent will give you a more detailed look.
Now look at the graph closely. What do we see? Just about every thing you would see in a “real” price chart: trends,support and resistance, double tops, bounces of ma’s, etc.
This brings up the question. Is the real market just the result of random price action or is it a presentation of a pseudo-random price action? Clearly, if one looked at the eur/usd during 2008, the tanking of the euro could be explained by macro-economic forces— the flight to the safe-haven u.s. dollar currency. Or was it just randomness? Look at the range of the downtrend in 2008 on the weekly chart and compare it with the ranges of other trend movements.
And what about the rest of the year? How much is the price action on a real chart due to randomness and how much to real fundamentals?
As architects of ea’s, one thing clearly bubbles up in my opinion: trends happen, even in a random generated process. Therefore, if you write ea’s that are structured around trends, this will give you the best chance of success.
P.S. I thought I was able to attach the spreadsheet but it doesn't look like it worked. If you are interested, send me a message and I can e-mail it to you.
Does anyone know how I can attach the worksheet so everyone on this forum can use it?