The Question of Randomness

[Tams 20]

I humbly disagree. Profitability of a trading system is defined by your ability to press winners and to run fast in the face of adversity.

Mathematicians (or any scholar, for that matter) make for terrible traders.

Most think they have to be right in order to win, hence they fail to bail if the trade turns bad. Yep, wide generalizations FTW :rofl:

 

But we disgress...

 

how did you come up with this conclusion?

is this a personal hunch? or a "FACT"?

[Do Or Die 10]

Profitability of a trading system is defined by your ability to press winners and to run fast in the face of adversity.

 

Profit comes from positive expectancy. Trading is a negative sum game.

 

This is not about the question of disagreeing, I'm a little surprised because this is the first time I've heard such perception.

[zdo 20]

Avarice,

 

I understand what you're saying but " Profitability of a trading system is defined by your ability to press winners and to run fast in the face of adversity" is not a generalization.

 

"Profitability of a trading system is defined by your ability to press winners and to run fast in the face of adversity " is highly system specific, made to appear to be a ubiquitous generalization by not explicitly identifying what kind of system... it's not a safe robust generalization ...

 

zdo

[MightyMouse 12]

After getting 19 out of 20 correct on the 2nd try at Aurora, I'd have to say that the human eye can learn to identify the difference between a real data series and a computer generated series, and do it very quickly. But is this skill useful for profit?

 

To the point regarding technical analysis: Most people who come into trading never question the roots of technical analysis or the nature of financial (price) data.That usually comes later after finding that the past may not do a very reliable job of predicting the future, despite how many lines or types of analysis are applied. Traders should be required to take a course in econometrics to learn about how to identify the true nature of price data and how to handle nonstationary time series. Financial data has stochastic trend. Because price is not deterministic, all those fancy TA indicators are useless when applied to raw price data, as they are all designed to work with stationary data.

 

The real kicker is that in most cases those moving averages that are so often applied to price by traders to predict future price movement have a lower order of integration than the underlying price data! It's like trying to predict a complex system with a simple system. Exactly how is that going to work? If only traders could profit from buying or selling a moving average (rather than the underlying price). Then it would be relatively easy to profit.

 

Does something make you think that a trader can't profit from trading moving average?

[Tams 20]

When I first came across this ages ago it made me sit up and it did the same when I came across it again recently. Take a look at the chart I have attached. What is it you ask? Oil, gold, S&P 500. Nope, it's nothing. It's a chart which I had excel generate simply by getting it to choose a tick higher, a tick lower or unchanged from the last price. But it looks very familiar doesn't it?

 

Now we all know the markets aren't random , but really when you look at an example like this it makes you wonder. What are the implications of such an exercise?

 

Here are some thoughts:

 

1. There is an element of perceived randomness to the market at times which may or may not perpetuate further 'random' activity.
    
   
2. Markets are not random at all as conditions present at any given moment, caused the subsequent price movement. However, when you look at historical charts, they have the appearance of being randomly generated as there is no context when viewing historical prices.
    
   
3. Unless we assign useful context to market movements, any analysis which is done is never going to be much better than random.
    
   
4. Would you trade the product in my chart if you knew it were completely random???
   

 

What do you guys reckon?

 

if human behavior is random, the market will be random,

because the market is a reflection of human behavior.

[cjforex 10]

There has been some interesting math relating to this exact question.

 

The process goes something like this:

 

1. Assume that whatever you were trying to evaluate as a prediction method got lucky by random chance, and given enough chances, could get lucky with random data.

 

2. Set this as your null hypothesis.

 

3. Prove it's not true. Disprove your null hypothesis.

 

The latest math solving this problem even had a patent issued and was incorporated into analysis software for the financial markets.

 

Google: white's reality check bootstrap data-snooping

 

Use combinations of the above words to see papers related to the topic.

 

The process of the solution is not that difficult, and there are lots of clever methods including beating the monkey.

 

If you are in the business of evaluating strategies then it's useful stuff.

 

Something i am very interested in as well if anyone wants to swap ideas.

 

Cj

[MightyMouse 12]

Profit comes from positive expectancy. Trading is a negative sum game.

 

This is not about the question of disagreeing, I'm a little surprised because this is the first time I've heard such perception.

 

He could not be more right. Profits come from knowing how to take more money from the market than you leave in the market. Huge profits come from being at the right place at the right time.

[Do Or Die 10]

He could not be more right. Profits come from knowing how to take more money from the market than you leave in the market. Huge profits come from being at the right place at the right time.

 

Maybe we are overlapping things.

 

"Huge profits come from being at the right place at the right time." You need a setup with positive expectancy to accomplish this. You cannot simple make profits by buying out anywhere and following 'money management'.

 

Timing and money management are two totally different things. Timing is achieved through selection of non-random pattens only (your 'edge'). Timing has no meaning in purely random data.. Money management can maximize your productivity, not create it. Again this is the same reason mathematicians do not rule the market.

 

Try backtesting same system on random data and real data- while considering slippage and commissions.

[jackj 10]

Maybe we are overlapping things.

 

"Huge profits come from being at the right place at the right time." You need a setup with positive expectancy to accomplish this. You cannot simple make profits by buying out anywhere and following 'money management'.

 

Timing and money management are two totally different things. Timing is achieved through selection of non-random pattens only (your 'edge'). Timing has no meaning in purely random data.. Money management can maximize your productivity, not create it. Again this is the same reason mathematicians do not rule the market.

 

Try backtesting same system on random data and real data- while considering slippage and commissions.

 

For someone with the capability, it should be quite possible. It sure seems as though mm alone could produce profits. To eliminate slippage, assume all limit fills. Enter anywhere with 2 contracts and a 2 tick stop. On winners, take 1 contract off at +2 ticks, and move to b/e +1, or +2 with some other target #2 of +4, +6, etc. or some trailing stop. Over a long enough series, you'll have half of the chart having runs of varying lengths below your entry, and hence of 2 tick losses (x 2 contracts + commissions); and half of the time have at least 2 tick winners, with some 4 tick, 6 tick, 8 tick, etc., depending on how one configures the target #2 or trailing stop. Why shouldn't this create enough to offset the commissions, as low as they are now? Just 1 tick of extra (over 2 ticks) profit offsets commissions. I'm not saying this IS right; just that it sounds like it would work. There appear to be some here with much a much higher math background than I who can maybe point out the flaw(s) in this thinking.

[jackj 10]

Also, Do or Die, thanks for those links in the other thread you pointed out, on the studies done on Relative Strength. Interesting. I thought when studies supporting a non-random market were brought up that the paper/book by Andrew Lo was being referenced. Here's that one: Contents for Lo & MacKinlay: A Non-Random Walk Down Wall Street

 

Also, aside from MA's, which price often doesn't react as strongly or consistently to, and hence one wonders if it's random when it does, trendline support - to me - seems much more assured and predictable, whether horizontal due to prior price action there, or angled in a trending market. But I have no objective proof of that, other than looking at a chart and noting (in the case of horizontal s/r) that if price turned there previously, you at least get a greater than "average" reaction there again. Which, as was previously pointed out by Tams(?), makes sense as price is caused by human behavior, and people see what price did there previously, and react to that with their buying/selling. And, hence, not random. As to Do or Die's assertion that one wouldn't trade that (double-top) all alone, I'd say that one could, albeit for a scalp. That you'll get a predictable-enough stronger than average reaction there, even if it subsequently breaks through. Enough for a scalp.

[MightyMouse 12]

Maybe we are overlapping things.

 

"Huge profits come from being at the right place at the right time." You need a setup with positive expectancy to accomplish this. You cannot simple make profits by buying out anywhere and following 'money management'.

 

Timing and money management are two totally different things. Timing is achieved through selection of non-random pattens only (your 'edge'). Timing has no meaning in purely random data.. Money management can maximize your productivity, not create it. Again this is the same reason mathematicians do not rule the market.

 

Try backtesting same system on random data and real data- while considering slippage and commissions.

 

Huge profits come from being at the right place at the right time and knowing how to take more money from the market than the market takes from you.

 

Being at the the right place at the right time is luck. knowing how to take money from the market is not. Knowing how to take money from the market is your consistently is the "edge".

 

One can certainly not have an edge and be at the right time at the right place and make a huge profit.

[pmwhite 10]

Does something make you think that a trader can't profit from trading moving average?

 

I'm sure it's possible to create profitable systems out of moving averages in the short term or for a trader to profit from such in the short term. I have done this. But long term profitability is or should be the goal of any trader or system. The problem is that because the data that the average is applied to is of a higher order than the moving average, and because price data has stochastic trend, meaning the mean moves over time, and can have infinate variance, a moving average (which is designed to work on mean stationary data) does not adapt very well (nor can it).

 

To create a profitable trading strategy using moving averages requires that the average adapt to current market conditions, which isn't possible just by applying a MA to price and optimizing it to past data. The reason this isn't possible is because the price data follows a stochastic trend, meaning the trend is not deterministic, follows no set pattern and because the mean moves and price doesn't have to return to the mean, the whole approach is flawed. As a result, the average will give false signals and will not do a very good job of describing the actual trading opportunities. Differencing and detrending the data will actually create data that is of the same order of integration as the MA. So a MA applied to differenced, detrended data should be useful. The kicker though, is that you don't buy or sell the diff/detrended data, you buy/sell the raw prices, so while a MA applied to differenced / detrended data my prove useful as an indicator, finding a fixed relationship that can be exploited is still a must if one wants to profit long-term from using a moving average strategy.

[pmwhite 10]

if human behavior is random, the market will be random,

because the market is a reflection of human behavior.

 

The logical implication of this is that if human behavior ISN'T random (which it generally is not) then the market will be non-random. However, just because a trader who places orders does them for a particular economic reason that isn't random, does not mean that the combination of all those determined actions of all market participants are not random. I think it is useful to replace the term "random" with non-deterministic or unpredictable. Can you (or anyone) really predict what orders will come into the market next to drive price up or down? If the answer is no, then I would argue that the markets are non-deterministic and may be considered a random walk for analytical purposes.

 

On a slightly different topic, Galton showed that when a large number of random events (coin tosses) are combined, they form a normal distribution. A reference may be found (among many sites) here: The Normal Curve and Galton's Board

[Tams 20]

The logical implication of this is that if human behavior ISN'T random (which it generally is not) then the market will be non-random. However, just because a trader who places orders does them for a particular economic reason that isn't random, does not mean that the combination of all those determined actions of all market participants are not random. I think it is useful to replace the term "random" with non-deterministic or unpredictable. Can you (or anyone) really predict what orders will come into the market next to drive price up or down? If the answer is no, then I would argue that the markets are non-deterministic and may be considered a random walk for analytical purposes.

 

On a slightly different topic, Galton showed that when a large number of random events (coin tosses) are combined, they form a normal distribution. A reference may be found (among many sites) here: The Normal Curve and Galton's Board

You are going down a path that was not my intent. But you are free to conjure up deductions or "creations" of your imagination. After all, none of these are new.

[jackj 10]

Huge profits come from being at the right place at the right time and knowing how to take more money from the market than the market takes from you.

 

Being at the the right place at the right time is luck. knowing how to take money from the market is not. Knowing how to take money from the market is your consistently is the "edge".

 

One can certainly not have an edge and be at the right time at the right place and make a huge profit.

 

I believe you have a couple of typo's in your post, though I think I understand what you're saying, MM. But, so that I don't misunderstand, would you care to share anything at all by way of an explanation as to HOW one does the above? I mean, really, you make money by taking more out than you're giving?! Who'd have thunk that...

 

But I think you're agreeing with me, that the edge is (or in my words, CAN be) in the money management aspect. Though I'm not sure how you view the "being in the right place at the right time" aspect of it, particularly when you've said that that part is random. Again, I'd appreciate any expounding that you can do on your thoughts, and how best to take more than one gives.

[jackj 10]

Re: PM White's posts on MA's, I thought (and think) that I was out of my league earlier when someone opined re: mathematicians and trading. But I think this is a perfect example of what that poster may have been alluding to. Mr. White explained a great mathematical, objective, random data-based analysis of what may or may not be supposed from that data. But that all is pointless, as the data is not simply random (though I guess that's what we're trying to prove - or not - in this thread). So, I'll just say that my assertion is that as people are those who determine what price is printed on that next price bar, and they factor in what's come before it (therefore it's not an independent event), there is a reason that a moving average may have more of an effect that posited in those posts.

 

Because the reality is that traders, who are the ones who determine where that next price prints, are watching a moving average and are often (enough) trading based upon that. As well as trendlines. On some timeframes more than on others. I'm always wary of my own perceptions of charts - always keeping in mind Taleb's Fooled by Randomness book and principles - but I've seen enough daily charts that show a very strong reaction right at the 50 and 200 sma that I'm convinced that it isn't merely a coincidence. No objective proof, mind you, but perhaps Prof. Lo's paper/book has that.

[MightyMouse 12]

I'm sure it's possible to create profitable systems out of moving averages in the short term or for a trader to profit from such in the short term. I have done this. But long term profitability is or should be the goal of any trader or system. The problem is that because the data that the average is applied to is of a higher order than the moving average, and because price data has stochastic trend, meaning the mean moves over time, and can have infinate variance, a moving average (which is designed to work on mean stationary data) does not adapt very well (nor can it).

 

To create a profitable trading strategy using moving averages requires that the average adapt to current market conditions, which isn't possible just by applying a MA to price and optimizing it to past data. The reason this isn't possible is because the price data follows a stochastic trend, meaning the trend is not deterministic, follows no set pattern and because the mean moves and price doesn't have to return to the mean, the whole approach is flawed. As a result, the average will give false signals and will not do a very good job of describing the actual trading opportunities. Differencing and detrending the data will actually create data that is of the same order of integration as the MA. So a MA applied to differenced, detrended data should be useful. The kicker though, is that you don't buy or sell the diff/detrended data, you buy/sell the raw prices, so while a MA applied to differenced / detrended data my prove useful as an indicator, finding a fixed relationship that can be exploited is still a must if one wants to profit long-term from using a moving average strategy.

 

Seems like you won't be using a moving average.

[Peterthemonkey 10]

The logical implication of this is that if human behavior ISN'T random (which it generally is not) then the market will be non-random. However, just because a trader who places orders does them for a particular economic reason that isn't random, does not mean that the combination of all those determined actions of all market participants are not random. I think it is useful to replace the term "random" with non-deterministic or unpredictable. Can you (or anyone) really predict what orders will come into the market next to drive price up or down? If the answer is no, then I would argue that the markets are non-deterministic and may be considered a random walk for analytical purposes.

 

On a slightly different topic, Galton showed that when a large number of random events (coin tosses) are combined, they form a normal distribution. A reference may be found (among many sites) here: The Normal Curve and Galton's Board

 

Think of it as cause and effect.

 

The specific circumstances affecting a single trader lead to his response, as you noted, This is also true of all traders in the available pool (ie. currently at work) at any given time. All of these individually specific determinants, intertwine to become the determinant for the price action at any given time.

 

Simply because we can not know with certainty all the factors that will affect any one individual at any point in time and the response it will generate, much less a group, does not infer that the result generated is random.

 

If you replicate the cause you replicate the effect.

 

The observable price action in a chart refutes randomness. Random price action would look like this: 101, 102, 500, 0.25, 50........... It wouldn't be confined by the succession of factors affecting current events into the future.

 

These factors impose order to price movements. The fact we can not predict it with certainty signifies we don't fully understand how it functions, are unable to analyze all the variables, or ignore some of the factors which determine the effects manifested at any given time, giving the appearance of a random outcome. Thus, there is a lack of certainly, of an always inconclusive analysis, and a reliance on a degree of luck in trading. Yet, just as the price action demonstrates a certain orderliness, we can reduce the degree of luck in any trade by carefully choosing our trades through analysis, as opposed to trading at random, at any time without a reason.

[MightyMouse 12]

I believe you have a couple of typo's in your post, though I think I understand what you're saying, MM. But, so that I don't misunderstand, would you care to share anything at all by way of an explanation as to HOW one does the above? I mean, really, you make money by taking more out than you're giving?! Who'd have thunk that...

 

But I think you're agreeing with me, that the edge is (or in my words, CAN be) in the money management aspect. Though I'm not sure how you view the "being in the right place at the right time" aspect of it, particularly when you've said that that part is random. Again, I'd appreciate any expounding that you can do on your thoughts, and how best to take more than one gives.

 

Wow. Yes typos. I was in a hurry.

 

Money management is important. Risk management, I think, is more important.

 

Quickly cut losses and admit that you might be wrong right away.Small losses are awesome compared to big losses. Doing so requires you to have a certain level of humility.

 

When you are in a profitable trade, look for reasons to stay in the trade instead of reasons to get out. In doing so, you will frequently donate a decent profit, but it allows you to achieve larger profits. I am not suggesting that you recklessly press every profitable trade. On the other hand, always be open to the possibility that something bigger could be happening. If something bigger is happening, you want to leverage your position to take advantage of it.

 

Humility and fearlessness are not items that you will find on a chart. How or why you enter make less of a difference than what you do when you are in.

 

I had a thread that detailed my trades. You need to find a method that you are comfortable with.

[pmwhite 10]

Seems like you won't be using a moving average.

 

I can see how you would draw that conclusion but that isn't correct. The point I'm trying to make is that averages are tools that need to be applied to the correct type of data. Much like a child who is handed a hammer who then sees nails everywhere, technical traders have averages and therefore attempt to use them on all types of price data and in all types of ways without evaluating the nature of the tool or the underlying data.

 

Price data can be differenced and detrended. It is possible to apply averages to this type of data to get more useful results. However, as I stated before, this doesn't solve the complete problem but surely is better than 99% of the currently available technical indicators. In fact most of them would work better when applied to differenced and detrended data because most technical indicators are based on moving averages which are designed to work with time series with a stationary mean. By differencing and detrending the data it is possible to force the price data to have a stationary mean and thus price becomes a deterministic system.

[MightyMouse 12]

I can see how you would draw that conclusion but that isn't correct. The point I'm trying to make is that averages are tools that need to be applied to the correct type of data. Much like a child who is handed a hammer who then sees nails everywhere, technical traders have averages and therefore attempt to use them on all types of price data and in all types of ways without evaluating the nature of the tool or the underlying data.

 

Price data can be differenced and detrended. It is possible to apply averages to this type of data to get more useful results. However, as I stated before, this doesn't solve the complete problem but surely is better than 99% of the currently available technical indicators. In fact most of them would work better when applied to differenced and detrended data because most technical indicators are based on moving averages which are designed to work with time series with a stationary mean. By differencing and detrending the data it is possible to force the price data to have a stationary mean and thus price becomes a deterministic system.

 

My mistake, then, if you do intend to use MA in your trading. I do not use them, but I do see times where using a MA would have been the best solution so if someone was using one the way I would have, he would have squeezed a lot of money out of the market. Hindsight.

 

If one takes such a rigorous approach to technical analysis, she will end up standing on the sidelines, questioning whether she can ever develop a system based on lagging indicators. The problem becomes whether or not the mean remains stationary or not over the period that you are trading and using the MA.

 

I am not suggesting that a rigorous approach is a wrong approach. If you are going to rely on TA then the conclusions you draw after careful analysis using a rigorous approach will be the right conclusion.

[SunTrader 10]

Most indicators are based on rate of change.

 

Moving averages are just one variation, albeit the most popular - especially with noobs.

[Tams 20]

Most indicators are based on rate of change.

 

Moving averages are just one variation, albeit the most popular - especially with noobs.

 

you are calling 99% of the population noob?

[SunTrader 10]

you are calling 99% of the population noob?

Yes if 99% use MA's. I don't think they do though, but someone has to make up the 90% quotient who lose.

[cjforex 10]

That moving average correlation makes lots of assumptions.