Are share prices random?

[darkhorse]

Originally posted by Gordon Gekko

Traden4Alpha is right, this has been an overlooked post, imo. is the answer to this question yes or no?

The only real answer is 'it depends,' because it completely does. A discretionary trader will be looking at dozens if not hundreds of variables comprising the whole picture- the lifespan of the trend, the relative action of sectors and industries, indexes, treasuries, news, the macro picture....so this small piece of data would likely be less than one one hundredth of one percent of the info mosaic that leads to a trading decision.

A mechanical trader, on the other hand, will have a system with set rules that treats all data in a binary way, meaning some information it pays rapt attention to (1) and most information it completely ignores (0). A computer does not 'glance' or 'peek'- it looks at what you tell it to and nothing else. A mechanical trader who does not take percentage days up or down into account will not have a rule set that makes use of this information. Whether he should or not is another question entirely. If a mechanical trader Does have a rule that takes percentage up days or down days into account, it is still no doubt built into a broader program.


Better to phrase the question another way?

p.s. by the way alpha, how does a trader shoot himself with a rope?

 

[Gordon Gekko]

Originally posted by darkhorse
Better to phrase the question another way?

sure, go ahead! hehe

Originally posted by darkhorse
p.s. by the way alpha, how does a trader shoot himself with a rope?

had to laugh at that..

 

[Traden4Alpha]

Originally posted by Gordon Gekko

Traden4Alpha,

i agree, what jperl asked is a good question. my question to you is, yes or no, in regards to what jperl asked? hehe

<b>Answer 1: Strong yes: </b> At first glance, it sure seems like these stocks have nonrandomly large numbers of downdays (compared to a naive expectation of 50:50 distribution of up:down days). Lets just look at the most extreme example: the case of LRCX and its extremely low 34 upday statistic. Assuming a binomially distribution on up/down days (i.e., the coin toss model), we can expect to see a run with 34 or fewer updays (or heads) out of 100 days(or tosses) about 0.12% of the time. Boy'o'boy, that sure seems unlikely! Must be nonrandom!

<b>Answer 2: Maybe yes: </b> At second glance, the story is less compelling when we check for survivorship bias. The chance that any stock (any coin) in a set of 100 stocks (coins) having a run with 34 or fewer updays (or heads) out of 100 days(or tosses) is 11.3%. OK, now the hypothesis that some stocks have nonrandomly high numbers of downdays seems a little shakier.

<b>Answer 3: Who knows: </b> But the real answer is that this analysis is not adequate for making a trading decision. (in fact the analysis is inadequate on two levels). First, the analysis tells us little about whether the upday/downday ratio for a set of 100 days is predictive of the direction of movement on subsequent days. At best, we can conclude that some stocks seem to have excessively large numbers of downdays. Whether that high frequency of downdays persists into the future (or is "stationary" in statistical parlance) is unknown. Second, the analysis tells nothing of the magnitude of the up and down movements. Although it is unlikely, in my experience, it is possible for a stock to be moving upward in price while having predominately downdays (i.e., large numbers of small down movements offset by occasion upsurges). Without a more careful analysis of how prior frequency of downdays correlates with future direction AND MAGNITUDE of downdays, we cannot really say how to trade these stocks based on this data.


Note that all of this ignores darkhorse's issue of how this type of data interacts with other data used by either discretionary or mechanical traders. This brings us into the realm of conditional probabilities and more complex decision rules. These have their own set of problems: the more variables you look at, the greater the number of contingencies, the rarer each contingency, the harder it is to collect enough data to test for valid patterns in all combinations of variables. Brings us back to "keep it simple."


<b>Re: how does a trader shoot himself with a rope? </b> Hmmm....Maybe if a trader tries to use a trend line as a life line, gets roped into an overleveraged position and blows up their account?!?!?!?!??? I got the phrase from the title of a book on C programming that detailed the rather numerous ways to create buggy software with that programming language. I like the phrase as it really epitomizes the mental/logical/mathematical/statistical errors that people make (myself included) in both creating and executing trading systems.

Cheers all,
-Traden4Alpha

 

[Gordon Gekko]

Originally posted by Traden4Alpha

<b>Answer 1: Strong yes: </b> At first glance, it sure seems like these stocks have nonrandomly large numbers of downdays (compared to a naive expectation of 50:50 distribution of up:down days). Lets just look at the most extreme example: the case of LRCX and its extremely low 34 upday statistic. Assuming a binomially distribution on up/down days (i.e., the coin toss model), we can expect to see a run with 34 or fewer updays (or heads) out of 100 days(or tosses) about 0.12% of the time. Boy'o'boy, that sure seems unlikely! Must be nonrandom!

<b>Answer 2: Maybe yes: </b> At second glance, the story is less compelling when we check for survivorship bias. The chance that any stock (any coin) in a set of 100 stocks (coins) having a run with 34 or fewer updays (or heads) out of 100 days(or tosses) is 11.3%. OK, now the hypothesis that some stocks have nonrandomly high numbers of downdays seems a little shakier.

<b>Answer 3: Who knows: </b> But the real answer is that this analysis is not adequate for making a trading decision. (in fact the analysis is inadequate on two levels). First, the analysis tells us little about whether the upday/downday ratio for a set of 100 days is predictive of the direction of movement on subsequent days. At best, we can conclude that some stocks seem to have excessively large numbers of downdays. Whether that high frequency of downdays persists into the future (or is "stationary" in statistical parlance) is unknown. Second, the analysis tells nothing of the magnitude of the up and down movements. Although it is unlikely, in my experience, it is possible for a stock to be moving upward in price while having predominately downdays (i.e., large numbers of small down movements offset by occasion upsurges). Without a more careful analysis of how prior frequency of downdays correlates with future direction AND MAGNITUDE of downdays, we cannot really say how to trade these stocks based on this data.


Note that all of this ignores darkhorse's issue of how this type of data interacts with other data used by either discretionary or mechanical traders. This brings us into the realm of conditional probabilities and more complex decision rules. These have their own set of problems: the more variables you look at, the greater the number of contingencies, the rarer each contingency, the harder it is to collect enough data to test for valid patterns in all combinations of variables. Brings us back to "keep it simple."

this makes me ask myself..ok so, that said, what do we actually do then? how can we "keep it simple?" this may be similar to darkhorse's question of "Better to phrase the question another way?"

all in all, i'm still confused when it comes to entries..do they matter or not? my gut says they probably do in some way..but i don't know how to find a "good" one without getting complicated. everyone says to "keep it simple," but quite frankly, i don't know how! lol hey, at least i can admit i have a problem..that's the first step!

 

[darkhorse]

Originally posted by Gordon Gekko

everyone says to "keep it simple," but quite frankly, i don't know how! lol hey, at least i can admit i have a problem..that's the first step!

Full circle again.

"Keep it simple" should be taken as keep it simple for YOU, i.e. make sure you are comfortable with your methods and understand why and how they should work.

Some folks don't like uncertainty and fuzziness, so they develop hard and fast rules and become mechanical traders implementing a step by step concrete process. Their weakness is an inability to deal with lots of information and the emotional stress of decisions on the fly- so they use computer generated signals instead of intuitive analysis. Their strength is the rigid discipline to always follow a clear cut plan and not to second guess.

Other folks don't like super restrictive rules and prefer to have logical motives behind their actions, i.e. they need to understand the 'why' as well as the 'how' and so they become discretionary traders. Their strength is the ability to handle uncertainty, process information flows quickly, and make intelligent decisions with confidence. Their weakness is that the fuzziness of their methods can work against them if they are not highly disciplined, and they can have trouble sticking with a trade unless they have strong analysis based conviction.

Which suits you better, mechanical or discretionary? All traders are BOTH to some degree- systems traders have a discretionary element to their day to day decisions, and discretionary traders may have mechanical rules in terms of position sizing, stop placement, entry and exit patterns etc. But in all likelihood, YOUR personality will push YOU in one direction or the other.

So your question should be, do I prefer the comforting concreteness of hard and fast rules from beginning to end, or would I rather embrace the freedom of fuzziness and try to 'figure things out' instead?

If you are more systems based in your personality, go do research on basic trading systems, entry signals, exit signals, etc. Buy a book like 'Trading Systems and Methods' by Perry Kaufman. In fact you probably want to buy a book like that anyway, less than the cost of a single trade and will answer questions you haven't even thought of yet.

If you are more discretionary based in your personality, learn to base your entries on information flow. Figure out the technical patterns you like to make use of, the way you like to enter trades-and then embrace the world of information to determine when to enter (and when to lay chilly). There is tons of free macro analysis on the web (most everyone gives away the big picture analysis because it is the specific recommendations people pay for). Start reading, start paying attention to stuff, develop a thesis on why markets move and where they might move and what affects what, etc. etc.

Big picture is that discretionary trading is tougher and more demanding than systems based but that's also why the biggest traders in the world are discretionary. I started out almost entirely mechanical and moved towards discretionary later as my style evolved.

 

[Gordon Gekko]

Originally posted by darkhorse
Full circle again.

"Keep it simple" should be taken as keep it simple for YOU, i.e. make sure you are comfortable with your methods and understand why and how they should work.

Some folks don't like uncertainty and fuzziness, so they develop hard and fast rules and become mechanical traders implementing a step by step concrete process. Their weakness is an inability to deal with lots of information and the emotional stress of decisions on the fly- so they use computer generated signals instead of intuitive analysis. Their strength is the rigid discipline to always follow a clear cut plan and not to second guess.

Other folks don't like super restrictive rules and prefer to have logical motives behind their actions, i.e. they need to understand the 'why' as well as the 'how' and so they become discretionary traders. Their strength is the ability to handle uncertainty, process information flows quickly, and make intelligent decisions with confidence. Their weakness is that the fuzziness of their methods can work against them if they are not highly disciplined, and they can have trouble sticking with a trade unless they have strong analysis based conviction.

Which suits you better, mechanical or discretionary? All traders are BOTH to some degree- systems traders have a discretionary element to their day to day decisions, and discretionary traders may have mechanical rules in terms of position sizing, stop placement, entry and exit patterns etc. But in all likelihood, YOUR personality will push YOU in one direction or the other.

So your question should be, do I prefer the comforting concreteness of hard and fast rules from beginning to end, or would I rather embrace the freedom of fuzziness and try to 'figure things out' instead?

If you are more systems based in your personality, go do research on basic trading systems, entry signals, exit signals, etc. Buy a book like 'Trading Systems and Methods' by Perry Kaufman. In fact you probably want to buy a book like that anyway, less than the cost of a single trade and will answer questions you haven't even thought of yet.

If you are more discretionary based in your personality, learn to base your entries on information flow. Figure out the technical patterns you like to make use of, the way you like to enter trades-and then embrace the world of information to determine when to enter (and when to lay chilly). There is tons of free macro analysis on the web (most everyone gives away the big picture analysis because it is the specific recommendations people pay for). Start reading, start paying attention to stuff, develop a thesis on why markets move and where they might move and what affects what, etc. etc.?

damn, i was leaning towards, "hard and fast rules from beginning to end," til i read...

Originally posted by darkhorse
Big picture is that discretionary trading is tougher and more demanding than systems based but that's also why the biggest traders in the world are discretionary. I started out almost entirely mechanical and moved towards discretionary later as my style evolved.

...but like you said, i may evolve into a discretionary trader.

p.s. once again darkhorse makes a great post!

 

[Kymar]

...how many times have I found myself in a poorly calculated trade... short above support... long below resistance... and wished that, if ever, maybe just this once, price movement could be random, if only to give me at least a 50/50 chance of getting out unhurt...

 

[dotslashfuture]

i think everybody needs at least one or two hard rules that they never break, otherwise there will be no consistency in their results. Basically the trader has to artificially impose some kind of order onto their trading or else they are pure gamblers and they would be better off just putting the money in their mattress.

 

[Traden4Alpha]

I like darkhorse's dichotomy of mechanical vs. discretionary traders (I'm in the mechanical camp myself) -- every trader needs to find approach that they feel comfortable with.

My concern is that there are valid reasons for limiting the complexity of the trading approach (regardless of whether one is "comfortable" with a higher level of complexity). With too many variables, its just too easy to get lost in all the myriad combinations of variables be up, down, at key levels, etc. With too many parameters, its just too easy to overfit the system to prior data. With screening too many different stocks or markets, it just too easy to find something just just happens to seem to trade well with you system.

If Einstein were a trader, he'd probably say:
<b>A trading system should be as simple as possible, but no simpler!</b>

Of course, Einstein would probably go on to blow-up his account. After this he would say "Although, god does not play dice with the universe, but he does play dice with the markets."

I wish I had the rules for the optimal complexity of a trading system -- guess this is something I really need to work on for myself.

Time to go think some more (or am I thinking too much?)
-Traden4Alpha

 

[darkhorse]

Originally posted by dotslashfuture
i think everybody needs at least one or two hard rules that they never break, otherwise there will be no consistency in their results. Basically the trader has to artificially impose some kind of order onto their trading or else they are pure gamblers and they would be better off just putting the money in their mattress.

very true. no successful trader is 'purely' discretionary, some basic rules have to by rhythmic or systematic even if never written down (and discretionary trading is ultimately systematic anyway because our minds are arguably systematic).

and on the other side, because even the best systems occasionally throw out signals that no trader in their right mind would take due to extenuating circumstances, i highly doubt there are any 'pure' mechanical traders either (thus implemention of a mechanical system is ultimately discretionary).