No registration required! (Why?)

The Gambler's Fallacy


Has anybody applied The Gambler's Fallacy to trading?

Briefly what this says is "The gambler's fallacy is a logical fallacy involving the mistaken belief that past events will affect future events when dealing with random activities..." etc.

The idea is that the dice or roulette wheel does not have a memory. So if red came up 10 times in a row on the roulette table then the chance of it coming up red the next time is not affected by the previous results.

I am interested to know how this applies to trading.
It is a good book to read. It is easy and fast and if nothing else gives you a good sense of history from that period. It is obviously also a good analogy to many psychological aspects of trading.
I just read a couple of pages. It looks very interesting!!
quote:
Originally posted by elite trader

Has anybody applied The Gambler's Fallacy to trading?

Briefly what this says is "The gambler's fallacy is a logical fallacy involving the mistaken belief that past events will affect future events when dealing with random activities..." etc.

The idea is that the dice or roulette wheel does not have a memory. So if red came up 10 times in a row on the roulette table then the chance of it coming up red the next time is not affected by the previous results.

I am interested to know how this applies to trading.



What a great question...


As a ten year plus veteran in statistical analysis, it's mind numbing how many times the word "random" is applied to philosophies and even scientific approaches. Yet, to date, nobody has ever proved that there is truly any such thing as "random", nor I think will they.

Just as Fibonacci's observation of primes, (we know as Fib series), is not an "answer" to trading, it does answer many things, one of which is that the markets are anything but random.

But then, Fib primes are a "Continuum" in all things, not a tool first off. Pick any distance, any ruler, any starting point and anything else a part of it will have a fib relationship, to the original reference, etc.

Just as the intersections and angles of our veins and arteries share the same mathematical relationship with tree branches and roots, nothing in nature will have a truly random domain or event.

Likewise, anything undergoing change, will have some measure of underlying order that it's variations represent. Not all things will let us measure and predict the next outcome perfectly, but certainly over a span of time and events, models of reliability emerge and where they can be measured, they will have some measure of predictability.

As for the gambler's fallacy, more times than not, emotions and greed over-ride logic and a failure to maintain a reasonable defense becomes an undoing regardless of how successful an offense may "seem" at first. Trading is not so far off from the same, as many seasoned traders have tried to impart, over and over again, to emphasize loss protection over that of profit seeking.

Defense does not win a game, but protects one from the extent of loss in order to establish an offense that can win some portion of a game. If the offense is adequate beyond the cost of defense, some measure of bias wins out. The market is generally impartial and therefore less akin to a sports team and more likely to behave methodically, unless a substantial public event impacts it heavily and pointedly, (anomaly). The market doesn't "gear up" or "answer back", like an opponent in gaming, but follows other more natural analogs.

A roulette wheel has bearings, protrusions, as does the ball suffer imbalance and motions which repeat their cycles of variation and a magnitude of repetition effects it in the end. So, mechanical examples make poor subjects and I have made decent money on roulette red/black bias. You get invited to leave, for using it, if successful.

But... take 100 people in a room and ask them to write down a number from 1 to 5 privately and what happens? Invariably, (I have done this many times), the number 3 is the most frequently chosen and 1 & 5 the least chosen.

Why? What caused the most people to choose 3 every time? Is it a positive influence directing them, or a negative influence? How does that apply to the markets?

Most suggest that we tend to avoid adversity as a function of fear. Most seek the center as the least extreme, the most comfortable to feel that we chose "reasonably".

We also tend to believe what we see with our eyes and when we first see such discrete and robust angles and patterns in the market, (appearing to be two dimensional), the mind tells us, "How hard can it be to grab 30% of these and put that little bit in my pocket?" About as hard as predicting where the next branch will emerge on a growing tree limb, with 30% accuracy to all the other branch locations on the tree; not so two dimensional, after all.

But, if you spill a hundred like balls down a slide with 100 like pins protruding evenly spaced, there are still far more balls that collect to the center at the bottom, despite hitting each other and the pins on the way down, (or because of). They form a "central distribution" or "Bell Curve", given that all other things are relatively balanced. They're not subject to emotion, yet they still behave with the same natural tendency.

One of the things that attracted me to mypivots.com, was it's focus on pivot points and some of daytrading's homework on pivot point analysis. Surprisingly, there are very few decent web locations that truly delve into how pivot points, (S/R's), relate to the market, market maker's application and how the herd mentality and bull/bear pressures play a role in creating those variations we know as "the market".

Gambler's philosophy is not a fallacy imo, but how well one does their homework, applies patience, perseverance and is steadfast on loss protection will ultimately determine the difference between if one becomes a trader, or a gambler, or how much of either we become as experience tunes us. I'm fine with 50/50 odds, as long as the size of the wins outweigh the size of the losses.

Apply the human intuitive to several reliable ES Gartleys making consistent Tea Cups, find the back-side, two-stage correction at .312, exceeding .618 and revisiting .500 down the back of the main Gartley and with enough space between pivots having been hit reliably, one can very easily apply a 1:5 risk reward ratio and put 4-5 points in the bag or cover the loss + commissions quite safely as the main Gartley completes a 1.0 retracement. But, blindly try to guess your way through 10 moves going full pivot-to-pivot with your thumb on the button and you'll lose more than you take home every time, because you didn't or wouldn't apply all the perception available intuitively, to choose reliably and seek every advantage. You gambled, rather than determining the reliable components and covering the loss first.

Knowing what distractions and attractions fail gamblers, will likely remain the best lessons that traders can take away from the industry of gambling. For sure, we'll never succeed in "Bluffing" the market into folding the pot to us.

Checkers anyone?
 




quote:
Originally posted by urbansound

Just as the intersections and angles of our veins and arteries share the same mathematical relationship with tree branches and roots, nothing in nature will have a truly random domain or event.

So a human, by being a carbon based organic life form, will not be random. That, and its extrapolation of humans as a group, sounds reasonable.
quote:
...but certainly over a span of time and events, models of reliability emerge and where they can be measured, they will have some measure of predictability.

There is the question that when something becomes even slightly predictable that people will start to front-run it to "get in early" and that will halt the move to the "predicted" resistance/support area. We were discussing that in Evidence-Based Technical Analysis
quote:
...Trading is not so far off from the same, as many seasoned traders have tried to impart, over and over again, to emphasize loss protection over that of profit seeking.

I completely agree. Most traders approach trading as a "profit making" endeavour and not a "loss limiting" exercise. The natural genetically wired behavior is to see profits and not to limit losses.
quote:
Defense does not win a game, but protects one from the extent of loss in order to establish an offense that can win some portion of a game. If the offense is adequate beyond the cost of defense, some measure of bias wins out.

Nice way to word the probability paradigm.
quote:
The market doesn't "gear up" or "answer back", like an opponent in gaming, but follows other more natural analogs.

So in gaming you can push your opponent into a "more" predicable pattern by doing something to them. If you physically hurt them for example you "know" that they are going to focus more energy on revenge and that type of behavior is more predicable; their strategy or movements for revenge. As you say, the market cannot be psychologically manipulated like that.
quote:
So, mechanical examples make poor subjects and I have made decent money on roulette red/black bias. You get invited to leave, for using it, if successful.

Were you so successful that you were invited to leave? If so, and if I may ask, when and where?

Also, I'm curious about your performance once you had discovered a mechanical imbalance in the roulette wheel. The imbalance must have caused either red or black to come up more than 50% of the time in order to give you the edge. After that I am guessing that you consistently bet the same amount on that color over a period of time. So the type of performance that I am interested in is: (1) how much did you risk (2) over what period of time (3) for what return? And from that did you ever calculate and hourly salary that you gained from the exercise?
quote:
We also tend to believe what we see with our eyes and when we first see such discrete and robust angles and patterns in the market, (appearing to be two dimensional), the mind tells us, "How hard can it be to grab 30% of these and put that little bit in my pocket?" About as hard as predicting where the next branch will emerge on a growing tree limb, with 30% accuracy to all the other branch locations on the tree; not so two dimensional, after all.

I have often wondered if an autistic person gifted with pattern/number recognition such as Dustin Hoffman played in Rainman could be tested using a high speed playback of daily data from the market. You would stop the market at random points and ask the subject which way the pattern is going to go and then measure the success rate. I am guessing that the success rate would be random 50/50 at the beginning but over time I am thinking that they would develop a higher probability selection skill.
quote:
One of the things that attracted me to mypivots.com, was it's focus on pivot points and some of daytrading's homework on pivot point analysis.

I wish that I had more time to analyze more of the umpteen million theories and possible formulae that are out there.
quote:
Surprisingly, there are very few decent web locations that truly delve into how pivot points, (S/R's), relate to the market, market maker's application and how the herd mentality and bull/bear pressures play a role in creating those variations we know as "the market".

I think that most trading web sites are geared towards the business aspect of making money out of traders as a service. This is not unreasonable because they existed before the Internet as newsletters. MyPivots.com started as an accident: I was helping a friend calculate some pivot points and each day I would email him a spreadsheet with the pivot points. He was having problems receiving the emails and so I put the spreadsheet on a web page so that he could get to the numbers. He told some friends and so on and that's how it got started, on trader helping another and not as a business. That is probably why the information here is on average more useful to the average trader than most other trading orientated web sites.
quote:
I'm fine with 50/50 odds, as long as the size of the wins outweigh the size of the losses.

This is often such a difficult thing to explain to other traders. You can have 30% win ratio and yet still be highly profitable if those 30% get you 10 points per win and the 70% lose you 0.5 points per loss. This is witnessed by ads for trading systems that boast 90% win ratios.
quote:
Apply the human intuitive to several reliable ES Gartleys...

What is a Gartley?
quote:
...we'll never succeed in "Bluffing" the market into folding the pot to us.

I enjoy reading your posts urbansound - you always have some tidbit of information that I can take away from it - something that I didn't know before. It also opens up the mind to interesting possibilities and stimulates it in different directions.

From your chat about defense and offense have you ever read The Art of War?
quote:
Originally posted by day trading
[quote:...but certainly over a span of time and events, models of reliability emerge and where they can be measured, they will have some measure of predictability.]



There is the question that when something becomes even slightly predictable that people will start to front-run it to "get in early" and that will halt the move to the "predicted" resistance/support area. We were discussing that in Evidence-Based Technical Analysis


(Hi again :) It's important to keep scope with the fact that if front-running could impede the natural order of pivots, the market wouldn't have direction any longer.

We tend to consider Fibonacci as a tool, where it's actually a continuum in all things. Since the market moves in relation with Fib order and pivots are calculated from ratios of prior market moves, the pivots inherently are also a direct subset of prior Fib order. It quickly becomes paradoxic.

(A little OT from gambler's fallacy but...) I'm presently designing a program that filters market maker moves away from the rest of the herd, in order to give a visual aspect of where, when and to what degree market makers are influencing the market, versus herd. The results are still raw yet, but a couple of cool things show up.

1) Sometimes even market makers get caught up in herd action.
2) Market Makers tend to "spoon feed" and "carrot dangle" the herd after an initial boost of volume off the pivot line. By spoon feeding between pivots they expose the least amount of non-profitable contracts to keep the herd stimulated.
3) Sometimes the market loses steam, just shy of pivot threshold and in comes a few more spoonfuls.
4) When pivot begins to become realized, MM stop and/or limit orders flood in like cattle running from the great Chicago fire.

I monitor tick data to extract the differences, but you can see it to a pretty fair degree just with a CCI tool also, since the CCI normalizes the moving average as a straight zero line.

quote:

[quote:So, mechanical examples make poor subjects and I have made decent money on roulette red/black bias. You get invited to leave, for using it, if successful.]


Were you so successful that you were invited to leave? If so, and if I may ask, when and where?

Also, I'm curious about your performance once you had discovered a mechanical imbalance in the roulette wheel. The imbalance must have caused either red or black to come up more than 50% of the time in order to give you the edge. After that I am guessing that you consistently bet the same amount on that color over a period of time. So the type of performance that I am interested in is: (1) how much did you risk (2) over what period of time (3) for what return? And from that did you ever calculate and hourly salary that you gained from the exercise?


I was not invited to leave and it was an impulsive test I decided to try while vacationing in the central Michigan area. I chose a double down approach to each bet I placed which signaled to the pits that eventually I would balance at zero and self limit. For that reason they allowed me to continue, I believe, but it was clear they were watching with interest as well.

I started with $20.00 and at one point inside of about 35 minutes or so, was up to $700 and change. I was eventually taken out by nine blacks in a row at about 2 hours in.

BTW I didn't look for an imbalanced wheel. All things suffer imperfection which will cause variation that will give way to order inside what "seems" to be random. I simply timed my entry as Red, after I would see four black in a row show up and stay doubling red until my double hit red. Then I would lay back and watch for the next series of same color events and then place in the opposite color, doubling again.

Jumping in and out of the table is also discouraged by pit bosses, but since I was playing on four tables all at the same time, they weren't perceiving the manner by which I was gaging when I would put up heavier money, versus laying back. By working multiple tables they saw me as contributing to them 4 fold, when actually I was keeping them from comparing notes. ;-) Again, I didn't stay long, made a social event of it for fun and clearly did not pose a risk to them.

I have actually seen 11 blacks in a row, on another occasion.

quote:

I have often wondered if an autistic person gifted with pattern/number recognition such as Dustin Hoffman played in Rainman could be tested using a high speed playback of daily data from the market.

Savant behaviour is truly an interesting phenomena. Then again, we have yet to invent a computer that can hold a candle to the terabyte/second processing rate that simply typical humans possess.

Still, we can't feed our poor, but we can plan for Mars. Go figure.


quote:
What is a Gartley?


[url]http://www.harmonictrader.com/gcontroversy.htm[/url]
[url]http://www.harmonictrader.com/price_patterns.htm[/url]

Notice also that the typical T-cup and handle is formed by a series of major/minor Gartley combination, or minor followed by major. The typical head and shoulders is usually a minor/major/minor Gartley or one of it's derivatives.

All Garts and derivatives will have some explicit expression of a very solid fib ratio between two or more parts. When those can be foreseen to land on or very near pivots, it supports probability of a retrace beyond 1.0 and can give nice confidence to seeing a 1.272 or 1.618 continuation or reversal/continuation; a nice place to drop half and ride the remainder to a discretionary exit.



quote:
Originally posted by day trading

I enjoy reading your posts urbansound - you always have some tidbit of information that I can take away from it - something that I didn't know before. It also opens up the mind to interesting possibilities and stimulates it in different directions.

From your chat about defense and offense have you ever read The Art of War?



Thank you for the kind words. You have many solid contributors that have gathered around the mypivots forums and some nice contributions show up in many places here. Sharing information is what it's about ;-)

I have not read The Art of War and in fact rarely read books at all. I'm typically 16 hours a day either writing API interfaces, programs or trying to find why and how bear/bull relationships stretch the market the way they do.

I do fancy watching some football on occasions and oddly one of my reasons is to take note of commentators prognostications pre-game and then listen to how their opinions change as the game progresses.
quote:
Originally posted by elite trader

Okay, here is another probability question:

Suppose you have 3 doors and behind one of those is a hidden treasure.

You pick door A and someone opens door B for you and reveals that there is no treasure behind door B. You are now given a chance to change your selection before the final door opening. Do you stick with your original choice (door A) or do you switch to door C?

What is your decision? And why did you decide that?


I recently had to explain this Monty Hall problem to someone and I came up with what I thought was a better explanation that ones that I've seen before.

Imagine that you play this game 9 times. If you stick with your original choice you would expect to win 3 of the games because you have a one in three chance.

For the 6 games that you lost, the prize will be behind the unopened door that you didn't switch to. i.e. had you switched doors you would have won 6 times for the 9 games that you played.

Ergo, switching doors improves your probability of winning to 66.6% and staying with the original door gives you a 33.3% chance of winning.

Now try the same logic with 99 or 999 or 9,999 games. Each game represents a trade. After a trade is in place more information is introduced to the problem by means of more price and volume information from the rest of the market.
quote:
Originally posted by day trading

[quote]Originally posted by elite trader

Okay, here is another probability question:

Suppose you have 3 doors and behind one of those is a hidden treasure.
[...]

I recently had to explain this Monty Hall problem to someone and I came up with what I thought was a better explanation that ones that I've seen before.
[...]



Except for the human bias of the "yabut" factor...

Yeah, but... It is a statistically proven fact that humans when given the choice will always pick the "safest" number within the realm of perception. In most cases the number 2 would always be the most common chosen value as it lies "centrally" between 1 and 3, (feels the safest).

Given the next safest opportunity and having already chosen door 1, most will A) resist change due to fear and B) 1 is considered a "safer" number compared to 3 as it is less significant in comparison.

Now, if we shut Monty Hall's expensive suit coat in door number 2 and then simply take what's behind all three doors, (even Carol), then as long as we can run faster than those who would stop us, our chances of obtaining the greatest yield improve even more.

(I think I just dated myself with that Carol thing... Oh boy)
quote:
Originally posted by urbansound
...statistically proven fact that humans when given the choice will always pick the "safest" number within the realm of perception. In most cases the number 2 would always be the most common chosen value as it lies "centrally" between 1 and 3, (feels the safest)...

So, as a day trader, are we trading against predictable humans that will be drawn to round numbers and other "predictable" areas on a chart? Or are we trading against machines that are void of emotion?

Are there enough humans trading in the market place to cause the machines to fail often enough such that it's more likely that the human predictability will prevail.

I enjoyed your other comments!
UrbanSound, you and I are saying the same thing. Simply using price bars you cannot predict the next bar. The predictability of future bars is in the patterns. Such patterns include S/R, Fibs and such. It is the formation that gives predictive value because from that you are able to see what other traders are thinking and hoping for.