Trading Lessons/Insights From Coin Flipping

[Nine_Ender]

Quote from Mike805:

This response is indicative of your inherent bias in the way you view markets, trading, investing etc.

The theory presented here is very sound. If you had read the reference materials/threads, you'd be rethinking your own assumptions.

We are not talking about personal biases here nor are we talking about some fundamental reason for price movements. We are talking about random behavior.

The context of this thread is the mathemetical representation of a random stochiastic process, and, it's potential insights in the realm of trading. This is not some grand unified theory of what "stocks really are", it is a glimpse into what statistics can offer.

In fact, you've missed the point so badly that perhaps you're not alone, so I'll make it very clear; when you don't understand randomness, you will never understand what is not random. It is the correlation versus causation fallacy in a nutshell.

Case in point, how do you know the fundamental driver in the BVF example is statistically significant? Which elements of that price move repeat and what are the causes? How do you segregate potential causes into random versus non-random?

Again, claiming "just look at the chart" is not evidence that you're looking at a non-random process (although in BVF's case this might be true). I can easily run a random time series process a few million times and from that set of runs you will see a couple of trends (i.e. random artificial price movements) that look just like BVF's "run"....

From your post it is clear that it is you that doesn't understand random behaviour and the onus of proof on theory.

Your claim that you could find a random trend that looks like BVF is total crap. You can't. Period. You are suggesting that a stock could go up 85% in a few months due to randomness in trading. Well if you truly believe that, I suggest you buy some housing in Florida and hope that the randomness of housing prices makes you 85%.

To those with their feet firmly in the ground, we all know BVF went up primarily for fundamental reasons, and housing prices went down in Florida for fundamental reasons.

Obviously you don't have that firm a grip. Maybe just flip a coin on that housing decision.

 

[Mike805]

Quote from Nine_Ender:

Your claim that you could find a random trend that looks like BVF is total crap. You can't. Period. You are suggesting that a stock could go up 85% in a few months due to randomness in trading. Well if you truly believe that, I suggest you buy some housing in Florida and hope that the randomness of housing prices makes you 85%.

To those with their feet firmly in the ground, we all know BVF went up primarily for fundamental reasons, and housing prices went down in Florida for fundamental reasons.

You didn't read my post very carefully. I'm disappointed...

Many here have posted examples of random trends that look very similar to many stock market moves. There is plenty of literature out there that shows "trending" examples with simple random walks. That said I do not dismiss the catalyst theory and have several models that exploit such price moves.

My suggestion was that a stock could go up 85% for any number of reasons. Most of which might as well fall into the correlation versus causation fallacy. I asked you to delve into the differences...

I never stated my belief system, I'm stating a grey area that you seem to think is black and white.

Be careful with having your feet "firmly on the ground", its hard to move out of the way of a steam-roller when one is convinced that it will turn due to obvious fundamental reasons...

 

[Nine_Ender]

Quote from Mike805:

You didn't read my post very carefully. I'm disappointed...

Many here have posted examples of random trends that look very similar to many stock market moves. There is plenty of literature out there that shows "trending" examples with simple random walks. That said I do not dismiss the catalyst theory and have several models that exploit such price moves.

My suggestion was that a stock could go up 85% for any number of reasons. Most of which might as well fall into the correlation versus causation fallacy. I asked you to delve into the differences...

I never stated my belief system, I'm stating a grey area that you seem to think is black and white.

Be careful with having your feet "firmly on the ground", its hard to move out of the way of a steam-roller when one is convinced that it will turn due to obvious fundamental reasons...

I see very little relevance to "coin flipping" theory to trading.
Please post some real time trades where "coin flipping" theory
was an important consideration. I'm not interested in unproven theory or academic arguments going nowhere. The onus of proof rests with you on this.

With respect to the "steam roller", I trade trends and the "steam roller" is my friend.

 

[masterjaz_99]

Hi Mike and Others,
A simple comparison, BVF vs. an excel spreadsheet. 50-50 probability with +1 for win, -1 for loss. Sum them up, plot it up, guess what you get...ANYTHING YOU WANT, with enough time. The entire point of the this thread is the existence of a distribution of results based on a coin flip. The results can be good, bad, awesome, crap, or flat. Hit F9 (recalculate command in excel) enough times and you'll see results that will blow you mind or make you want to blow your mind away. That is the point, you have to see 'through' the noise and probabalistic results to have a true edge in the market.

Have fun playing,
Masterjaz

 

[masterjaz_99]

Random results

Cannot upload spreadsheet, but here's all you need. In 1 cell (A1) use formula =rand()
in next column (use B1) =if(A1>=0.5,1,-1)
in next column (use C1) =B1, then C2 on use =B#+C(#-1)

Much harder to write than create. I'll try reposting later. Elitetrader giving me a 'database error'

 

[masterjaz_99]

Actually should be .xlsx

 

[nLepwa]

Quote from masterjaz_99:

That is the point, you have to see 'through' the noise and probabalistic results to have a true edge in the market.

Have fun playing,
Masterjaz

Or you can trade the noise.

Ninna

 

[masterjaz_99]

Quote from nLepwa:

Or you can trade the noise.

Ninna

Agreed, that's the way I've been leaning as of late...due exactly to this topic and these results...we'll see if it works.

Masterjaz

 

[intradaybill]

Why are you wasting you time with coin tossing? This has nothing to do with the market. In probability theory, the outcome of a coin toss is not depended at all on any other toss, previous or future. In the market, each price move depends to a lesser or larger extend on the previous move. This is obvious since

P(i+1) = P(i) + dP(i)

The change in price dp has a deterministic and a stochastic component as follows

dP(i) = D(i) + S(i)

S(i) depends on news, events and other random information.

D(i) depends on previous moves, it is a variable that can be forecasted easily.

So the answer is to trade when there is no random information coming in, or at least, when there is less expectation for any random information of great value to the market coming in. All you need is a model to forecast D(i). That is your edge.

Some people trade before or during unemployment, ppi, or GDP reports and then complain about randomness. Stupidity, there is plenty.

 

[masterjaz_99]

Quote from intradaybill:

Why are you wasting you time with coin tossing?

Because I previously posted that a year of real trading resulted in near perfect 50-50 coin flip results. Similarly, it can be shown that price moves (once accounted for various components i.e. volatility) may be modeled by a normal distribution.

I agree that individual ticks do move in a less probablistic fashion, the net move, may or may not be probablistic. Also note, I would say 'we' meaning me and possibly others, don't believe the market is random, it is probablistic, and that probablity includes the probability the distribution may change (skewness, kurtosis, distribution, etc.).

In the end, you model the market one way, we model it a different. Doesn't make one method right or wrong. I sure as heck don't want to be modeling (random) news, etc.

I don't know, just my thoughts