Sociopath Community

AliceInWonderland said: 

I've never come across this paper. 

It's among the original `robust` derivations, although it's not the first. It derives the random walk from the efficient market hypothesis (a few axioms).

In so far as an 'original derivation' goes I'm not sure there is one per say as it relates to Random Walk theory. 

Bacheilier's Theory of Speculation is the OG I guess while the real mathematical framework begins with Levy, Wiener, and Ito.

I'm not sure of any real conclusive combinations of these into singular ideas until the 60's when Fama and Cooter start to put out their Hypothesis. 

I guess Fama is closer to what I'm looking for.

I assume you are trying verify EMH? Proof of its validity? I don't think you can and in fact the opposite has already been done with Grossman and Stiglitz

No, I believe models are only accurate insofar as the underlying assumptions that go into building them are accurate. Even Samuelson in that paper is saying that people shouldn't read into his results too much, because his assumptions are not realistic. I've talked about my philosophy a bit, and I'd rather understand the fundamentals and what has been done to understand what will go wrong when the market defies any given hypothesis. So the next time a book tells me that the stock prices follow Brownian motion, I'll know what the context is.

Besides, implementing small corrections to an otherwise workable model is often better than starting from scratch. I'd rather think that the stock market is driven by supply and demand, and the price changes can have stochastic elements to them because the number of investors can vary stochastically, the information flow can't be predicted, and different people will evaluate the price of a stock differently due to variable reasons.

However, to say that the stock market follows a random walk without even attempting to understand why that should be the case is a sin.

If I understand correctly, the Grossman and Stiglitz model is essentially pointing out that the efficiency depends on the profitability of applying new information. So there should be a balance between the two. However, that isn't my main worry about the EMH/random walk hypothesis; my main issue with it is that the supply/demand itself has stochastic elements to them, people won't agree on what the fair price of a stock is, and how to apply any given information is largely subjective. What I do not understand is how much each element is expected to contribute.

In so far as proving the validity of random walk theory as an approach at all goes, join the club. 

This field is a mess. 

I feel like the starting point is off, and I don't believe the random walk hypothesis. I believe there are stochastic unpredictable elements to the stock market but I don't buy that the pricing itself is wholly random. I'm trying to see what I can get to with a supply/demand model, but there isn't too much about modeling that in the literature.

I tried making a quick model, and it gave OK results. However, I realized that the model is flawed in one fundamental way; I have to presume to know about the underlying distribution of stock investors (how they would bid/ask on the market), whereas I only can observe the apparent distribution of stock investors (the ones who in fact do make bids/asks). Relating the two has everything to do with the pricing model that I adopt, but I can't relate the two distributions without being able to read the mind of every stock investor (have the perfect pricing model). It's a pretty useless model if I need all the world's information to use it. So I'm trying to see if there are any cheats I can apply (poissonian process to model the inflow/outflow of investors or other tricks) which would allow me to relate the two without being a mind reader.

AppleGenius said: 

AliceInWonderland said: 

I've never come across this paper. 

It's among the original `robust` derivations, although it's not the first. It derives the random walk from the efficient market hypothesis (a few axioms).

I read the paper and I am not convinced ether. 

Seemingly we have a set of micro fluctuations being random which doesn't necessarily imply that the entire time-series is a random process. 

In so far as an 'original derivation' goes I'm not sure there is one per say as it relates to Random Walk theory. 

Bacheilier's Theory of Speculation is the OG I guess while the real mathematical framework begins with Levy, Wiener, and Ito.

I'm not sure of any real conclusive combinations of these into singular ideas until the 60's when Fama and Cooter start to put out their Hypothesis. 

I guess Fama is closer to what I'm looking for.

If you get a hold of his Phd thesis let me know, it's more robust than his general paper. 

The Axioms are there though. 

Essentially all the assumptions stem from the information utilized by the speculator and when it gets priced in. 

I assume you are trying verify EMH? Proof of its validity? I don't think you can and in fact the opposite has already been done with Grossman and Stiglitz

No, I believe models are only accurate insofar as the underlying assumptions that go into building them are accurate. Even Samuelson in that paper is saying that people shouldn't read into his results too much, because his assumptions are not realistic. I've talked about my philosophy a bit, and I'd rather understand the fundamentals and what has been done to understand what will go wrong when the market defies any given hypothesis. So the next time a book tells me that the stock prices follow Brownian motion, I'll know what the context is.

Besides, implementing small corrections to an otherwise workable model is often better than starting from scratch. I'd rather think that the stock market is driven by supply and demand, and the price changes can have stochastic elements to them because the number of investors can vary stochastically, the information flow can't be predicted, and different people will evaluate the price of a stock differently due to variable reasons.

However, to say that the stock market follows a random walk without even attempting to understand why that should be the case is a sin.

I agree with everything you've said here. 

If I understand correctly, the Grossman and Stiglitz model is essentially pointing out that the efficiency depends on the profitability of applying new information. So there should be a balance between the two. However, that isn't my main worry about the EMH/random walk hypothesis; my main issue with it is that the supply/demand itself has stochastic elements to them, people won't agree on what the fair price of a stock is, and how to apply any given information is largely subjective. What I do not understand is how much each element is expected to contribute.

Right, this is my misunderstanding as well. 

There is obvious random fluctuations but I have seen very little mention of how much of the system's fluctuations can be considered random. 

 

In so far as proving the validity of random walk theory as an approach at all goes, join the club. 

This field is a mess. 

I feel like the starting point is off, and I don't believe the random walk hypothesis. I believe there are stochastic unpredictable elements to the stock market but I don't buy that the pricing itself is wholly random. I'm trying to see what I can get to with a supply/demand model, but there isn't too much about modeling that in the literature.

Fundamentally what we want to test is the randomness of a nonlinear time-series. 

I have seen very little of this in financial literature but lots in general mathematical literature pertaining to chaotic systems. 

The issue seems to be that you can't without making assumptions about the structure and complexity of the dynamical system.

I tried making a quick model, and it gave OK results. However, I realized that the model is flawed in one fundamental way; I have to presume to know about the underlying distribution of stock investors (how they would bid/ask on the market), whereas I only can observe the apparent distribution of stock investors (the ones who in fact do make bids/asks). Relating the two has everything to do with the pricing model that I adopt, but I can't relate the two distributions without being able to read the mind of every stock investor (have the perfect pricing model). It's a pretty useless model if I need all the world's information to use it. So I'm trying to see if there are any cheats I can apply (poissonian process to model the inflow/outflow of investors or other tricks) which would allow me to relate the two without being a mind reader.

I will tell you that the majority of all stock investors in the retail market do not believe the markets to be random and this informs their behavior. Having said this the majority do not trade off of supply and demand either - value investing is simply not profitable right now and hasn't been for a decade at least. Everything is built on what is called momentum. 

This is a classic debate mind you, behavioral investment vs value investment. The debate is over the nature and roles of intrinsic value and psychology. 

Castle-In-The-Aire Theory vs Firm Foundation Theory

Just a quick response here, because I'm not sure if I'm expressing myself properly:

AliceInWonderland said: 

I will tell you that the majority of all stock investors in the retail market do not believe the markets to be random and this informs their behavior. Having said this the majority do not trade off of supply and demand either - value investing is simply not profitable right now and hasn't been for a decade at least. Everything is built on what is called momentum. 

Just to clarify, when I say supply and demand, I refer to the underlying process which drives momentum (and mean reversions). If you have more supply, the price will go down, irrespective of the timescales involved or the underlying motivation. I'm not proposing that supply and demand is what affects people's understanding of the fair price of a stock.

It will, however, drive the immediate drive of the stock. If people are bidding at 300 USD, and I think the fair price of the stock is 500 USD, I will still bid at ~300 USD (because I can). Understanding of the fair price can, in turn, drive the supply/demand. So I'm not saying it's the other way around -- that would be silly. Instead, I'm saying it seems like a more natural (fundamental) starting point than the stock price itself.

If you get a hold of his Phd thesis let me know, it's more robust than his general paper.

The Axioms are there though.

Essentially all the assumptions stem from the information utilized by the speculator and when it gets priced in.

Sure.

Fundamentally what we want to test is the randomness of a nonlinear time-series.

I have seen very little of this in financial literature but lots in general mathematical literature pertaining to chaotic systems.

The issue seems to be that you can't without making assumptions about the structure and complexity of the dynamical system.

Well, I'll let you know once I've solved that :P

Nah, but I believe there's some missing information which we won't have access to, which I still see as the main problem. Anyway, I'll need to get back to the drawing board and write this up, otherwise I'm just trying to explain vague ideas without rigor.

AppleGenius said: 

Just a quick response here, because I'm not sure if I'm expressing myself properly:

AliceInWonderland said: 

I will tell you that the majority of all stock investors in the retail market do not believe the markets to be random and this informs their behavior. Having said this the majority do not trade off of supply and demand either - value investing is simply not profitable right now and hasn't been for a decade at least. Everything is built on what is called momentum. 

Just to clarify, when I say supply and demand, I refer to the underlying process which drives momentum (and mean reversions). If you have more supply, the price will go down, irrespective of the timescales involved or the underlying motivation. I'm not proposing that supply and demand is what affects people's understanding of the fair price of a stock.

Right, so the ratio between buyer and seller volume. 

It will, however, drive the immediate drive of the stock. If people are bidding at 300 USD, and I think the fair price of the stock is 500 USD, I will still bid at ~300 USD (because I can). Understanding of the fair price can, in turn, drive the supply/demand. So I'm not saying it's the other way around -- that would be silly. Instead, I'm saying it seems like a more natural (fundamental) starting point than the stock price itself.

Okay I think I understand. 

So you want to find the relationship between bid/ask volume and value changes in the underlying? 

AliceInWonderland said: 

Okay I think I understand. 

So you want to find the relationship between bid/ask volume and value changes in the underlying? 

Yes, mostly; I'd rather talk about the bid-ask distribution as opposed to volume, but essentially yes.

Basically, given a bid/ask distribution, a fair price distribution, underlying number of bid/ask traders, and a pricing model, find the price distribution at time t+(delta t). My understanding is that the usual random walk theory proposes that the step is essentially determined by Brownian motion. However, I would be more comfortable seeing a similar result arise from those components that I mentioned (under some reasonable assumptions).

I've seen some discussion on related topics, but they've been pretty unsatisfactory. So, I decided to just try to look for an answer myself without having too many ideas distracting me. I also don't want to end up simply copy-catting others, so I'd rather get to my own result first and then compare and contrasts with others' ideas.

I cannot find myself accepting that the Efficient Market Hypothesis to be true given the markets do not seem to be a Martingale nor Markovian in nature.

Martingales are a type of stochastic process in which you will find a set of successive events in which the next event is conditional with only the latest event. If I have some stock price x_n+1 I want to know the value of and a set of other prices x1,x2,x3,…xn for which I do know the value, then x_n+1 is conditional only on x_n. I find this not to be the case given my experience with trading via mean regression which fundamentally relies on the market having memory and a future price depending on a set of past prices. The dependence is necessary because the strategy predicts a future price based on the average of a set of past prices and assumes a return to that average price. I have found this strategy to work remarkably well over the past few months though I am still in the infancy of making it mathematically rigorous, for now I am implementing the strategy and observing its success. To go beyond the antidotal it seems behavioral finance has made a lot of progress EMH as it has documented several phenomena such as the short-term reversal effect (which is similar to my ideas), long-term reversals, and long-horizon overaction.

As a final note, the market itself is made up of human actors who do the pricing. They cannot all be considered rational as not all can be considered irrational, but a vast majority of these individuals use past prices to predict future prices. As such one can suspect that the idea of self-fulfilling prophecy common in technical analysis can be considered as highly plausible especially given how common group think is in markets. If people are doing the pricing than people could be considered to be the market and as such it’s hard to consider these people as making decisions in a Markovian manner.

behavioral finance > technical analysis and fundamental analysis

MartinLooterKing said: 

behavioral finance > technical analysis and fundamental analysis

 That seems to be the case. 

 People tell markets how to move and markets tell people how to react. 

 Analytical Analysis through you Stochastics really has not changed conceptually since the 60's when it was first introduced. All the growth seems to be a side effect of new product offerings along with the efficiency of trading greatly increasing thanks to computers and algos.  

Alice taught me her trading method and this is what I pulled this morning. $992 in 45 minutes. I'm never wagecucking again!

PalePeach said: 

Alice taught me her trading method and this is what I pulled this morning. $992 in 45 minutes. I'm never wagecucking again!

 good job ya worthless nigger