Predicting randomness
- Thread starter oddiduro
- Start date
Hmmm, this could be a key post here. So, can the independent variable affect the dependent variable in a predictable way?
There are some professionals that do not think so. ( I of course, am a rank amatuer)
In theory, if you had very detailed knowledge of the entire financial sytem and you had excruciatingly detailed knowledge about how all traders will react to an event, then yes in theory it seems possible to do this....NUMERICALLY! (you cannot abtain a closed mathematical 'solution' to a many interacting body problem, which is what markets are with each participiant being an 'atom'). In physics we call this statisical mechanics.
The practical reality is that this is impossible to get such information, unless there are three people in a market perhaps.
You need to be careful though about the variables you select, the are not all connected causually and correlation is not causation.
I think the best people can do, (in terms of predicting) is 1) look for situations where it takes some time for all participants to respond. Then if you think you know what they will do (in the aggregate average sense) you can front run them, 2) work in markets that are 'news quiet' and keep your time horizon small, or 3) use a very coarse set of parameters (like buy and hold for 20 years)
This is getting harder with everyone on the net and computers doing trading. Outside of this I can't see just how the heck you can predict the externals like 9/11, or make (precise) long term predictions on a nonlinear system other than of course the basic ideas that economy grows overall historically- but then we are talking about just how precise a prediction we are trying to make as well.
Let me clarify the last statement. The weather is a nonlinear dynamical system but we still can predict that it will be colder in january than july. When we get to the finer level though, the two week forecast of rain, exact temp, ect. it isn't possible to go much beyond that. In other words we can predict averages but not the specifics. This is simply what I said above as well, if you coarsen your prediction then then it gets easier.
The analogy to the markets is not complete as the markets do not behave according to the laws of physics, the rules can seem to change over time. There may be average effects that have some predictability in the statisical sense but they will be nowhere near as predictable as the climate.
There are some professionals that do not think so. ( I of course, am a rank amatuer)
In theory, if you had very detailed knowledge of the entire financial sytem and you had excruciatingly detailed knowledge about how all traders will react to an event, then yes in theory it seems possible to do this....NUMERICALLY! (you cannot abtain a closed mathematical 'solution' to a many interacting body problem, which is what markets are with each participiant being an 'atom'). In physics we call this statisical mechanics.
The practical reality is that this is impossible to get such information, unless there are three people in a market perhaps.
You need to be careful though about the variables you select, the are not all connected causually and correlation is not causation.
I think the best people can do, (in terms of predicting) is 1) look for situations where it takes some time for all participants to respond. Then if you think you know what they will do (in the aggregate average sense) you can front run them, 2) work in markets that are 'news quiet' and keep your time horizon small, or 3) use a very coarse set of parameters (like buy and hold for 20 years)
This is getting harder with everyone on the net and computers doing trading. Outside of this I can't see just how the heck you can predict the externals like 9/11, or make (precise) long term predictions on a nonlinear system other than of course the basic ideas that economy grows overall historically- but then we are talking about just how precise a prediction we are trying to make as well.
Let me clarify the last statement. The weather is a nonlinear dynamical system but we still can predict that it will be colder in january than july. When we get to the finer level though, the two week forecast of rain, exact temp, ect. it isn't possible to go much beyond that. In other words we can predict averages but not the specifics. This is simply what I said above as well, if you coarsen your prediction then then it gets easier.
The analogy to the markets is not complete as the markets do not behave according to the laws of physics, the rules can seem to change over time. There may be average effects that have some predictability in the statisical sense but they will be nowhere near as predictable as the climate.
There is of course no harm whatsoever in theoretical debate on all the factors that may or may not subscribe to market conditions.
However where the underlying conundrum is to take the most money you can out of the market by trading, you need the answers to do so. The theoretical debate does not give precise answers and they are not the practical answers needed anyhow.
Problem solving will work at all levels although the harder the problem the greater the intelligence and knowledge required. As is a common feature in the independent trading fraternity, answers are proposed or tried without knowing what the questions are or by simply ignoring the need to to establish the questions.
Without the addressing the questions, the answers for example can be this or that system you borrow or invent and you hope it will work or half work.
However where the underlying conundrum is to take the most money you can out of the market by trading, you need the answers to do so. The theoretical debate does not give precise answers and they are not the practical answers needed anyhow.
Problem solving will work at all levels although the harder the problem the greater the intelligence and knowledge required. As is a common feature in the independent trading fraternity, answers are proposed or tried without knowing what the questions are or by simply ignoring the need to to establish the questions.
Without the addressing the questions, the answers for example can be this or that system you borrow or invent and you hope it will work or half work.
Cheese,
I both agree and disagree. Most people don't need theory to make a system, they can stumble onto it by pure empiricism.
I however have not been so lucky so I need to look deeper. I am more like Taleb in the article, I have seen my black swan.
You dismiss theory too quickly though, if you work hard enough you can translate it into trading systems. After all, this is what the PhDs do on wall street and they would not be around if it hasn't worked.
I both agree and disagree. Most people don't need theory to make a system, they can stumble onto it by pure empiricism.
I however have not been so lucky so I need to look deeper. I am more like Taleb in the article, I have seen my black swan.
You dismiss theory too quickly though, if you work hard enough you can translate it into trading systems. After all, this is what the PhDs do on wall street and they would not be around if it hasn't worked.
Perseus
I appreciate what you say.
The nearest you get to a starting point is your second option namely "work in markets that are 'news quiet' and keep your time horizon small".
You don't need the market to be 'news quiet' in something like index futures. News at best is merely a distortion to the pattern, either waylaying it for a little time or reinforcing it.
Keep the time horizon small. This is a good observation. The near term gives you everything. If you can optimize or maximize what the market daily offers, then you can arrive for each day with a big enough shovel to scoop out the market each day.
I appreciate what you say.
The nearest you get to a starting point is your second option namely "work in markets that are 'news quiet' and keep your time horizon small".
You don't need the market to be 'news quiet' in something like index futures. News at best is merely a distortion to the pattern, either waylaying it for a little time or reinforcing it.
Keep the time horizon small. This is a good observation. The near term gives you everything. If you can optimize or maximize what the market daily offers, then you can arrive for each day with a big enough shovel to scoop out the market each day.
I was just curious how you connected a dynamic, non-static view of the market with randomness in the first place.Quote from oddiduro:
I think we have moved into talking about the non-randomness of randomness.
I was all set to give a long-winded diatribe on the futility of statistical reliability (along with all other "technical" analyses) in a dynamic system before I read your last post -- but it seems I'd be preaching to the choir.
Good thread though, thanks.
Humans are random, if they were not random we would be living in Utopia by now. But since humans are self gratifying in their pursuits of Utopia the goal is never reached simply because their reach for self gratification can and will come at the expense of others.
We see this also in the markets, the learned will empty the pockets of the unlearned.
Life continues on....
PS, Bush is not random though he is a simple "DOG and PONY" show. First he used the "Saddam" card to drum up the war. Second he used the "TERROR" card to enter the war. Now he is using the "PATRIOT" card to defend a losing war and the publics beating drum for real answers not just rhetoric.
As they would say on the World Poker Tour show when you have a no win hand: .....he is drawing dead. How sad is that?....
We see this also in the markets, the learned will empty the pockets of the unlearned.
Life continues on....
PS, Bush is not random though he is a simple "DOG and PONY" show. First he used the "Saddam" card to drum up the war. Second he used the "TERROR" card to enter the war. Now he is using the "PATRIOT" card to defend a losing war and the publics beating drum for real answers not just rhetoric.
As they would say on the World Poker Tour show when you have a no win hand: .....he is drawing dead. How sad is that?....
OK, I'll predict a few stocks to try to disprove the "unpredictable randomness" theory.
Trying to predict all market action all the time is an exercise in futility. However, it is useful to recognize there are pockets of non-randomness that appear in the market as fleeting opportunities to exploit for profit. Price action in the market is not always efficient for all stocks all the time -- as the Worden Brothers put it, "it's not a stock market, it's a market of stocks."
And so one strategy that works is to scan stocks for market inefficiencies. Key principle: Buying and selling is not always reflected in the price! This is often apparent in low volume small cap stocks where it can take a month or more for buyers to chew through the available inventory; once the inventory is exhausted, price must go up to satisfy the buying demand. Here's an example:
I'll go out on a limb and say these stocks -- IDN, SBEI, VPS, NETM -- will go up between now and the next several weeks due to price inefficiencies I'm seeing in the charts. My prediction assumes there are no major news events that move the entire market.
I predict these will go up at least 10% in the next few weeks or so: http://stockcharts.com/candleglance/?IDN,SBEI,NETM,VPS|C|G17
Trying to predict all market action all the time is an exercise in futility. However, it is useful to recognize there are pockets of non-randomness that appear in the market as fleeting opportunities to exploit for profit. Price action in the market is not always efficient for all stocks all the time -- as the Worden Brothers put it, "it's not a stock market, it's a market of stocks."
And so one strategy that works is to scan stocks for market inefficiencies. Key principle: Buying and selling is not always reflected in the price! This is often apparent in low volume small cap stocks where it can take a month or more for buyers to chew through the available inventory; once the inventory is exhausted, price must go up to satisfy the buying demand. Here's an example:
I'll go out on a limb and say these stocks -- IDN, SBEI, VPS, NETM -- will go up between now and the next several weeks due to price inefficiencies I'm seeing in the charts. My prediction assumes there are no major news events that move the entire market.
I predict these will go up at least 10% in the next few weeks or so: http://stockcharts.com/candleglance/?IDN,SBEI,NETM,VPS|C|G17