Yeah we need diversions once the market is closed
What TA tool can detect random walk data?
- Thread starter ivanbaj
- Start date
Yeah we need diversions once the market is closed
Since traders spend so much time staring at charts, it's natural to ask questions like this from time to time. As it turns out, it doesn't matter if you see the same patterns, indicator readings, etc. on a market-based chart or on a simulated chart. The fact that you do simply means these are the inherent properties of charts. You just need to do what you've always done and look for the higher probability events.
Take AAPL
It is random that Apple was the company to succeed with the iPod.
It could have been any of these:
Phillips
RealNetworks
Rio
Sony
...
etc.
and few hundred or thousand garage companies in 1999-2000
But here were are, with Apple the winer. It is a complete random act. Natural selection.
How can one profit?
One strategy is to buy all of the players and hope one will make it. But how do you know how many players are in a particular domain. Why would we you think that people will be crazy about buying MP3 players. Do you even know the boundaries of the domain? From the small iPod, Apple went to compete with Nokia and RIM etc... Also some of the players could be private companies. Unless you are a venture capitalist you can't get in. If one has the skills to identify the playing field and is able to invest in most players while the cost is extremely low this can produce predictable results without the need of luck.
The other option is to monitor the space until the natural selection of the randomness starts to show a clear winner. Then load up on the winner's stock. But... most likely the stock price will raise faster then one can decide who the actual winner is.The risk will rise faster than the signs of a clear winner. If one has the skills to find out the winer while the price/risk is still low then this can/should work without the need of luck.
Or some sort of both approaches. Monitor the space till few potential winners emerge and buy all of them early in the process.
Can this be applied to TA? What if a group of stocks is selected that using our skills we determine that at least one of them will make a big move. We don't know which one. We buy all of them and wait. I think there was a guy that bought all $1 stocks after the big crash in 1929. He did well.
It is highly unlikely that there are millions given away by sloppy traders on the big volume stocks or futures. In a very low volume stocks ...maybe.
But on ES and the likes it is impossible for me to believe that the players will be giving a way millions. We all know that big %. 80% or more of the trades are done by computers. It is not that difficult to randomize the transactions so the information is hidden. The small guys that show their cards are slowly eaten by the commissions and the market makers. You can't profiting from them unless you are MM.
The day-trading FOREX guys (the small ones) are the most amusing bunch of all. Surely there could not be any TA tool that will work on the data given to them.
Here is another idea. Buy/sell OTM options. One day you will make it big. Just regularly buy your "Lotto" ticket.
Or when once or maximum twice in a lifetime the opportunity presents itself. Load up and if it works retire and never look back.
Any other ideas?
It is random that Apple was the company to succeed with the iPod.
It could have been any of these:
Phillips
RealNetworks
Rio
Sony
...
etc.
and few hundred or thousand garage companies in 1999-2000
But here were are, with Apple the winer. It is a complete random act. Natural selection.
How can one profit?
One strategy is to buy all of the players and hope one will make it. But how do you know how many players are in a particular domain. Why would we you think that people will be crazy about buying MP3 players. Do you even know the boundaries of the domain? From the small iPod, Apple went to compete with Nokia and RIM etc... Also some of the players could be private companies. Unless you are a venture capitalist you can't get in. If one has the skills to identify the playing field and is able to invest in most players while the cost is extremely low this can produce predictable results without the need of luck.
The other option is to monitor the space until the natural selection of the randomness starts to show a clear winner. Then load up on the winner's stock. But... most likely the stock price will raise faster then one can decide who the actual winner is.The risk will rise faster than the signs of a clear winner. If one has the skills to find out the winer while the price/risk is still low then this can/should work without the need of luck.
Or some sort of both approaches. Monitor the space till few potential winners emerge and buy all of them early in the process.
Can this be applied to TA? What if a group of stocks is selected that using our skills we determine that at least one of them will make a big move. We don't know which one. We buy all of them and wait. I think there was a guy that bought all $1 stocks after the big crash in 1929. He did well.
It is highly unlikely that there are millions given away by sloppy traders on the big volume stocks or futures. In a very low volume stocks ...maybe.
But on ES and the likes it is impossible for me to believe that the players will be giving a way millions. We all know that big %. 80% or more of the trades are done by computers. It is not that difficult to randomize the transactions so the information is hidden. The small guys that show their cards are slowly eaten by the commissions and the market makers. You can't profiting from them unless you are MM.
The day-trading FOREX guys (the small ones) are the most amusing bunch of all. Surely there could not be any TA tool that will work on the data given to them.
Here is another idea. Buy/sell OTM options. One day you will make it big. Just regularly buy your "Lotto" ticket.
Or when once or maximum twice in a lifetime the opportunity presents itself. Load up and if it works retire and never look back.
Any other ideas?
It could have been any one of those companies but it was Apple. This was over the past 4 years. Since Apple released the first iPod it was a hit and the stock price has risen on that success since 2006 or whenever it first came out. It is not random that sales of the thing exploded and kept growing over time and Apple created a branded image to maintain its advantage. You may not be able to pick the start of the trend but randomness goes to the wayside as buying activity, increased revenues, higher forecasts/realizations lead to the expected behavior of continued buying.
Of course I over simplified it greatly, but the point is the move from $80 at the start of 2007 to $251 right now over the past 3.5 years was not a random drift during a recession, it was an explosive trend that still shows little sign of slowdown. Again over simplification so forgive me but the point is nothing like this is random, Apple spends hundreds of millions to ensure its brand identity is secure and leads to growth.
Of course I over simplified it greatly, but the point is the move from $80 at the start of 2007 to $251 right now over the past 3.5 years was not a random drift during a recession, it was an explosive trend that still shows little sign of slowdown. Again over simplification so forgive me but the point is nothing like this is random, Apple spends hundreds of millions to ensure its brand identity is secure and leads to growth.
If it's not random walk, what are the other possibilities? For example you could model data as random walk or white noise or abcxyz. Take all of these models into account, and you should be able to somewhat tell how to classify different data.
Basically, you want somebody to *give* you an indicator that detects profitable set-ups!!! First, nobody in their right mind would do that. Second, I doubt it even exists. The market can't be reversed engineered for causal data points. The market can only be forward-tested under stated parameters. You have to do the work yourself.
The only issue I see as to whether you care about the Random Walk or not - is application to your trading system.
Most of my trading is automated - I don't click the mouse and make decisions.
Therefore the models I use is very dependent upon the data that feeds the algos. If that data has any market bias, that is measurable - (not random) - then you can improve the system.
Right now, I do NOT account for random walk (I also don't believe it). I am just wondering what other's think and have they applied it to an automated system.
Most of my trading is automated - I don't click the mouse and make decisions.
Therefore the models I use is very dependent upon the data that feeds the algos. If that data has any market bias, that is measurable - (not random) - then you can improve the system.
Right now, I do NOT account for random walk (I also don't believe it). I am just wondering what other's think and have they applied it to an automated system.
If your system does any kind of time series analysis or modeling, then ideally it should take into account random/chaotic/noise influence. If your system is not based on time series analysis but is instead based on technical analysis or other methods, then randomness won't matter.
Book search for "random walk time series":
http://www.amazon.com/s/ref=nb_sb_n...ield-keywords=random+walk+time+series&x=0&y=0
"If your system does any kind of time series analysis or modeling, then ideally it should take into account random/chaotic/noise influence."
We do use a levy flight in our algo, but have avoided using random walk data. We also do have a volatility noise band that is a fairly decent filter.
I guess if you feel the markets are NOT a random walk, then to use random walk in any model kind of refutes your assumption. Since I don't believe in random walk (while it may look random) - we have not used random generation in the modeling or even our time series analysis.
I can tell you it is difficult to build around and not using random generation models, since that seems to be the status quo.
I still find it odd that people use normal distribution functions.
But what do I know....
We do use a levy flight in our algo, but have avoided using random walk data. We also do have a volatility noise band that is a fairly decent filter.
I guess if you feel the markets are NOT a random walk, then to use random walk in any model kind of refutes your assumption. Since I don't believe in random walk (while it may look random) - we have not used random generation in the modeling or even our time series analysis.
I can tell you it is difficult to build around and not using random generation models, since that seems to be the status quo.
I still find it odd that people use normal distribution functions.
But what do I know....
============
Good TA indicator ;
your brain
The reason why the market is not random ;
is most people are in the market for several good or not so good reasons,
not a random walk/mindless stroll.
The only ''random walk'' i ever saw is a female friend shopping.LOL. But that two,was not really real random, the purpose was to buy produce in an emotion-fun way.LOL
Founder of IBD[Investors Business Daily];
gives you several more good reasons , why its NOT random.See His books.