Why I don't believe in current Machine Learning: Emergence

[Jack.Yarn]

Let's break down your statement

The movement in the markets is not random but the way humans cause price changes is beyond any man made algorithm.

A staggering percentage of trading is done by machines, not just in the US but in the world:

Not Just Stocks … High Frequency Traders Might Be Manipulating Futures, Options, Bonds, Currency and Commodities Markets As Well

http://www.washingtonsblog.com/2010...currency-and-commodities-markets-as-well.html

High Frequency Trading Dominates UK Stock Market

http://www.washingtonsblog.com/2011/01/high-frequency-trading-dominates-uk-stock-market.html

And in stocks it is a staggering 86% done by machines. Morgan Stanley:

"Trading by “real” investors is taking up the smallest share of US stock market volumes [since Morgan Stanley started keeping track 10 years ago.]

The findings highlight how US trading activity is increasingly being fuelled by fast turnover of shares by independent firms and the market-making desks of brokerages, many using high-frequency trading engines. [actually all of the market-making desks are using it.]

The proportion of US trading activity represented by buy and sell orders from mutual funds, hedge funds, pensions and brokerages, referred to as “real money” or institutional investors, accounted for just16 per centof total market volume in the form of buying, and13 per centvia selling in the final quarter of last year, according to analysis by Morgan Stanley’s Quantitative and Derivative Strategies group."

Therefore, your conclusion

It has no rational logic to it and therefore will flummox both man and machine.

does not follow from your premise.

Q.E.D.

 

[Jack.Yarn]

It baffles me, that on a trading site, people don't know the most basic facts about the markets. Probably because people are few and far between in markets anymore!

Sun Tzu

“Whoever is first in the field and awaits the coming of the enemy, will be fresh for the fight; whoever is second in the field and has to hasten to battle will arrive exhausted.”

―Sun Tzu,The Art of War

 

[maxpi]

I dabble in coding trading tools and systems. It's stunningly difficult to make a definition of what I see on a chart such that a computer can work with it. Makes me appreciate the human mind a lot more actually.

I worked for a German company. They were hugely successful and I discovered that one of their secrets was that they knew how to keep the computers between employees ears working well. They treated workers well, they treated me well, I'm inventive and capable and they got a lot of good work out of me. I was glad to be there. Some cheapshit Americans bought the place and immediately they lost the output from the computer between my ears. Instead I made a list of what I could get out of the place before getting out of the place and I used the computer between my ears to complete that list rather than working on their shit...

I kind of wonder how we are going to get all those superior introvert doctors and lawyers to move over so machines can serve us? I guess if we render them obsolete they will go eventually but they make the rules so I'm guessing they will pass laws that obstruct us. Maybe the Insurance company money will shove them off their self-appointed thrones?

 

[Rogerahan]

I think I understand the point you are trying to get across and your reasoning. And I would definitely say trading from some sort of ML black box system which utilises spurious relationships would be dangerous. However, I don't think that's how ML is generally used.

For example, In the medical field they use ML. They use it to find out the likelihood of someone developing a disease (which is literally a case of life or death). But they also use it to gain better insight into the potential causes of a disease, and the significance of a factor in contributing to a persons diagnosis.

As a caveat though I have read about a least one profitable trader who trades like the way you said in 'Hedge Fund Market Wizards'. His name is Jaffray Woodriff and he runs Quantitative Investment Management (QIM) with over $5b in funds under management. He literally generates hundreds of secondary variables based on price (i.e. volatility), and will basically data mine the combination of those variables to give him a new and profitable trading model.

But even in the example above, he isn't using non logical variables, he is using direct derivatives of the price.

I would say the majority of people responsible for fund allocation would not allocate any major amount of capital to a system that does not have some sort of rational or understandable economic underpinning to it. This could be based on a new economic or market hypothesis, or a more traditional one.

For instance Ray Dalio who only uses Fundamental Macro data, does not make discretionary decisions on trades, he says and I quote " Those criteria-I call them principles-are systematised.....We don't make decisions about individual positions".

I am sure his company has used machine learning in some aspect of its business, maybe to find how to parameterize and relate the factors of his trading models based on some sort of weighted and parameterized look back period.

Here is a quote from Business Insider;

"Dalio told Ackman at the Harbor Investment Conference earlier this month that AI already factors into Bridgewater's investment strategy. He explained:

I can be short or long anything in the world, and I'm short or long practically everything. I don't have any bias, so I do it in a very fundamental way... we use a lot of artificial intelligence type of approaches to think about portfolio theory. I use a lot of financial engineering to basically take a whole bunch of uncorrelated bets..."

The problem with the Human brain, is that sometimes we see and conclude things we want to see and conclude. Because if there is not some sort of definite quantifiable way to define something, then it becomes whatever we want it to be at the time. There needs to be objectivity in analysis.

I believe ML can aid with our analysis and objectivity, as long as it used in the right ways.

 

[Jack.Yarn]

Elon Musk joins more than 100 tech bosses calling for ban on killer robots

• In an open letter to the UN, a group of 116 robotics specialists — including Tesla's Elon Musk and Google's Mustafa Suleyman — urged the organization to take action against "weapons of terror."
• "We do not have long to act. Once this Pandora's box is opened, it will be hard to close," the letter said.
• A UN group had been scheduled to meet on Monday to discuss autonomous weapons, including drones, automated machine guns and tanks, but the meeting has been postponed until November.

Sam Meredith |@smeredith19
Published 2 Hours Ago | Updated 33 Mins AgoCNBC.com

https://www.cnbc.com/2017/08/21/kil...-tech-bosses-to-call-for-an-outright-ban.html

 

[OddTrader]

Elon Musk joins more than 100 tech bosses calling for ban on killer robots

• In an open letter to the UN, a group of 116 robotics specialists — including Tesla's Elon Musk and Google's Mustafa Suleyman — urged the organization to take action against "weapons of terror."
• "We do not have long to act. Once this Pandora's box is opened, it will be hard to close," the letter said.
• A UN group had been scheduled to meet on Monday to discuss autonomous weapons, including drones, automated machine guns and tanks, but the meeting has been postponed until November.

Sam Meredith |@smeredith19
Published 2 Hours Ago | Updated 33 Mins AgoCNBC.com

https://www.cnbc.com/2017/08/21/kil...-tech-bosses-to-call-for-an-outright-ban.html

I totally agree.

Once a country or a political leader think there will be no human life of their own country would be lost in a war, the whole world will be finished due to unlimited wars.

We don't really need to travel around the world in two hours! The money for developing this kind of technologies should be spent on homelessness, etc.


We don't want a future society offering no employment for no income to spend on any thing. The whole economy in developed countries could be collapsed! However, underdeveloped countries can survive.

https://en.wikipedia.org/wiki/Amish

 

[fan27]

As a caveat though I have read about a least one profitable trader who trades like the way you said in 'Hedge Fund Market Wizards'. His name is Jaffray Woodriff and he runs Quantitative Investment Management (QIM) with over $5b in funds under management. He literally generates hundreds of secondary variables based on price (i.e. volatility), and will basically data mine the combination of those variables to give him a new and profitable trading model.

Interesting. That is exactly the approach I am taking with my new backtest engine. Good to know I am in good company.

fan27

 

[Jack.Yarn]

...


We don't want a future society offering no employment for no income to spend on any thing. The whole economy in developed countries could be collapsed! However, underdeveloped countries can survive.

That's a really interesting insight that never occurred to me!

 

[OddTrader]

OddTrader said:
...


We don't want a future society offering no employment for no income to spend on any thing. The whole economy in developed countries could be collapsed! However, underdeveloped countries can survive.

That's a really interesting insight that never occurred to me!

UBI could be merely a poisonous pill. A government cannot afford much payment forever for all citizens without firstly receiving enough taxes from limited number of income-generating citizens.

Maintaining operating expenses on basic infrastructure/services would require much money.

When combining small number of tax-paying citizens and huge payout for UBI and national operating expenses, there would be minimal exposable 'income' for every individuals/citizens, probably the tax-paying citizens included.

That means no any money would be spending on anything for new clothes, dinning, rental/buying-house, new-cars, bank-saving-account, education (for what? Jobs?), entertainments, etc.

Besides, who will be paying off for the costs/finance of producing/deploying robots when nobody else coming in for dinning, shopping, travelling, zero-delivery demand of pizzas, minimal online shopping on Amazon.com, etc.

The whole value-chain of consumer economy would be broken down. Especially after a few generations of launching advance robots that will be getting more smart, versatile, sophisticated, very intelligent, very very creative and tricky/slippery to a degree even beyond human's control (due to limited IQ for humans), and very expensive.

An example is: an Advance robot would be reasonably instructed not to follow any external instructions coming from any external humans or robots that the Advance robot itself, after technical evaluation, considers them of inferior and low IQ.

Any venture investors joining this kind of AI projects to replace human jobs could truly kill many of developing/developed economies.

That would be just like a huge prison for all people!

 

[OddTrader]

" An example is: an Advance robot would be reasonably instructed not to follow any external instructions coming from any external humans or robots that the Advance robot itself, after technical evaluation, considers them of inferior and low IQ. "

A kid could ask, "Dad, go to the middle of the heavy traffic highway to pick up my toy just dropped, please!"