A Year With No Setups

[DionysusToast]

I've read that there are only really 7 basic story plotlines. So perhaps every story you've read isn't as different as you think. But perhaps that was kinda your point

Read Chuck Phalaniuk & get back to me on that one. Start with Pygmy.

A lot seems to have been said about mechanical not working. There are algorithms in the market that work. Does anyone dispute this? So mechanical can work.

I met a guy at a **** up not long back - day before the Singapore GP I think it was. He was 41 & retired. He'd been a trader at a Canadian Bank. After retiring he became an advisor to a firm that developed trading algo's. When he told me this, I said "arbitrage?" he said "yes".

The way that retailers trade is generally "guess the direction". A lot of us retailers make a presumption that this is what trading is all about.

If you get your commissions down to zero, have a co-located server and some clever math, you can, for a while do very well out of arbitrage. Still, it's a competitive market and by it's very nature (just like HFT) you end up going for smaller & smaller chunks of change.

So - whilst there is lots of evidence that trading algos work, I have never seen any that they are trying to pin the tail on the donkey like a retail trader.

I don't trade purely mechanically. But I noticed you said there's not a cat's chance in hell you could program your method. I have often wondered this. And I suppose in theory it should be doable, it would just be a nightmare to do, and probably not worth the time. By the time I'd finish, my methods (or the market) may have changed slightly and I'd have to redo it. What is there about any method that isn't relying on some information that a computer could obtain and interpret as you do?

For many years the best chess players in the world said that a computer would never be able to beat the best human player. That computers were too predictable, and would fall into traps a human could set. Eventually a program came along that beat the best. I'd guess that there are already many mechanical systems trading the market daily that do better than what you or I can achieve.

Chess is a problem with a finite solution. There are 64 squares and each piece can only move in a pre-defined way. The markets are not finite and this is what causes some issues in programming certain techniques.

The thing is, many people believe that there is a mechanical solution but one has never been demonstrated. Not on this board, nor other boards. Not in any trading books or skype rooms - not that I have seen.

I think the halfway house mechanical system is something that is eminently doable. The halfway house being a mechanical system that will work in certain conditions. This still leaves you to decide when to turn the thing on or off. I think the robot you leave on day in day out unattended to make $$$ for you is a myth.

Anyway - what does that have to do with tying one hand behind your back, sir?

 

[Hotch]

For many years the best chess players in the world said that a computer would never be able to beat the best human player. That computers were too predictable, and would fall into traps a human could set. Eventually a program came along that beat the best. I'd guess that there are already many mechanical systems trading the market daily that do better than what you or I can achieve.

More than that, Chess is actually solvable, and will be solved soon(ish). They did draughts a few years ago, and Chess is the same just a lot more leg work.

 

[DionysusToast]

More than that, Chess is actually solvable, and will be solved soon(ish).

This is rather depressing, isn't it?

How about poker?

 

[Hotch]

This is rather depressing, isn't it?

How about poker?

It is. I was discussing with my dad, lots of manufacturing jobs have gone to robots, but 50 years down the line most of the current jobs will be done by computers, of course we may come up with new more complex jobs, meaning more time spent in education, but it's inevitable that a large proportion of the population is going to become obsolete (in employment terms). Then what happens...

As for poker, I'm sure there are a fair few poker bots which play honest and clean up against the fish, higher stakes is less likely as the players will play mathematically correct, but you could probably measure snap calls etc and given enough data make a good show of it.

 

[DionysusToast]

It is. I was discussing with my dad, lots of manufacturing jobs have gone to robots, but 50 years down the line most of the current jobs will be done by computers, of course we may come up with new more complex jobs, meaning more time spent in education, but it's inevitable that a large proportion of the population is going to become obsolete (in employment terms). Then what happens....

Maybe - but the human race does not have to go in that direction. Personally, I'd love to automate Missus Toast. Then I could switch her off from time to time.

As for poker, I'm sure there are a fair few poker bots which play honest and clean up against the fish, higher stakes is less likely as the players will play mathematically correct, but you could probably measure snap calls etc and given enough data make a good show of it.

There's 2 parts to playing poker.

First of all is playing the odds according to your hand.
Second is trying to read your opponents behaviour to figure if they are bluffing or not.

If you played the odds, and only bet high when your hand was good, you would not make any money. Your opponent would know by your betting behaviour when you had a good hand and when you didn't. The likely outcome is that no bets would be placed. If you do not have a good hand, you don't bet. If you have a good hand, everyone knows it and so they don't bet.

Hence, pitching a computer against a pro or a computer in this situation would likely end in stalemate.

 

[scose-no-doubt]

I can only play poker in the second way. I see outliers occur far too frequently to place any faith in a mathematical approach but I'm just an unlucky guy in general and never win games of chance. I think that shaped my approach to trading too tbh. Probably why I just don't buy the classic "set-up" as discussed on T2W. S'all qualitative to me but that should also encompass the numbers.

 

[Hotch]

That's what I was saying DT, to use your 2 parts bit.

Against fish, you only need to play mathematically correct. They're that bad. Against pros they're on a similar level and you'd have to start looking for tells which complicates it somewhat, still plausible though, especially as the computer isn't going to give a tell. Of course the pro could try and fool the computer, but it's against human nature to be random (that's why you should always get a lucky dip on the lottery instead of picking numbers), so I reckon it'll be done eventually.

 

[robster970]

More than that, Chess is actually solvable, and will be solved soon(ish). They did draughts a few years ago, and Chess is the same just a lot more leg work.

Analysing to depth is still too complex at the moment. That's why dedicated chess software relies upon game histories. All it takes is for a Grand Master to go off piste and the software is in trouble.

Read about IBM's deep blue and some of the controversy surrounding the Kasparov games.

With a rating of circa 1200, I get floored by chess software. I think you need to up around 1800+ to be getting one over on these things.

 

[Hotch]

Analysing to depth is still too complex at the moment. That's why dedicated chess software relies upon game histories. All it takes is for a Grand Master to go off piste and the software is in trouble.

Read about IBM's deep blue and some of the controversy surrounding the Kasparov games.

With a rating of circa 1200, I get floored by chess software. I think you need to up around 1800+ to be getting one over on these things.

You misunderstand me robster, I'm not saying that they'll make a computer that will beat anyone, I'm talking about them actually SOLVING it, as in, "this is the optimal move for every situation". There's a difference.

Checkers Is Solved

 

[robster970]

You misunderstand me robster, I'm not saying that they'll make a computer that will beat anyone, I'm talking about them actually SOLVING it, as in, "this is the optimal move for every situation". There's a difference.

Checkers Is Solved

Gotcha now. Yes I agree it's only a matter of time.

 

[Lord Flasheart]

yes but it will all have to be based on historic results,just like everything else,so it may have been greta over the last 100 years,but the next 100 it might not work on

 

[Uilen]

There is nothing wrong with using charts. You just cant use setups. Just for the sake of this discussion of course.

I guess I'd be just trading the news, as it curbs the randomness of the price for some time. Sort of "can't beat them, join them".

 

[DionysusToast]

That's what I was saying DT, to use your 2 parts bit.

Against fish, you only need to play mathematically correct. They're that bad. Against pros they're on a similar level and you'd have to start looking for tells which complicates it somewhat, still plausible though, especially as the computer isn't going to give a tell. Of course the pro could try and fool the computer, but it's against human nature to be random (that's why you should always get a lucky dip on the lottery instead of picking numbers), so I reckon it'll be done eventually.

The computer of course will give a MASSIVE 'tell'. The computer would place higher bets on better hands. There's the tell.

Hence, the computer will never make any money from the game.

People will make money from people. Computers against pro players or computers against computers will be a wash. The nature of the game is that it is a level playing field without the bluffing.