Making a more scientific trading strategy

[NVP]

The problem with trying to find a 'scientific' trading method and strategy is basically this -

Most people are trying to find rules/methods with a load of past price data and a £400 Dell computer. Some are even doing it by hand.

But the big banks, funds and trading houses are collectivly spending probably in excess of $5billion per year on research and development, and have been doing so for the last 15 years.

That means 2 things, first they're looking at the same patterns you're looking at because their computers are tasked with looking at every possible price pattern there is, plus the ones you didn't know existed.

Also, look at their results. Generally they're not pretty unless in a big bull market in which cases their results are excellent. But then again so are the results of monkeys throwing darts at a quote screen (wherever the dart hits, that's the insturment to buy).

What this all means is it's generally futile for anyone to look for price patterns that can be programmed into a computer. Yes, it has and can work, but not for many and only really those with the best minds and research capabilities. But even then the results are probably going to be in the 10%-30% a year bracket (risk/adjusted though which is good).

10%-30% a year will basically get most people on this board nowhere as you have to assume most people don't have 6 or 7 figure accounts.

So what to do?

As others have suggested here, the big money is to be made with understanding how markets work and then becoming a discretionary trader. If you could take everyone here on this board who's making proper money and has been for a minimum of a few years, I'd bet big money that 98%+ of them are discretionary traders........

Good luck

good post...........dont follow dumb rules.....READ the market ......all the clues are there.....and the higher the timeframe the bigger the players are that will react to the moves

 

[ronblack]

anley,

The dot matrix printer, the windshield washer and a host other inventions were made by individuals, not huge corporations. Do not ever try to justify your lack of knowledge or failures using an appeal to the power of corporations. Corporations are made of people and even if they have more resources, it is the people that make the difference.

It sounds like you do not know where the big profits of large corporations come from. They mostly come from designing and pricing products for institutional clients, not from discretionary or system trading.

 

[the hare]

Corporations are made of people and even if they have more resources, it is the people that make the difference.

Very true, and most coroprations are completely dysfunctional. Anyone with the brains or motivation to figure this stuff out certainly isnt going to be working for someone else

 

[anley]

It sounds like you do not know where the big profits of large corporations come from. They mostly come from designing and pricing products for institutional clients, not from discretionary or system trading.

I know exactly where the profits come from, fees and working the bid-offer spread.

But the fact that many of these banks don't build and trade their own mechanical systems tells you all you need to know about the chances of an individual designing and operating one.

The days of simple systems that worked (in the 70s and 80s) are long gone. Now it's a money/research/brains race and the man in the street hasn't really got a chance.

But if anyone thinks I'm wrong then be my guest and try. This is what I think will happen -

You'll reaslise about 1-5 years down the line that your efforts produce nothing more than a fair return at best (assuming you understand this is a risk/adjusted world). Then if you're still interested in the markets you'll switch tacts and start to trade an analyse on a discretionary basis. Then if you're good the profits will come rolling in and you'll curse yourself for getting sidetracked with stupid system development which you now understand to have been a complete waste of time.

Always exceptions to the rule (above) but very very few........

Good luck

 

[sj8070]

This is a very interesting thread, a good discussion of the potential and pitfalls of trading systems.

Whilst it is clear to everyone that there is no mathematical "solution" to the markets and never will be, this does not mean that an edge cannot be developed on a medium term time frame.

Importantly, price action most certainly is not equivalent to Brownian motion (a truly random process), if you require proof beyond reasonable doubt of this, PM me, I have written a paper on the subject.

Therefore, if you accept that markets are not perfectly random (i.e. in some instances the expected change in price is non zero or non risk free rate), then you accept that a strategy can produce a positive expectation on P&L.

To me, whether this is discretionary or mechanical is immaterial. There is no form of analysis that the human mind can do that a computer cannot- even if you are a believe in gut instinct / feeling this is often the snap assimilation of large amounts of data done in the background, or unconsciously.

Practically speaking, having worked at a bank I can tell you that producing these sorts of systems or strategies is not within their remit at all. I can assure you there is no laboratory of coders tackling this problem in any major bank.

Hedge funds may well be different of course. I think members of this forum would be surprised how many hedge funds actually employ totally automated systems based purely on technicals as a key driver of return.

Whilst discretionary trading may well be more responsive to changes in market patterns, I'm loathe to play down the importance and profitability of automation.

 

[tomorton]

Hi anley - No doubt any system a large corporation would even consider using is going to be a lot more sophisticated than anything useful for the small account private individual. And I could never develop somethig to match it. It would be like me trying to develop an airliner in competition with Boeing.

Then agan, I 'm not trying to build an airliner, I just need to mackle together a glider, or even a hang-glider. People achieve this level of flight from garages and workshops all over the world. That's all I need to do and I am sure I can do it.

But another point you make, which I'm less clear on is that simple systems worked in the 70s and 80s but will now fail. Apart from the fact that the markets are bigger and work faster, what has changed to torpedo these systems? As per my earlier post, whatever it is isn't obvious on price charts.

 

[anley]

But another point you make, which I'm less clear on is that simple systems worked in the 70s and 80s but will now fail. Apart from the fact that the markets are bigger and work faster, what has changed to torpedo these systems? As per my earlier post, whatever it is isn't obvious on price charts.

Simple, everyone has cottoned on. Back in the 70s/80s there were very few people using technicals or even computers (there wern't many of them as well).

Take John Hendry (the Liverpool owner) as a good example. If you ever meet him ask him if system trading was simple compared to now when he started back in the 70s/80s. I bet he gives you a wry smile.

Same with options, pre 1980 most option market makers hadn't a clue about the maths involved for pricing, they priced them on orderflow and market view. Then the maths guys came in and took the old style market makers to the cleaners.

You also saw this with LIFFE options in the late 1980s, early part of the 90s. The Americans rolled into London (not the big banks, the independent option companies and traders) and took MOST of the money against their UK counterparts, why because the Americans were so much more sophisticted in their pricings, strategies and software. But people in the markets aren't that stupid, so the UK fellas got their act together over the next year and could then compete.

The point I'm making is that 98% of the time when it comes to systems and strategies, money, time and brains will win. That is life in the 21st century. Now the average retail system developer might have some brains but that's not going to win many prizes if he doesn't have time and money on his side.

Plus, I guarantee that whatever idea he comes up with - mix a positive ADR with a 37 period moving average combined with a Gap opening of less than 0.75% of previous close and the 240 MACD is bullish etc will have been tested and retested over the last 15 years about 100,001 times. In effect any possible and probable market scenario has been tested, front and backwards.

 

[tomorton]

Thanks anley, food for thought. Hmmm, I know nothing of trading pre-1999, barring having read Jesse Livermore a couple of years ago.

I guess it's tempting to assume that every simple system has been tested to the nth degree by somebody somewhere working for a large corporation. But that doesn't mean that it won't work, it just proves it's not what they want.

You and I and plenty of other individuals on this site could make money out of the simplest two MA crossover system, and of course in theory so could the funds and banks - but although it might be something we would do, it isn't what they're looking for so they will ignore it. That doesn't mean it can't work. I have to refer back to the glider v's airliner analogy. Would anyone pay me for a ticket on a glider I built in my garage? No, of course not, but if I just want to ride the thermals, it will do for me. I don't need to sell 500 tickets per flight and jet to 30,000ft, so I'm not competing with Boeing.

 

[VielGeld]

Imo, support and resistance should be the basis for any strategy.

I'm biased since I'm a newbie and it's all I use, but S/R is so incredibly prevalent and apparent that I can't see any strat not taking them into account. They can at least be used to set up rudimentary ranges for a trade's success/failure.

If you're interested in scientific trade ideas, I would recommend anything Victor Niederhoffer. His book is a classic, and you can read his blog here: Daily Speculations. You might not agree with the way he sets them up, but the mentality is probably what you're looking for.

 

[Pat494]

I was a real sucker, years ago for neural networks, swarm tech etc. and spent money on them
It looked like they were the answer to all trading problems. But sadly they didn't live up to the sales hype. I must have spent 100s of hours. The results weren't bad but not quite good enough. Just waiting for some kid in his bedroom to invent the next level. The regulars seem to have stalled and run out of ideas.

 

[the hare]

What really aggrieves me is the observation that all systems fail unexpectedly - sometimes they consistently succeed when price is in a trend and volatility is high, sometimes they consistently fail in a market of the same general chart / indicator characteristics.

This is an extremely good point. I honestly do think that its a case of being fooled by randomness. Take a TA based system, run several instances of the same strategy with minor changes in indicator parameters, and then run each instance across multiple instruments, and multiple timeframes.

You'll end up with something that resembles a bunch of equity curves generated from completely random systems. The parameter sets that performed well during a given period can often return the worst results over another period (and as you say, in what looks to be similar market conditions)

You'll also find that if you run random systems, there's often a strong correlation between the results that they produce, and the results obtained from TA based strategies (perhaps not surprising really as both are actually random)

I suppose the natural reaction initially is to try to find another filter, to make something that worked previously, work again, but this this just compounds the curve fitting /data mining bias problem. I'd probably argure that if you want to go down the mechanical route, the trick is to accept the reality of the situation, and to try and exploit that.

 

[DionysusToast]

I realise what I have written above sounds like a discussion on chess, which is indeed how I see trading.

I disagree - trading is more like this:

Squeal Like A Pig (Deliverance) - YouTube

 

[TakeYourProfits]

I know exactly where the profits come from, fees and working the bid-offer spread.

But the fact that many of these banks don't build and trade their own mechanical systems tells you all you need to know about the chances of an individual designing and operating one.

The days of simple systems that worked (in the 70s and 80s) are long gone. Now it's a money/research/brains race and the man in the street hasn't really got a chance.

But if anyone thinks I'm wrong then be my guest and try. This is what I think will happen -

You'll reaslise about 1-5 years down the line that your efforts produce nothing more than a fair return at best (assuming you understand this is a risk/adjusted world). Then if you're still interested in the markets you'll switch tacts and start to trade an analyse on a discretionary basis. Then if you're good the profits will come rolling in and you'll curse yourself for getting sidetracked with stupid system development which you now understand to have been a complete waste of time.

Always exceptions to the rule (above) but very very few........

Good luck

you have a point

 

[TakeYourProfits]

This is a very interesting thread, a good discussion of the potential and pitfalls of trading systems.

Whilst it is clear to everyone that there is no mathematical "solution" to the markets and never will be, this does not mean that an edge cannot be developed on a medium term time frame.

Importantly, price action most certainly is not equivalent to Brownian motion (a truly random process), if you require proof beyond reasonable doubt of this, PM me, I have written a paper on the subject.

Therefore, if you accept that markets are not perfectly random (i.e. in some instances the expected change in price is non zero or non risk free rate), then you accept that a strategy can produce a positive expectation on P&L.

To me, whether this is discretionary or mechanical is immaterial. There is no form of analysis that the human mind can do that a computer cannot- even if you are a believe in gut instinct / feeling this is often the snap assimilation of large amounts of data done in the background, or unconsciously.

Practically speaking, having worked at a bank I can tell you that producing these sorts of systems or strategies is not within their remit at all. I can assure you there is no laboratory of coders tackling this problem in any major bank.

Hedge funds may well be different of course. I think members of this forum would be surprised how many hedge funds actually employ totally automated systems based purely on technicals as a key driver of return.

Whilst discretionary trading may well be more responsive to changes in market patterns, I'm loathe to play down the importance and profitability of automation.

"Whilst it is clear to everyone that there is no mathematical "solution" to the markets and never will be" posted by sj8070




are you kidding me??? math defines EVERY SINGLE ASPECT OF OUR WORLD... even trading and it is an extremely powerful tool to be used in trading...

Calculus IS the market...

for example take a look at a typical calculus argument:


The Limit of f(x) as the function of X approaches a Defined variable of a function of the given curve.... ie... the Market

go to Khan Academy

its a free educational resource where you can learn college level calculus 1,2,3 Differential Equations & probabilities & Statistics... when you grasp those concepts, you will grasp and be able to visualize the mathematics involved in the market....

 

[DionysusToast]

"Whilst it is clear to everyone that there is no mathematical "solution" to the markets and never will be" posted by sj8070




are you kidding me??? math defines EVERY SINGLE ASPECT OF OUR WORLD... even trading and it is an extremely powerful tool to be used in trading...

Calculus IS the market...

for example take a look at a typical calculus argument:


The Limit of f(x) as the function of X approaches a Defined variable of a function of the given curve.... ie... the Market

go to Khan Academy

its a free educational resource where you can learn college level calculus 1,2,3 Differential Equations & probabilities & Statistics... when you grasp those concepts, you will grasp and be able to visualize the mathematics involved in the market....

People move the markets, not maths.

 

[TakeYourProfits]

People move the markets, not maths.

math is a method of understanding the world around us, and that includes trading.... in itself the market is nothing more than the psychological measurement of the combined decisions and activities made by all of the humans, the people, as you say, in the world...

the way in which we interact with each other is measured, whether you would like to admit it or not using math.... the values of each security traded are quantified using NUMBERS like 1,2,3,4,5... its ALMOST like a binary measurement of 1's & 0's if you will, just a tad bit more complex... which is why we use math to understand it... because when we have a mathematical understanding of the market, we have a fundamentally deeper level of understanding simply because of the structure of the market itself!!!

but hey dont ask me... im nobody, i just trade my own account in a consistently profitable fashion everyday and thats all i do, using MATH... just ask those "Math Guys" from the 80's that somebody mentioned earlier who cleaned house when they were more efficiently able to price the options market using the black scholes MATHEMATICAL model...

I once heard a story that said all the traders in the trading pits gave a standing ovation to black & scholes when they visited the floor one day for the innovation they took to the market that made it possible to trade options efficiently... using math... without math, the options market wouldnt be what it is today... OUR ENTIRE WORLD WOULDNT BE WHAT IT IS TODAY....

so i dont know what to tell you , but i think anybody would be hard pressed to make an any kind of real argument against Mathematics in effective and consistently profitable trading activities...

so good luck figuring out a way to measure your peoples without using math, just make sure you use tight stop losses please... preferably defined by price structure itself... but do whatever you want...

Trade the market on your own terms... that is the best policy... Math works for me... A trading plan can be made using absolutely anything, so don't think that anyone way is going to give you the holy grail.... professional money managers, only win an average of 4 out of 10 trades they make... MAYBE even 5... the key is trade, risk, and money management... not trading strategy... (well you have to at least have a positive expectancy, which unfortunately for you is more math, but you get what i mean...)

anything is good so long as its objective, consistent, and that you stay disciplined to the money and risk management rules you have created to trade on your own TERMS.... :clap:

 

[anley]

I'm going to leave it like this -

By all means use and test systems, I used to do it myself and yes, with hindsight it was generally a waste of time. Having said that you do learn plenty about the markets and technicals so the time is still of value.

But if you want the best chance of making money, and excellent money at that, go discretionary. Plus, it's a far more interesting way to look at the markets and trade.

Good luck!

 

[DionysusToast]

math is a method of understanding the world around us, and that includes trading.... in itself the market is nothing more than the psychological measurement of the combined decisions and activities made by all of the humans, the people, as you say, in the world...

the way in which we interact with each other is measured, whether you would like to admit it or not using math.... the values of each security traded are quantified using NUMBERS like 1,2,3,4,5... its ALMOST like a binary measurement of 1's & 0's if you will, just a tad bit more complex... which is why we use math to understand it... because when we have a mathematical understanding of the market, we have a fundamentally deeper level of understanding simply because of the structure of the market itself!!!

but hey dont ask me... im nobody, i just trade my own account in a consistently profitable fashion everyday and thats all i do, using MATH... just ask those "Math Guys" from the 80's that somebody mentioned earlier who cleaned house when they were more efficiently able to price the options market using the black scholes MATHEMATICAL model...

I once heard a story that said all the traders in the trading pits gave a standing ovation to black & scholes when they visited the floor one day for the innovation they took to the market that made it possible to trade options efficiently... using math... without math, the options market wouldnt be what it is today... OUR ENTIRE WORLD WOULDNT BE WHAT IT IS TODAY....

so i dont know what to tell you , but i think anybody would be hard pressed to make an any kind of real argument against Mathematics in effective and consistently profitable trading activities...

so good luck figuring out a way to measure your peoples without using math, just make sure you use tight stop losses please... preferably defined by price structure itself... but do whatever you want...

Trade the market on your own terms... that is the best policy... Math works for me... A trading plan can be made using absolutely anything, so don't think that anyone way is going to give you the holy grail.... professional money managers, only win an average of 4 out of 10 trades they make... MAYBE even 5... the key is trade, risk, and money management... not trading strategy... (well you have to at least have a positive expectancy, which unfortunately for you is more math, but you get what i mean...)

anything is good so long as its objective, consistent, and that you stay disciplined to the money and risk management rules you have created to trade on your own TERMS.... :clap:

Math has only the ability to model what has occured in the markets.

To trade, you have to be ahead of your competitors. Math will not do that for you.

You cannot, for example, mathematically resolve the game of poker. The reason for this is quite simple - much of the game is based on bluffing. The markets are the same. Math will never be able to figure out if the opponent is bluffing or not, it will only be able to play the hand according to probability.

For this reason, the math driven poker solution will ALWAYS lose. Math will dictate that the best hand is worth a higher stake. The opponents will know that math never bluffs and so whenever math dictates a higher stake, the opponents will know his hand & fold.

Common Sense > Math

 

[rawrschach]

I also attempt to use a scientific approach-

Define a 'scientific approach'.

 

[rawrschach]

Math has only the ability to model what has occured in the markets.

To trade, you have to be ahead of your competitors. Math will not do that for you.

You cannot, for example, mathematically resolve the game of poker. The reason for this is quite simple - much of the game is based on bluffing. The markets are the same. Math will never be able to figure out if the opponent is bluffing or not, it will only be able to play the hand according to probability.

For this reason, the math driven poker solution will ALWAYS lose. Math will dictate that the best hand is worth a higher stake. The opponents will know that math never bluffs and so whenever math dictates a higher stake, the opponents will know his hand & fold.

Common Sense > Math

Not *entirely* true. There's nothing wrong with math. The problem is that the Finance/Econ world uses the wrong math.

When I was doing my MSc in Econ every model started with 'Assume a well behaved i.i.d. distribution', 'no-bubble/ponzi conditions'; i.e. zero relevance to reality. Problem is that people learn this bull**** and get jobs in banks/government and apply their 'perfect situation' models to real life.

Of course academics don't want to admit that real life outcomes have a Mandelbrotian distribution because that would put them all out of jobs. If you embrace the fat tail instead of try to assume it away math can help you make a ****load of money from trading.

The problem with a 'scientific' approach to trading is that contemporary scientific methodology emphasises only the observable whilst a truly discretionary approach builds on largely intangible experience which is what turns into your edge.