Random trading systems - "monkey with a pin"

[ryandj2222]

Ok now I am sure I am going to get some LOLz from this but it's a genuine question.

I've seen a few articles and papers about random trading strategies and how they can beat a fair chunk of investors performance quite regularly.

Things like...
PLOS ONE: Investment Strategies Used as Spectroscopy of Financial Markets Reveal New Stylized Facts

PLOS ONE: Investment Strategies Used as Spectroscopy of Financial Markets Reveal New Stylized Facts

Computer Simulation Suggests That The Best Investment Strategy Is A Random One - Forbes

the best strategy for investment isn’t any kind of plan at all. Rather, the authors of a new study suggest, the best way to invest is to invest randomly.

What do people think about this?

I think it sounds interesting and could be fun to try, with a very diverse set of shares, funds, etfs or something, but I would want to use some risk management on it.

Thoughts?

 

[Shakone]

I would suggest you try it (paper trade it, not real money) and you may learn something useful.

It seems to me that with any strategy there is an element of randomness, so the articles seem to be comparing strategies with different degrees of randomness. How random is the strategy that is examined in the articles? Is it that the system goes long on a randomly selected stock...or that it randomly selects a stock and then randomly chooses long or short at a random time?

 

[Liquid validity]

Ok now I am sure I am going to get some LOLz from this but it's a genuine question.

I've seen a few articles and papers about random trading strategies and how they can beat a fair chunk of investors performance quite regularly.

Things like...
PLOS ONE: Investment Strategies Used as Spectroscopy of Financial Markets Reveal New Stylized Facts


Computer Simulation Suggests That The Best Investment Strategy Is A Random One - Forbes


What do people think about this?

I think it sounds interesting and could be fun to try, with a very diverse set of shares, funds, etfs or something, but I would want to use some risk management on it.

Thoughts?

No lulz from me.
This is the approach I take

 

[ryandj2222]

No lulz from me.
This is the approach I take

Does it work well for you?

How do you randomly pick the stocks/ instruments to trade? and do you add in any other rules?

 

[Liquid validity]

Does it work well for you?

How do you randomly pick the stocks/ instruments to trade? and do you add in any other rules?

It works acceptably well.
I'm not sure I would advocate randomly picking an instrument, but thats just me.
Other rules are essential.
Random entry alone does not work,its the exits and trade management
that make it 'work'.

 

[shadowninja]

I've messed about with this in the past. You could even pick an arbitrary reason for entering (a plane flies overhead). Whilst it's true that managing the trade is important, if you keep taking duff entries, you're going to throw away money slowly. Perhaps if you have DOM you can get out for a tick or scratch a trade but if you're playing the retail trading game then you'll just lose money through the spread over the long term.

Of course, just because it didn't work for me, doesn't mean you won't become a billionaire by 2014.

 

[ryandj2222]

I've messed about with this in the past. You could even pick an arbitrary reason for entering (a plane flies overhead). Whilst it's true that managing the trade is important, if you keep taking duff entries, you're going to throw away money slowly. Perhaps if you have DOM you can get out for a tick or scratch a trade but if you're playing the retail trading game then you'll just lose money through the spread over the long term.

Of course, just because it didn't work for me, doesn't mean you won't become a billionaire by 2014.

I'd doubt very much to become a billionaire, or even a millionaire, by next year. I think the appeal to me is the laziness approach. If I could make more or less the same as I do now (which is around 10-20% a year for a couple of years medium term trading) without really spending any time on it, it might be nice.

Something to think about anyway. Yes too many duff entries and it would blow the account, but that's the same for "planned" entries too... each one we never know if it will go the right way, so it's the same for a randomly planned entry.

 

[counter_violent]

Oh good, an interesting topic for a change.

 

[cablemonster]

Ok now I am sure I am going to get some LOLz from this but it's a genuine question.

'monkey with a pin' - racist

 

[shadowninja]

Something to think about anyway. Yes too many duff entries and it would blow the account, but that's the same for "planned" entries too... each one we never know if it will go the right way, so it's the same for a randomly planned entry.

That really doesn't make sense. If you have a way of trading that is profitable, then all entries are valid and should be taken because you know that ultimately you are going to be profitable. Bad entries only come about because you don't have a system that actually brings long term profit.

To expand, when you open a trade, because it's completely random, you won't know whether the conditions (because there are none) are right or not so whether to hold, cut or scratch. Mind you, you could just make up another arbitrary reason for closing early eg the plane you saw banked left.

In any case, try it and see how you get on.

 

[ryandj2222]

That really doesn't make sense. If you have a way of trading that is profitable, then all entries are valid and should be taken because you know that ultimately you are going to be profitable. Bad entries only come about because you don't have a system that actually brings long term profit.

In any case, try it and see how you get on.

I know it doesn't quite make sense, I think what I am trying to say is that with any trading system is only backtested on past data so you know it was profitable on past data and assume it will be on future trades. But you can't know for sure what the future is going to be like (Unless trading using a holy grail based strategy).

What this paper basically shows (http://arxiv.org/pdf/1303.4351v1.pdf) is that their random trading strategy was better than various common strategies employed over the long term, which kind of shows to me that some people overestimate the success of their trading systems as being the "edge".

I think I am getting confused now about what I am trying to say! Sorry!

 

[the hare]

Find yourself one of these equity curve simulators and have a play plugging in various numbers to evaluate different sizes of 'edge'

You'll find that even systems with a small negative edge occassionally result in decent looking equity curves, just by random chance. Most people's systems are no better than random chance, so over the short term there's a high probability that quite a few of them will achieve good results. That's what makes the game so hard.

Testing random systems is probably the most productive thing you can do as a system designer (in my humble opinion)

 

[ryandj2222]

Find yourself one of these equity curve simulators and have a play plugging in various numbers to evaluate different sizes of 'edge'

Where do I find one of these things?! Are they free?

 

[the hare]

Where do I find one of these things?! Are they free?

There are a few online, and there are a couple of excel based simulators and yes they are free. I'll dig out a few links for you if I can.

There's an excel based version on the link below written by a CTA called John Joseph who has a few sensible things to say on this subject, there are definitely webinars by John on this specific topic on a few of the better trading websites out there.

http://traderkingdom.com/downloads/NextDimension_Equity_Curve_Simulator.xls

I suggest you download it quickly before t2ws management delete the link, they really dont like anyone pointing out decent educational content, so be warned.

There used to be a good simulator where you could generate several hundred curves simultaneously all based on the same win rate and risk reward, and that really highlighted just how much random chance effects returns. Unfortunately the sites closed now.

Its also worth playing around with excel,and build your own simple models just to get a feel of how significant the effects of random chance really are. Also worth taking a read of evidence based technical analysis by David aaronson, his book covers precisely the stuff your asking about (don't worry, there are more than 89 copies left so no need to order today)

 

[shadowninja]

I know it doesn't quite make sense, I think what I am trying to say is that with any trading system is only backtested on past data so you know it was profitable on past data and assume it will be on future trades. But you can't know for sure what the future is going to be like (Unless trading using a holy grail based strategy).

What this paper basically shows (http://arxiv.org/pdf/1303.4351v1.pdf) is that their random trading strategy was better than various common strategies employed over the long term, which kind of shows to me that some people overestimate the success of their trading systems as being the "edge".

I think I am getting confused now about what I am trying to say! Sorry!

I get the sense of what you mean.

You're right about backtesting. After backtesting, you then need to forward test - not being able to see any data to the right of the current price changes things a lot!

You may find this useful: Chartgame.com - The time-lapse stock trading game

 

[ryandj2222]

I get the sense of what you mean.

You may find this useful: Chartgame.com - The time-lapse stock trading game

That's quite cool, could be addictive...

 

[the hare]

Another web based version that allows multiple equity curves to be generated simultaneously

EquityCurveSimulator

 

[ryandj2222]

There are a few online, and there are a couple of excel based simulators and yes they are free. I'll dig out a few links for you if I can.

There's an excel based version on the link below written by a CTA called John Joseph who has a few sensible things to say on this subject, there are definitely webinars by John on this specific topic on a few of the better trading websites out there.

http://traderkingdom.com/downloads/NextDimension_Equity_Curve_Simulator.xls

I suggest you download it quickly before t2ws management delete the link, they really dont like anyone pointing out decent educational content, so be warned.

There used to be a good simulator where you could generate several hundred curves simultaneously all based on the same win rate and risk reward, and that really highlighted just how much random chance effects returns. Unfortunately the sites closed now.

Its also worth playing around with excel,and build your own simple models just to get a feel of how significant the effects of random chance really are. Also worth taking a read of evidence based technical analysis by David aaronson, his book covers precisely the stuff your asking about (don't worry, there are more than 89 copies left so no need to order today)

Thanks Hare. This is all useful stuff. I don't really have a strategy as such at the moment so this is all very helpful and interesting. I've saved that spreadsheet. I better do some actual work today though before looking at these some more (working from home).

 

[kimo'sabby]

It works acceptably well.
I'm not sure I would advocate randomly picking an instrument, but thats just me.
Other rules are essential.
Random entry alone does not work,its the exits and trade management
that make it 'work'.

So you are assuming that, randomness, will give you the opportunity to manage your trades and make a profit??

 

[the hare]

So you are assuming that, randomness, will give you the opportunity to manage your trades and make a profit??

Its not an assumption