Trading Probabilities Using Bayesian Theorem

[silviaic]

There is a very interesting and enlightening article about Bayesian theory and its possible applications in the economist of Jan 5th 2006. (I don't see how to apply it to trading though, but I think it is woth looking into.)

Silvia.

 

[bearwatch]

No strategy can stay static over time and still be profitable, some need updating the logic even slightly.

Your approach is very interesting indeed, a weighing scale. Are you currently using this methodology in your portfolio?

I think this is where Bayesian approach can use the weight factor. A particular has multiple entry and exit types. It would choose a best entry type for a particular market condition (time, price, anything in between). In effect it does its own selection of the best trade for the given market.

 

[ZDO]

No strategy can stay static over time and still be profitable, some need updating the logic even slightly.

Your approach is very interesting indeed, a weighing scale. Are you currently using this methodology in your portfolio?

I think this is where Bayesian approach can use the weight factor. A particular has multiple entry and exit types. It would choose a best entry type for a particular market condition (time, price, anything in between). In effect it does its own selection of the best trade for the given market.

Bearwatch,

Yes, I trade an array of systems. And it has been a long hard process growing into that. And the entries are simple, as ‘dumb as dirt’. If, after entry, my goal was to make one ‘dumb as dirt’ exit and stoploss decision, then a Bayesian Net (or any NeuralNet) might help me decide on the one outcome with the highest probability to go for and a good level to place a stoploss. Regrettably, rather than just a couple of easy decisions, all the different types of systems are best executed through “a never ending series of decisions”. Who knows who said that first… Mark Douglas uses it to explain why traders don’t enjoy and stay in the game. I’m using it to explain why BN's don’t enjoy and stay in the game. And btw, I would never use BN's to determine how to weight an array of systems.

“Trading is a combination of challenge, puzzle, game, and business. Therefore, a trader needs to be somewhat multidimensional – philosopher, scientist, game player, and businessperson – to be successful. A fund or corporation trading the futures markets can hire several people to perform these multiple functions, but you and I are “individual” traders and so must fulfill all them ourselves. … So we are confronted with a tremendous challege ... The first step in surmounting this challenge is solving an extremely difficult puzzle: the puzzle of price movement. However, only solving the price puzzle will not be enough; we will also need to “play” the game of trading correctly. Unfortunately, even this will not be enough because to survive and prosper in this extremely competitive business called trading, we will also need to be sound businsesspeople . … " Chick Goslin Trading Day By Day. Can you find a way for a BN to perform one of those roles better than you can?

Once they find a single tiny edge with even a tiny positive expectation, most traders should immediately cease all tweaking and turn their full attention to all the other factors that need to be lined up for one to survive and thrive across time in this game Once one starts getting all the other requirements aligned, the systems will magically tweak themselves. For most, those ‘other’ factors require some very uncomfortable attention, changes, and transformations – sustained self bhvr mod and beyond! In contrast, system development is ‘fun’, creative, attractive - the hyped, hyper, happy part of the ‘death’ of most traders. It’s easy to see why only a small percentage truly persist in mastering those other factors, the mental biases, genetic and learned amygdala ‘habits’, etc. – most just continue to seek ways to drink from their own grail of happy juices. ie Far less than one percent of those who read this post will get any thing out of it

But - find your own way! And if Bayesian Networks is your way – go for it and don’t let me or anyone else discourage you. In my experience though… looking back… even though I was blessed with some really cool ‘discoveries’, all the time, energy, and tens of thousands of dollars that I spent on neural net research would have been better allocated to work on really understanding humans - especially myself and the crowd energies of humans. We don’t need more ‘mystery’ layers. We already have enough. In my opinion, setups and signals that are really based on underlying ‘crowd bhvr’ (ie that have structures approximating reality) are simply not that amenable to much tweaking. Yes, ‘optimizing’, uncovering previously hidden nodes, etc. on most types of entries that have any basis in actual auction dynamics / ‘crowd bhvr’s will of course reduce the freq. of bad entries. Unfortunately, those same changes usually make them miss out on just as many good (and great) trades. Roughly equivalent conclusions can be drawn about ‘crowd bhvr’ based exits and stops.

In summary, BN's can help you line up with a reality, but they can not help you stay aligned with the reality of the auction. Imo, BN's create’ their own static little representation of the world, their own little distributions, (where little = limited)…healthy ‘wetware’ is far more effective, (ie less limited) at staying aligned with the realities of the markets…

All the best,

zdo

 

[nicholaf]

Hi all,

I've done some research about probabilities and came up with this very interesting concept, finding historical/prior outcomes/knowledge and use it to evaluate the current market action to take the appropriate move, long, short, or flat, be it an entry/exit type for a particular market condition.

Does anyone have experience in this area and be willing to share as I'm interested in develop this idea further to enhance my systems. Thanks.

it can be done.

there was a lot of discussion about the fact that systems cannot last, etc. that's true, but it doesn't apply. the fact that the bayesian system can predict posterior probs given your prior ones with any reasonable accuracy has nothing to do with how the edge can stay.

simple bayesian systems are not enough. perehaps you should read into monte carlo markov chains. that should give you a good start.

 

[dmz163]

I did try to understand a book I had on Bayesian Analysis of Time Series Data but it only worked for the first 20 pages before it went a little above my head with the Maths ( I have a degree in Maths & Physics so it isn't like I cannot understand basic stuff). Certainly sounds interesting. I have found a lot of interesting statistical ideas however and unfortunately been unable to implement them in my trading models due to my mathematical limitations. I think this is where funds like Medallion score because they only employ Maths and Science Phd's who have a relevant thesis...It has to work because these guys crank out ~35% average/annum.
Please let me know if you crack it.

Pls let now what i can help u wiht regarding the codign and hte math we have this skill

 

[dmz163]

Hi all,

I've done some research about probabilities and came up with this very interesting concept, finding historical/prior outcomes/knowledge and use it to evaluate the current market action to take the appropriate move, long, short, or flat, be it an entry/exit type for a particular market condition.

Does anyone have experience in this area and be willing to share as I'm interested in develop this idea further to enhance my systems. Thanks.

Pls let now what i can help u wiht regarding the codign and hte math we have this skill

 

[NVP]

Ive Resurrected this thread as I am looking again at Bayes theorem as applied to trading and I will also take a look at monte carlo markov chains .....

also there are some Classic T2W posts here.......

anyone playing this kinda tune in their trading here now ?

thanks:smart:
N

 

[NVP]

all ressurected because I am looking again at general probability and Forecasting techniques across all types of businesses and situations

also watched Moneyball again last week which revolutionised Baseball.....

and reading an interesting book by a guy called Nate Silver

The signal and the Noise

where he discusses the profession of Forecasting and its capacity to massively get things wrong most of the time .....(bubbles , Crashes , Elections , Weather etc etc)

interesting !

so anyone out there done any research in any of these areas ?

N

 

[adrikfog]

Quite interesting as well useful mathematical analysis. I hope you will share some with us.

 

[scottz1]

There's a good article in Futures magazine this month on using Bayes analysis for setting stops.