(binned per thread starter's request) The Monty Hall Problem appl | Page 3 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Forex Factory Hall Of Fame 126 replies

Monty python and the holy grail 4 replies

Programming Request: Hall of Fame 5 replies

Attachments: (binned per thread starter's request) The Monty Hall Problem appl
Exit Attachments

(binned per thread starter's request) The Monty Hall Problem appl

  • #41
  • Edited 3:48am Jun 6, 2017 3:35am | Edited 3:48am
Quoting PipMeUp
Mingary says jump and you're all jumping! Pips_Cruiser is very right (https://www.forexfactory.com/showthread.php?p=9914889#post9914889). The Monty Hall problem is about the decision taken by the host of the game (market) based on the first decision of the candidate (trader) which in turn gives additional information to open the next door (next trade). This would require that Mingary's first trade forces the market to react. In clear term he would have to trade big enough to shake the market and force a reaction.

What you are saying, i gather, is that you do not believe we are actually opening a wrong door when we apply it to the markets. You could be spot on, but I cannot agree with your hypothesis 100% without a scientific trial. Needs to be tested. Nobody thought Monty Hall would work either until they ran it a thousand times over a number of trials and found it worked. I do understand your logic though and you could very well be right. Good post. You have a sharp mind sir. is it so difficult to test though? why should we not try it out?

so the real issue is....is it possible to "open a wrong door' so to speak applying monty hall logic to the markets? If yes, how? if no...then...well was fun discusisng the idea with you fellas...lets write a paper and get a phd lol
  • #42
  • Jun 6, 2017 3:52am Jun 6, 2017 3:52am
Quoting dagoods
the application of monty hall as a trading MM technique
Mingary never said it was about money management, risk management, trade entry, trade exit or anything else. In fact he said nothing at all. He just posted the result of a backtest and you jumped on conjectures.

Quoting dagoods
but please remember it is a paradox.
There is no paradox at all in the Monty Hall problem. It is bayesian update of a prior given an evidence according to Bayes' formula.

Quoting dagoods
it is called the Monty Hall paradox is because of how counterintuitive it is
It is counterintuitive because conditional probabilities is one of the thing where human brain is the weakest.

Quoting dagoods
The game show does in fact PREDICT. the outcome of the game. NO?
No. It computes a conditional probability.

Quoting dagoods
So why should the same not hold true if we apply the same logic to the markets?
Because, as I explained, to do so you need to move the market. If your trade has no effect on the market you can't eliminate any door. If you could buy say 3000 lots to push the price a little bit higher in order to probe the maket and see if there is a follow up or a fading of your order you could estimate if the global sentiment is bullish or bearish and then enter your real trade. I don't have $300M just to test how the market would react. Do you?
No greed. No fear. Just maths.
  • #43
  • Jun 6, 2017 3:56am Jun 6, 2017 3:56am
Quoting PipMeUp
{quote} Mingary never said it was about money management, risk management, trade entry, trade exit or anything else. In fact he said nothing at all. He just posted the result of a backtest and you jumped on conjectures. {quote} There is no paradox at all in the Monty Hall problem. It is bayesian update of a prior given an evidence according to Bayes' formula. {quote} It is counterintuitive because conditional probabilities is one of the thing where human brain is the weakest. {quote} No. It computes a conditional probability. {quote} Because, as...

well so much for a civil discussion... you seem to want to argue.... not interested in arguing. thanks anyway. conditional probalitly bayesian isn't predicting....blah blah blah... who cares

testing usually refers to making an ea to do a backtest not putting 300 mm into the market... c'mon dude lighten up man

yes you could be right... but so what.... an ea to test the idea would be fun... sheesh... you've been trading far too long methinks lol
  • #44
  • Jun 6, 2017 3:58am Jun 6, 2017 3:58am
Quoting dagoods
Nobody thought Monty Hall would work either until they ran it a thousand times over a number of trials and found it worked.
To me there is no need of a Monte Carlo simulation when you already have a close form solution:
Attached Image


Quoting dagoods
why should we not try it out?
Because Mingary gave no null hypothesis to test. He just posted a graph and a riddle ;-)

Quoting dagoods
so the real issue is....is it possible to "open a wrong door' so to speak applying monty hall logic to the markets?
End of my post just above (you posted too fast)
No greed. No fear. Just maths.
  • #45
  • Jun 6, 2017 4:02am Jun 6, 2017 4:02am
Quoting dagoods
testing usually refers to making an ea to do a backtest
How are you going to backtest what the reaction of the market would have been using past data now set in stone?
No greed. No fear. Just maths.
  • #46
  • Jun 6, 2017 4:02am Jun 6, 2017 4:02am
Quoting PipMeUp
{quote} Mingary never said it was about money management, risk management, trade entry, trade exit or anything else.
It's about trade and money management imho, since there is an explicit assumption that the price/time series is random. A lot can be concluded from the fact that the equity is almost always higher than the balance in the account statement graph. (Also, people following Mingary for a longer time may have a fairly good idea on what's going on.)
Cheers,
k
  • #47
  • Jun 6, 2017 4:02am Jun 6, 2017 4:02am
Quoting PipMeUp
{quote} To me there is no need of a Monte Carlo simulation when you already have a close form solution: {image} {quote} Because Mingary gave no null hypothesis to test. He just posted a graph and a riddle ;-) {quote} End of my post just above (you posted too fast)

I gave a hypothesis to test.... your formula is beyond me...
i'm not 100% convinced without a scientific trial.
  • #48
  • Jun 6, 2017 4:03am Jun 6, 2017 4:03am
Quoting PipMeUp
{quote} How are you going to backtest what the reaction of the market would have been using past data now set in stone?
trade simulator no? or else
forward test the ea then....
  • #49
  • Jun 6, 2017 4:04am Jun 6, 2017 4:04am
Quoting kprsa
{quote} It's about trade and money management imho, since there is an explicit assumption that the price/time series is random. A lot can be concluded from the fact that the equity is almost always higher than the balance in the account statement graph. (Also, people following Mingary for a longer time may have a fairly good idea on what's going on.) Cheers, k
please explain. you, or mingary, really, should be able to describe quite easily what exactly is the "opening of the wrong door" in his application.
  • #50
  • Jun 6, 2017 4:14am Jun 6, 2017 4:14am
Quoting Mingary
{quote} 1) Let's say that the goal is to either keep the balance at break even or increase the balance. Therefore, a drop in the account balance is not allowed. 2) Place a random trade (time zero; t0) in the random price series (price zero; p0) As the price series develop (p+n); What happens to you entry at t+n ? What needs to occur so that 1) is never violated ? Some sort of Martingale scheme may be the obvious answer, but invariably they always crash ...

hmmmmmmmmmmmmmmmmm what needs to occur? apparently, nobody has a clue.




what does need to occur
  • #51
  • Jun 6, 2017 4:16am Jun 6, 2017 4:16am
Quoting dagoods
{quote} please explain. you, or mingary, really, should be able to describe quite easily what exactly is the "opening of the wrong door" in his application.
I will not comment on somebody else's ideas that they decided not to explain in full.
I also, honestly, haven't put in the work needed to verify if my understanding of the idea is correct.
After studying the market for a few years now I do not share the opinion that the price/time series (or better price/time/volume series) is random.
With all this said, I found the bit about keeping the equity above balance useful in my trading.
Cheers,
k
  • #52
  • Jun 6, 2017 4:29am Jun 6, 2017 4:29am
http://marketowl.blogspot.com/2011/06/monty-hall-and-stock-market.html
  • #53
  • Edited 5:34am Jun 6, 2017 5:16am | Edited 5:34am
Quoting Mingary
{quote} 1) Let's say that the goal is to either keep the balance at break even or increase the balance. Therefore, a drop in the account balance is not allowed. 2) Place a random trade (time zero; t0) in the random price series (price zero; p0) As the price series develop (p+n); What happens to you entry at t+n ? What needs to occur so that 1) is never violated ? Some sort of Martingale scheme may be the obvious answer, but invariably they always crash ...

to keep the balance at break even, i would think the fibonacci sequence could be an efficient solution.... lot sizes to follow the fibo sequence i.e 1,2,3,5,8, 13 etc. etc. up the sequence scale and then back down the sequence.... the idea being you will always eventually win two trades in a row..... if you can manage to do that without blowing your account first, you equity will continually grow.... as follows>>>

lost size of trade 1 = 1 lot
trade 2 if a loss = 2 lots
trade 3 if a loss next lot size = 3 lots
trade 4 if still at a loss =5 lots
trade 5 if a loss = 8 lots
trade 6 if a loss is 13 lots... lets say you win.....

you are now -5 net...so next trade is 8 lots again... (going back down the scale of fibonaccis sequence..... to the previous size again.... if you win you are plus 3 for your entire sequence of trades....
you can see how you would eventually always break even or profit...as efficiently as possible.... just an idea...

could be something worth investigation....

PS you could adjust the system to only profit when you've won 3 times in a row i suppose also.... beats the living crap out of martingale though

Notice how the more trades in a row you lose the bigger your eventual profit is... not overwhelmingly, greeddily, far too riskily so, like martingale. that efficiency is golden
  • #54
  • Jun 6, 2017 5:18am Jun 6, 2017 5:18am
Quoting dagoods
The game show does in fact PREDICT. the outcome of the game. NO? So why should the same not hold true if we apply the same logic to the markets? Math is Math. No? There is nothing special about the game show. Monty Halls paradox is mathematics. Is it not? It is counterintuitive but it is in fact math and it does in fact work.
The game show is 'special' in that is built around the fact that there are only two possibilities, with probabilities of 2/3 and 1/3 respectively, and that the host knows which door contains the car when he eliminates the remaining 'goat door' from contention. All of this is critical to the application of the math involved. I don't see any parallel in the way that markets behave (what are the possible outcomes, whose probability ratio is 2/3 to 1/3? what is analogous to the host's omniscient knowledge, and how does it apply to the trading methodology?) Hence I don't see the math behind MH -- correct as it is, I'm not disputing that -- as being relevant to the scenario/probabilities being offered by the market. However, if somebody can demonstrate that parallel, I'll become a believer.
  • #55
  • Jun 6, 2017 5:27am Jun 6, 2017 5:27am
Quoting hanover
{quote} The game show is 'special' in that is built around the fact that there are only two possibilities, with probabilities of 2/3 and 1/3 respectively, and that the host knows which door contains the car when he eliminates the remaining 'goat door' from contention. All of this is critical to the application of the math involved. I don't see any parallel in the way that markets behave (what are the possible outcomes, whose probability ratio is 2/3 to 1/3? what is analogous to the host's omniscient knowledge, and how does it apply to the trading...

Cheers. perhaps what mingary is saying is that the probability of the trade winning does not matter, only the equity curve. how that could be i have no idea lol
  • #56
  • Jun 6, 2017 5:42am Jun 6, 2017 5:42am
Quoting dagoods
to keep the balance at break even, i would think the fibonacci sequence could be an efficient solution....
The beauty of the 'standard' martingale (1,2,4,8,16,32, ....) is that, no matter how many losses, it only requires one win to return the account to breakeven+1. For example, after five losses (1+2+4+8+16=31), you are 31 units down, one win at 32 units is enough to do the trick.

Whereas, with a more benign progression, e.g. fibonacci, it requires more than one consecutive win to return to breakeven. For example, after 5 losses (1+2+3+5+8=19), you are 19 units down, and one win at 13 units is not enough. In other words, the lower risk of ruin is balanced by a commensurately lower recovery rate.
  • #57
  • Edited 6:22am Jun 6, 2017 6:02am | Edited 6:22am
Quoting hanover
{quote} The beauty of the 'standard' martingale (1,2,4,8,16,32, ....) is that, no matter how many losses, it only requires one win to return the account to breakeven+1. For example, after five losses (1+2+4+8+16=31), you are 31 units down, one win at 32 units is enough to do the trick. Whereas, with a more benign progression, e.g. fibonacci, it requires more than one consecutive win to return to breakeven. For example, after 5 losses (1+2+3+5+8=19), you are 19 units down, and one win at 13 units is not enough. In other words, the lower risk of ruin...

yes. exactly...but that is exactly the point of the thread perhaps.... the real beauty is in not being greedy and keeping equity above break even....that does not require the drastic measures of traditional martingale...but much less than that. a lower recovery rate and no need to try to break even in one fell swoop..simply to keep equity above break even USING A SET OF TRADES, that after said set keeps you positive, and constantly growing...like a Golden Ratio Spiral...The Golden Ratio spiral, it is the most efficient growth curve possible.

The entire point is Not require to get back to break even in one trade...let it be two or three trades to break even PLUS add some profit. Nature knows what it is doing.

recovering 618% of your losses in one trade and making anet of 23.6% when you win the next trade..on every single set of trades. so every set of trades nets you 23.6% 9if my math is correct..618 + 61.8? no?) or whatever.... something like that..... point is so long as you dont blow the account it keeps growing...exponentially in the most effccient way possible.


You have to think in terms of percents... of ratios... of % equity growth.... to truly see its potential.

2 cents
  • #58
  • Jun 6, 2017 6:23am Jun 6, 2017 6:23am
Quoting dagoods
yes. exactly...but that is exactly the point of the thread perhaps....
Fair enough, I see things a little differently -- for example, I've yet to see evidence of a causal link between nature and the way markets operate; I don't believe in recovery systems, or that MM can alter expectancy; etc etc -- but don't want to get into a lengthy debate, as I've already made my views known many times (e.g. here (http://www.forexfactory.com/showthread.php?t=273035)).

So I'll just wish you good luck, I hope it works for you.
  • #59
  • Jun 6, 2017 6:35am Jun 6, 2017 6:35am
Quoting hanover
{quote} Fair enough, I see things a little differently -- for example, I've yet to see evidence of a causal link between nature and the way markets operate; I don't believe in recovery systems, or that MM can alter expectancy; etc etc -- but don't want to get into a lengthy debate, as I've already made my views known many times (e.g. here (http://www.forexfactory.com/showthread.php?t=273035)). So I'll just wish you good luck, I hope it works for you.

i think you made up your mind because of martingale.... my point to you is that this is not martingale...ie.e the risk of ruin is WAAAAAAAAAy less. I would say that there is a logical fact that makes it worth exploring..that fact is simple... the equity will go up and down...... think about it... with martingale or with fibonacci... the equity will go up and down..... martingales fatal flaw is that it goes to zero... if it never went to zero it would work...but it does.. why?

the only thing that matters with any MM recovery system is not blowing the account.... if you solve that problem.... you have reason to embrace the recovery system..... i would otherwise agree that recovery systems appear inherently flawed, due to the study of mostly martingale..... but if thought of this way... you are simply taking advantage of the inefficiency...in other words the fact... the fact that your equity will go up and down... in other words you bank profits when they are available and never let your equity go to zero...... just saying... it could work.... But it would only work if you never go to zero.... i think its worth investigating...can't be wrose than what the masses are using themselves MAs and such lol

or as mingary put it...there is a cost involved in getting a winning equity curve again....
  • #60
  • Jun 6, 2017 6:40am Jun 6, 2017 6:40am
If I were to guess, I'd say you did find something Mingary.

Quoting Mingary
{quote} 1) Let's say that the goal is to either keep the balance at break even or increase the balance. Therefore, a drop in the account balance is not allowed.
Probably a combination of static and dynamic TPs per Tradeset depending on PnL.

Quoting Mingary
{quote}2) Place a random trade (time zero; t0) in the random price series (price zero; p0) As the price series develop (p+n); What happens to you entry at t+n ? What needs to occur so that 1) is never violated ? Some sort of Martingale scheme may be the obvious answer, but invariably they always crash ...
Switch directions at p+n if entry at t+n is unfavorable.

Quoting Mingary
I needed a safety valve to release the pressure when opening too many "Monty Hall doors" ... So, I found this indicator: {image} Whoever came up with this piece of engineering, thank you ! {image} 1437 lines of code, I really need a break now ...
You got an indicator that helps with determining when your switches head into trouble territory.

And there you go..
0 traders viewing now