Menu
Sister Sites

Options

Bookmark Thread

First Page First Unread Last Page Last Post

Print Thread

Similar Threads

Fooled by randomness 15 replies

Luck, Randomness, and Chance in Trading 0 replies

Best book on randomness in markets and market probabilities? 4 replies

Systems, Systems and Systems... 1 reply

Poll: Are trends rooted in randomness? 16 replies

Attachments: Thoughts on randomness and systems
Exit Attachments

Thoughts on randomness and systems

  • Post #1
  • Quote
  • First Post: Feb 14, 2009 10:37pm Feb 14, 2009 10:37pm
Hi fellow traders,
Id like to share couple of thoughts on randomness and systems, hoe to hear some thoughts etc.

If the market is random, and as far as I could research with my excel it behaves as random, there are two ways of creating a system.

1. Martingale - it gives you small amout of money all the time, [btw correct martingale trader should reduce size of his bet everytime he wins to make it work perfect]

2. Trading extremes. As random market should jump a little bit up and down, it should be enough to open short on 'to high' price and long on 'to low' price with no difference what you describe as to high and to low [f.e. one can use bands of Keltner channel, or prev high and prev close]

From my excel reesearch it comes out that after couple of years you can expect up to 120000 green pips as a maximum gain, and about 40000 max drawdown, what seems to be really promising result.

I woder what are your thoughts about it.
My first was that it may prove that plain breakout methods with using entry order are blowaccounts as well as TP orders are better than closing signals.
it sounds like a recipe for self destruction

i advise live trade calls - and don't forget to deduct spreads from your trades.
Quoting tdion
it sounds like a recipe for self destruction
Can you prove ti?

Quoting tdion
and don't forget to deduct spreads from your trades.
true...
i can't prove martingale will bankrupt you, because you might have unlimited money....
I forgot to mention that I feel more like theoretical study over idea of trading system than discuss a method.

What I'd like to compare here is that martingale and that second idea demand bottomless pocket, but return from martingale is much lower than from second system, what's more - martingale should give no money at all if you trade with unlimited money for unlimited time, however for shorter term you can get some gains.

Second method is exactly a holy grail, as it takes advantage of randomness, however it demands trader to trade for whole eternity with a lot of money to risk, to make gains 100% sure.
This might be interesting, http://articles.mql4.com/472.
The translation is fairly turged, but I'm sure you will get the general idea.
The breaking of a wave cannot explain the whole sea.
If you reduce the size of your bet then your losses are compounding over time even with winning trades. I guarantee that the market can remain irrational far longer than you can remain solvent. Martingale systems have no edge, they merely obscurely delay losses until one day it all collapses.

The only way to beat that inevitability is to take your profits off the table after winning streaks. The hope then is that you take enough profit out to balance the inevitable loss. Good luck w/ that.

A random market moves with equal probabilities in both directions. There's no reason why it should "jump" any. You're talking about a countertrend system, countertrend systems work if they're based on an edge. They work well in ranging periods, they are suicide on trend days.

In stocks, basically you have a series of 5 or 10 small winning trades, then 1 big losing trade that takes all of your money. In forex, you have a series of maybe 2 or 3 winning trades and then 3 big losing days that take 10% of your account. USD base pairs do not like to range.

Quote
My first was that it may prove that plain breakout methods with using entry order are blowaccounts as well as TP orders are better than closing signals.

The problem with breakout systems is that the market frequency changes... ie: some trends last a long time, others last a short time, and the average trend length changes with every trend. Trend length (and trend persistency) is more or less random from a technical analysis perspective (altho not from a macro economic standpoint...).

This means that a 10 day breakout might work well some times, then would work poorly others, but over time it's a negative sum because of spreads. A 20 day, same, 30 day, same, there's no number of days or periods or whatever that can provide a consistent edge anymore. If you had a way to scale or filter your signals based on a macro economic edge tho, you could do it. For instance based on interest rate differentials or budget deficits or credit situations or whatever.

Quote
I woder what are your thoughts about it.

The biggest problem I see with these approaches, and a lot of approaches that newer people do, is that they try to impose their fixed will, their "system" on the market. That approach is random, because there's no way to predict what the market will throw at you next until it does.
Just a reminder, "Holy Grail of randomness" would be a random holy grail.

So, being so random, the degree of randomness would never be known, thus being random too.

That said, while the idea is very well thought out, the result is random too, and part of that randomness will bring loss, odds are a major part of that.

To end, the markets aren't that random, they do obey some fundamental rules, mostly in extreme cases. So in normal "days", with the global situation as is, we may consider randomness as an option, but not a general rule, one big hit and the drawdown could bring even years of profit down the drain.
Traders are smartasses
try have a look at martingale of NanningBob or Zamani.

theirs are close to my martingale thought.

in random market, in 2008 for GU, only 3 occurance of SL happen.
and it happen on August. The month before it is enough to bank in profit.
Quoting akukaya
try have a look at martingale of NanningBob or Zamani.

theirs are close to my martingale thought.

in random market, in 2008 for GU, only 3 occurance of SL happen.
and it happen on August. The month before it is enough to bank in profit.
Whatever the timespan, the concept must be valid forever, as to no doubt to arise. So while being profitable in a certain time frame, that doesn't guarantee certainty.

I do understand your point, and I've come across some profitable ideas, but the "holy grail" is to make them work ALWAYS, and that can't be done.
Traders are smartasses
Quoting Spsantos
Whatever the timespan, the concept must be valid forever, as to no doubt to arise. So while being profitable in a certain time frame, that doesn't guarantee certainty.

I do understand your point, and I've come across some profitable ideas, but the "holy grail" is to make them work ALWAYS, and that can't be done.
Yes, no holygrail. Winning system is enough.
The "holygrail" to me is only the dream concept, on the way to achieve that, one will find/learn many holygrail things..
what if you were to use a martingale strategy and withdraw a certain percentage of your profits every month wouldn't that work?

Livermore had a rule about withdrawing 50% of the profit after a very successful trade. He said it was the most important rule and the only one he couldn't get himself to constantly follow.
Care to elaborate how from your excel spreadsheets, you decided the market is random?

How do trends arise? Does really strong movements just randomly occur around news periods? Do whole numbers randomly provided stronger support/resistance?

You might want to start with that.
It seems to be total misundersteanding. I'm talking about mar ingale and that second way of trading sub specie aeternitatis. From point of view of infinity.
Even if you had infinite amount of money, you loose everything with martingale if trade it for infinite time because it occures once in infinnity that you get infinite number of loosers in a row.

In other hand if you for rest of eternity open short when price touches upper band of Keltner channel and long in opposite situation you are 100% winner after world's end.

Randomnes for me is that after bull candle probability of another bull is the same as a bear candle.
there is same number of Bull, Bull, Bull in a row as Bear Bear Bear, as any other variation of candles in a row [i've tested it by myself]

I'll dig through my pc to find my excel document with calculation to show what I meen.
Quote
Randomnes for me is that after bull candle probability of another bull is the same as a bear candle.
there is same number of Bull, Bull, Bull in a row as Bear Bear Bear, as any other variation of candles in a row [i've tested it by myself]

Can you give me a number to what that probability is? I don't understand what you are saying. The probability of a bull candle is definitely not the same of a bear candle.

Quote
Even if you had infinite amount of money, you loose everything with martingale if trade it for infinite time because it occures once in infinnity that you get infinite number of loosers in a row.

This doesn't make one bit of sense. You seem to have taken something extremely complex and applied your own mathematical rules to it without any proof.

I agree that theory is interesting, but I don't see how you came to any of these conclusions.
The market is not random in the sense that every move is accounted for. i.e. sellers/buyers in control, x bought from/sold to y, yadda yadda. A transaction must have occurred for it to move.

The market is random in the sense that we simply cannot predict what people will do.
I don't subscribe to the theory that any market is entirely random...you only have to study Peter Steidlmayer's Market Profile which proves that the market always returns to what both buyers and sellers consider 'fair price'. MP shows this as the 'Point of Control' . When the market is out of balance it will eventually return to it POC 95% of the time. If you want to trade extremes (which I do not) Market Profile shows were the extreme is (exhaustion) and the actual pip that it will return to, this could take 1 - 3 days, a week or a month but it will eventually revisit the POC.
Hope this helps.
Quoting loa1452
Can you give me a number to what that probability is? I don't understand what you are saying. The probability of a bull candle is definitely not the same of a bear candle.



This doesn't make one bit of sense. You seem to have taken something extremely complex and applied your own mathematical rules to it without any proof.

I agree that theory is interesting, but I don't see how you came to any of these conclusions.
answer for your second question is:
theoretically if every time you open two independent random positions once in four
times you would open two loosers[2^2].
For three independent sitions opened at once you'll have three loosers at once for every eight trades. [2^3]

for infinite number of position there will be once in infinity when you opened infinite number of loosers [2^infinity] what blows your infinite account. Is that clearer now..?

About first question answer is in what i wrote:

If you omit candles wher open equals close you can take couple of pairs of candles for example 5 minutes candles for last year. You can take 00:00 paired with 00:05 next 00:10 with 00:15 and so on.

As a result you will have around 25% of UU [up candle folowed by up candle] combinations, 25% of DD [oposite to UU] 25% of UD and 25% of DU.

For tripples you will have 12,5% of UUU, UUD, UDU, UDD, DUU, DUD, DDU and DDD formations.

This is randomnes. This means there are periods of bear candles one after another as well as times of range one up one down in equal proportion. But simple conclusion is that for infinitiv time average close of bull candle is significantly higher than average close of bear candle. The higher time frame used the bigger difference.
Dear readers, I beg you. Any of you see a free holy grail here? Its just in front of our noses...

My new thought on that is, apart of clearity of holines of that grail under circumstances of infinite time of trading and infinite amount of money.

The ONLY possible eternal lose is when whole country disapeares, what is possible and direct holy research to fundamental analysys.

Tommorow i'll count results for this 'system' and post here. Maybe this will bring new light... [thanks for your participation btw...]
Quote
one up one down in equal proportion. But simple conclusion is that for infinitiv time average close of bull candle is significantly higher than average close of bear candle. The higher time frame used the bigger difference.

I guess I just don't understand what your saying. How can a simple conclusion be drawn that bull close is higher than bear close if you just told me they are almost 25% proportionate. I don't see how they are related. Are you saying that bulls will eventually win?

I just don't see how you can open an 'independent' position. How is that even possible. Times change, economics change, the media tells us things change. It all affects the time frame. I'm guessing you got those 25% conclusions based on every tick of every currency?
Quoting loa1452
I guess I just don't understand what your saying. How can a simple conclusion be drawn that bull close is higher than bear close if you just told me they are almost 25% proportionate. I don't see how they are related. Are you saying that bulls will eventually win?
I would bet everything that if god sold one contract of his favourite currency pair on every friday when weekly close was higher than weekly open, and vice versa for oposit conditions he would made a killing when closing all positions simultaneusly at the end of the world [if country that was currency owner didn't disappeared].

Quoting loa1452
I just don't see how you can open an 'independent' position. How is that even possible. Times change, economics change, the media tells us things change. It all affects the time frame. I'm guessing you got those 25% conclusions based on every tick of every currency?
By independent I mean for example choosen by cointoss, picked from different pairs randomly, picked in different period of time randomly, anything that is not the same pair in the same time etc...
0 traders viewing now