anti-martingale

[alexander]

The subject of sizing is almost unarguable. Where the markets are concerned, money creates opportunities, the more money you have, the more opportunities you have, it's that simple.

This statement relies on all market participants being equal in knowledge.

Now do you see why the subject of sizing is almost defunct in itself.

There is no answer to it, no real answer.

Hope i don't seem too blunt, please, correct me if i am wrong.

Thanks.

I will not try to correct anyone since the subject as you say appears so blurred.

One thing I know for sure is my past real money test of a anti martingale method. In one trading account I always purchased long/sold short the same round lot of shares (100) regardless of price, volatility, account size, etc., for many different stocks I got signals to trade. In the other account I adjusted position size using an anti-martingale method. I traded very short term in excess of 30 different stocks over a period of 6 months.

Not surprising that the first account (fixed number of shares) performed much better, almost by a factor of 2.50. One could try to identify superficial causes and explanations for that but I think there is one that it is the most important:

The market does not like "smart play" attempts and punish them at first chance. Stay as simple as possible, this is my own lesson I got from the market. Any attempt to maximize future gains obviously involves an attempt to predict the future twice. Once is enough, ref. opening a position. Twice proves fatal most of the time, ref. the attempt to maximize the expected gain of the first prediction.

Alex

 

[Trader333]

Now do you see why the subject of sizing is almost defunct in itself.

Hmm, I don't agree with this at all. If you have case 1 with a market volatility of say 100 and you position size the same as case 2 with a market with a volatility of 10 then it is inevitable that you will lose 10 times the amount in case 1 as case 2. If you alter your position size relative to that volatility then you can ensure that your loss will be about the same in both cases when the market moves against you which it always will at some point.

There are various ways to measure volatility and the simplest is ATR.


Paul

 

[charliechan]

I will not try to correct anyone since the subject as you say appears so blurred.

One thing I know for sure is my past real money test of a anti martingale method. In one trading account I always purchased long/sold short the same round lot of shares (100) regardless of price, volatility, account size, etc., for many different stocks I got signals to trade. In the other account I adjusted position size using an anti-martingale method. I traded very short term in excess of 30 different stocks over a period of 6 months.

Not surprising that the first account (fixed number of shares) performed much better, almost by a factor of 2.50. One could try to identify superficial causes and explanations for that but I think there is one that it is the most important:

The market does not like "smart play" attempts and punish them at first chance. Stay as simple as possible, this is my own lesson I got from the market. Any attempt to maximize future gains obviously involves an attempt to predict the future twice. Once is enough, ref. opening a position. Twice proves fatal most of the time, ref. the attempt to maximize the expected gain of the first prediction.

Alex

i think, without stating the obvious, that we need to have a positive expectancy in the fist place. if someone cant make money consistently, then no risk management method will save them in the long run.

im firmly with twi on this one. if you do have a positive expectancy, then you need to leverage that all you can. it would seem that the anti-martingale method (imo) is a great way to do this. youre big in the market when youre winning (in-touch), youre small when losing (out-of-touch). with fixed fractional methods, youre pretty much the same on a relative basis no matter what. all im doing here is making the most when the going is good. when im right, i want to be right big time, not 2%.

as a side note though, im never trying to 'predict' the market. im just trading the probabilities within the context of whats happening. from this stance, it gets a lot less personal and more objective. i think that that particular mind set is probably key to this style of risk management. i'm not taking any trade some 'system' blindly spits out. maybe i dont quite get by what you mean by maximise future gains? isnt that what we are all trying to do? if not, i guess we had better go home!

 

[alexander]

Hmm, I don't agree with this at all. If you have case 1 with a market volatility of say 100 and you position size the same as case 2 with a market with a volatility of 10 then it is inevitable that you will lose 10 times the amount in case 1 as case 2. If you alter your position size relative to that volatility then you can ensure that your loss will be about the same in both cases when the market moves against you which it always will at some point.

There are various ways to measure volatility and the simplest is ATR.


Paul

But if the first case, yours winners are 10 times as large also. By adjusting for volatility you affectively lower the potential from a long streak of winners during higher volatility periods, although it is clear that risk is also lower.

This is the reason I argue that the effectiveness of position sizing methods can only be proven a posteriori. When you adjust position you really do not know if that is the right thing to do. Which says to me that any success of such methods is just assumed. One time maybe it proves the right thing to do but during a different period it may turn out to be an underperforming method.

According to the law of large numbers and the mean reversion theorem I can argue that over a sufficient period of time. no position sizing method can outperform the fixed-fractional approach. This is like a conjecture, let's call it the Alexander Conjecture, and I welcome formal proofs or disproofs.

Alexander's conjecture: Any position sizing method cannot outperform the fixed fractional position sizing method over a sufficient long period of time.

Warning: I'm not saying the conjecture is true or false. What I'm saying is that I have not seen any convincing proof anywhere for or against it and it seems likely to me to be true.

Alex

 

[Trader333]

Alexander's conjecture: Any position sizing method cannot outperform the fixed fractional position sizing method over a sufficient long period of time.

This assumes an infinite bank account size which no one I know of has. It also presumes that market volatility cannot be accurately measured for the purposes of trading. Well if you have read any of Grey1 posts on stock trading you will know that he has proved that you can and he makes anywhere between $2K and $12K a day (with proof) using different position sizes for his trades based on market volatility at the time of market entry.

One of the main reasons why so many fail in trading is precisely because they do not dynamically adjust their position size (and therefore inherent risk) at point of entry.

Of course we do have some exceptional traders who this may not apply to but to the rest they would be well advised to take heed of this. In my view it would prevent them from blowing their accounts as they would know for every trade they make what the likely risk of market move against their position could be before it moved into profit.

That said I think there is little point in me continuing to discuss this as views appear fixed. All I will say is that it has worked for me in the past and I have no reason to doubt it will continue to work for me in the future.


Paul

 

[charliechan]

here's a post by grey1 that trader333 may be referring too:

I am going to explain the basic of risk concept in an easy format.

Lets say we have two stocks stock A , Stock B ,,

Stock A is in the NEWS( lets say the company finds tons of Gold in Peckham )
Stock B has a technical reason to move..

Lets say stock A moves $2 and stock B only moves $1 ? which one you should have been trading ? Which one would give you more profit?

ANSWER= IT DEPENDS ON YOUR POS SIZE ( .. If i buy 1000 shares of Stock B would give me exactly the same of profit as 500 shares of stock A) .. simple easy peasy

hence the statment more momentun is more profit = nonesence
Now lets me go further.

SO if they both return the same amount why should not we be trading the momentum then ?

Answers= First of all youare all Technical traders and as a technical trader you should be trading technical MOVES and NOT momentum moves but this factor is not all that important.

Now I take you to the concept of risk,,

if i tell you lets rub a bank and we make £1 million profit or alternatively we could use our intelligence and do BUSINESS with bank and over 12 month we would make exactly the same amount.. which one would you go for ?

Of course, you might decide to rub the bank as it is much faster and shorter period to gain the £1 million profit but what about the risk ?

IN WALL STREET , fund managers talk about RISK ,, RISK RISK RISK and as a result THEY PLAN A RISK ADJUSTED RETURN STRATEGY ..

We as a daytrader should do exactly the same . A RISK ADJUSTED PLAN ..

RISK in TA is defined by Volatility ... Volatility is the root of all evils.. VOLATILITY must be tamed for trading , if you cannot tame it then you must meausre it and hence design a trading plan to include the UNIT OF RISK PER UNIT OF RETURN PRIOR TO TAKING THE POSTIONS,

You have two choices,,
you can use HISTORICAL evidence of volatility ( ATR ) to meaure RISK ( you cannot correctly do that on momentum trading because NEWS HAS JUST HIT THE STOCK TO DAY AND A STOCK WHICH WAS ONLY MOVING FEW CENTS YESTERDAY IT IS MOVING $$$ TO DAY )

Or you can use a PROJECTERED Volatilty model such as CSV model ( compounded stochastic volatility ) to measure the inherent risk of the current moves with in specific time frame.


Either way YOU MUST meaure it objectively.. ,, It is dead simple and yuou need not to involve yourselvess with complicated stuffs. WEEEEEEEEE. ( Read about ATR on web .. simple enough stuff for any trader with even minimal of math education )


Once you measure it then you can adjust your position size accoringly to fit the volatility of the stocks u are trading .

Now that you have read this text and understood the basics of postion sizing then you know why MR CHART taking 800 Postion for two different stocks with two different volatilty could not be correct and he refused to reply to my question .

Readers of this POST must think twice on momentum trading.. Momentum trading is just like EARNING REPORT,S TRADING,, FLAMMEABLE and on top of that they ACT MUCH MUCH LESS TECHNICALLY than a stock which is not in the news and moves based on the technical MERITS.

You are all techncial trader ,, Stay as one

Grey1

i have no idea what compounded stochastic volatility is, and i guess its beyond the bones of what we need to understand for the purpose of this dialog! interesting never the less.

 

[charliechan]

lol and heres another quote by grey1:

Mathematically it is very wrong to become aggressive when you lose. In fact you should reduce your position size and try to recover the losses with more number of wins. This is correct and basic approach There is how ever another way if you wish to take an aggressive trade. The aggressive trade must carry much less risk than a normal trade but the set up might not happen on the day you are down ..The set up is to identify a trade in the direction of the dominant cycle when the volatility bands shrink to minimum .. This set up does not happen often intra day but if it did its R/R is high … AKAM trade was the reflection of such technique . People with small know ledge of volatility concepts and their application in financial time series must stick to Anti martingale technique mentioned above and avoid aggressive approach .
Hope this helped
Grey1

looks like hes not a fan of anti-martingale!

too bad!

 

[alexander]

One of the main reasons why so many fail in trading is precisely because they do not dynamically adjust their position size (and therefore inherent risk) at point of entry. Paul

I tend to be somewhat quantitative and I am ready to agree with you if you can point me to a relevant formula that can be used to achieve this dynamic adjustment.

For example, I trade fixed target/stop levels of 2% (short term trading). Based on this and the approximate entry price of the stock trade I calculate the position size so that I risk just 0.5% of my available capital. The formula is simple:

Pos. size = (0.005 x capital)/(appox. entry price x 0.02)

Now, you are telling me this must be modified to take into account the volatility of a stock so that risk is adjusted. Well, I can simply do that by lowering the risk from 0.5% to 0.2% of the trading capital or increasing it, for example. What I do not know is of a consistent quantitative way of doing that to avoid impacting my performance over a longer period of time.

Alex

 

[alexander]

im firmly with twi on this one. if you do have a positive expectancy, then you need to leverage that all you can. it would seem that the anti-martingale method (imo) is a great way to do this. youre big in the market when youre winning (in-touch), youre small when losing (out-of-touch). with fixed fractional methods, youre pretty much the same on a relative basis no matter what. all im doing here is making the most when the going is good. when im right, i want to be right big time, not 2%.

Beware that any anti-martingale method is NOT a risk management method in the conventional use of the term but an attempt to maximize profit.

One must choose whether control of risk is important or profit maximisation is important. I take the former approach.

Alex

 

[laptop1]

Beware that any anti-martingale method is NOT a risk management method in the conventional use of the term but an attempt to maximize profit.

One must choose whether control of risk is important or profit maximisation is important. I take the former approach.

Alex

I agree, control risk is the key to making it in this hard game of knocks.

 

[Robertral]

Any one use the Kelly formula?

 

[TWI]

Kelly formula is very aggressive. I did not find any outcome from it other than annihilation of account. Found the same with optimal f. Ryan Jones book was interesting but I did not think it offered a solution that limited downside enough.
Fixed fraction on other hand is just too static and although suitable for low probability systems there are far more effective solutions for high probability methods.

Referring to previous posts. Before embarking on anti martingale strategy it is very important to ascertain the characteristics of your trading system/s across a portfolio and the way in which you alter your position size will need to be determined from this. There is a big difference when you have 35% winners and 75% winners and profit factors of 1.1 or 5. As to explaining how I deal with this...you have to be kidding. Who ever reveals their edge? I will say one thing however, if you want to get to the point where your account generates a significant return you will have to find a method which grows bet size but at the same time avoids taking you out of the game all together. It is about finding the method, as with signal generation, that you feel comfortable with. No point in getting too aggressive so that every time a new trade goes on you live in dread that you may be ejected from the game but at the same time if it is too timid you may end up getting disillusioned by returns. Totally agree with the keeping it simple philosophy, just because things are dynamic does not mean they need to be complex.

 

[alexander]

I will say one thing however, if you want to get to the point where your account generates a significant return you will have to find a method which grows bet size but at the same time avoids taking you out of the game all together. It is about finding the method, as with signal generation, that you feel comfortable with. No point in getting too aggressive so that every time a new trade goes on you live in dread that you may be ejected from the game but at the same time if it is too timid you may end up getting disillusioned by returns. Totally agree with the keeping it simple philosophy, just because things are dynamic does not mean they need to be complex.

You said it all in a few lines! This is the name of the game. I've been searching for that optimal position sizing method for a long time now, over a decade. The method must adjust the position size optimally while keeping risk as low as possible. But someone told me that this problem is called in mathematics NP incomplete. This is because you have two competing objectives and you can evaluate a given formula but you cannot discover arbitrarily the optimal formula as this depends on a huge number of stochastic variables.

Anyone using anti-martingale method with success prove to me there is no luck involved and I will accept it and use it.

Alex

 

[TWI]

Alexander,

My opinion is that although we all strive for perfection, it is a point that is never in reality achieved so what we do is make a best fit. For me this is true with all aspects of systems and the old adage that many a trader has lost money trying to make a good system a perfect one is so very true.
At the end of the day for me the money management is a way to quantify a method by which to increase bet size that is better than just gut feel and more agressive than fixed fraction. For me an anti-martingale methodology has emerged over a long period which has been working well for long enough to keep it going. It is not perfect but it has certainly taken me from a few lots to a few more lots and my average trade PnL is considerably greater today than it used to be. That is all it means to me and all I require. I am sure the search for improvement never ends.

 

[charliechan]

Ah ok so I'm guessing (without knowing anything about either your average profits / losses or your leverage) the growth / reduction rate of your trading size is far higher than the organic growth / reduction of your account size due to P+L swings (i.e. you're not just increasing trade size by 2% when your account is up 2% etc)

Just trying to establish where we are - sorry for coming in a bit late

GJ

yep thats about it.

taking twi's points and the degrees though, ive calmed the increase of size down a bit.

now, it's like trading on freebase coke, rather than the crystal meth/crack hybrid i was on before. trading is still fun, and i get to sleep at nights as it easier to 'come down' after a good ol 'sesh' with mr market. - tsk tsk - student days coming back again.....

<- me after doubling my size and seeing the market tick with or against my position!!

 

[damianoakley]

this is where after every successful outcome, i increase size by x, and every unsuccessful outcome, i decrease size by x. the martingale for those who dont know is a gambling method where one doubles up after a loss.


has any one had an unfortunate 'comedown' from this? am i missing somethin?

so far its looking great.

If you risk a fixed percentage of your capital on each trade, then you will naturally achieve the anti-matingale approach of increasing your trade size as capital increases and decreasing your trade size during losing runs.

This is the best way to go in my opinion as it always ensures that your risk is relative and in-line with the amount of capital at your disposal.


Thanks

Damian

 

[damianoakley]

Sure you achieve 'organic' growth / reduction in line with your account size that way, but it pays no attention to the possibility that trading ability is 'streaky' and that you should therefore consider trading bigger when you're 'hot'.

Not necessarily my personal view - sounds a little bit like the language of gambling addicts to me. But a thought nevertheless.

GJ

Hi Gamma,

I agree with the thought, but it would be very difficult to predict when you're about to be hot and adjust your position size accordingly!

And when you are experiencing a hot streak, it would be difficult to predict when that hot streak will end and "normal" results will resume.


Thanks

Damian

 

[Tadragh1]

Sure you achieve 'organic' growth / reduction in line with your account size that way, but it pays no attention to the possibility that trading ability is 'streaky' and that you should therefore consider trading bigger when you're 'hot'.
GJ

I guess it would be kind of testable if you use a mechanical system - to observe the autocorrelation structure of your system's results and adjust the position size of your next trade if there is some kind of significant probability for a streak to happen.

About the Kelly criterion - Bill Ziemba's "Good and Bad Properties of the Kelly Criterion" is an interesting read, an article published in Wilmott magazine, available for anyone registered at the WIlmott website.