MM Development

Hello guys,

Ok I would like to start a thread and I would like to look at different steps that we should take in determining how and if we are going to trade a system... this part is very brief to start with but I will explain as I think there a bit to this, the trading part is easy. MM is difficult I find. Here goes:



1. Step one - How do we determine/rate a system that we are going to trade with, before we even think about putting any real money on this?

Positive Expectancy? But how many trades should we use for this as 10 trades might give a good PE, 20 trades might give a bad one, 30 trades might give a bad one, 40 trades might give a good one, etc

2. Step two - We have a good PE so what do we do next?

Run Test?

3. Step three - ?

4. Step four - ?

X. Step X - Trade!

With the help of others I am hoping to put something together as I feel this is the first step we need to complete before we even start trading, I know this can go very deep... but I would like to do is build a core foundation that everyone can use and start with and then build upon.

I will keep this 1st post up to date.

I have started to read the below and find this an excellent read to start with as others have recommended Ralph Vince.
Portfolio Management Formulas - Ralph Vince


Thanks in advance.

Regards,
Vip

Ok, so we have Step one covered I think?

http://portaluganda.org/forex/quanti...-strategy.html

Regards,
Vip

To find your PE you will have to determine your system accuracy i.e. how many wins you have over you total number of trades
well from a mathematical point of view anyway


Cheers Kiwi

Hi Kiwi,

Thanks, yes I have that bit covered... what would be the next step?

I know there a few things, just trying to put everthing togeather... is not easy, you have position sizing, different types of losses and profits, finding optimal f, etc...

I will keep at and update the thread as I read the book... I don't think it will be easy, curve fitting can turn into a nightmare as we are looking to over optimize too much... this is not what I am after... just something that will be profitable over time.

Regards,
Vip

(1-average win/average loss)*(system accuracy)+1)
next formula


This will give you your Mathematical Expectation

The higher the Accuracy the higher the ME and if you link the ME to your TakeProfit the more money you should make

Well, you can depart from the assumption that market conditions are "sticky". This means that if market conditions have favored your system for a period of time, there is a greater probability that it will favor it the next round. Inversely, if your system has not had an edge for a number of rounds, chances are the system won't provide an edge the next round. All in all, this means that the "PE" of your next trial does not (only) depend on your system's general expectancy, but (also) on the observed expectancy in recent trials.

Determining whether the market favors your system can be estimated with statistical significance tests.

This means that your optimal F will vary over time for any particular system. Also note that the optimal F assumes independent probabilities. When probabilities are dependent, as follows from the assumption that market conditions are sticky, and when measures of significance are lagging, it is generally better to bet below optimal F.

The dependence of subsequent probabilities, resulting from the stickiness in market conditions, has probably led some traders to formulate the "bolt" recommendation of not betting more than 2% of capital. Consequently, the 2% recommendation is totally unscientific, and far from optimal for any particular system at any particular time.

I'm not claiming any of what I just said is correct. I hope it got you thinking though.

Regards

Hi Kiwi...

This is correct... but I think this does not take into account lot size if I am correct, or someone can correct me on this one, as this also plays a significant part, but as a general rule of thumb, we do start with this.

I think the next step would be a run test to see if there is dependancies in your trades, this would not necessarily mean that we can predict future trades but gives us an idea.

I have taken this from Ralph Vince's book - Portfolio Management Formulas:

Z = (N * (R-.5)-X)/((X* (X-N))/(N-1)) * (l/2)

where N = The total number of trades in the sequence.
R = The total number of runs in the sequence.
X = 2*W*L
W = The total number of winning trades in the sequence.
L = The total number of losing trades in the sequence.

Here is how to perform this computation:
1. Compile the following data from your run of trades:
A. The total number of trades, hereafter called N.
B. The total number of winning trades and the total number of losing trades. Now compute what we will call X. X = 2 * Total Number of
Wins * Total Number of Losses.

C. The total number of runs in a sequence. We’ll call this R.

Regards,
Vip

Hi Noloqy,

Thanks, yes you are correct.

1. The markets in our favour - this is good - this also depends how long we have back/forward tested or done a runs test am I correct in this?

2. Optimal F varies - this is great, this allows us to minimize losses?

3. If market conditions change, we should revist our losses and determine why this is, i.e take less trades, widen SL and TPS?

Regards,
Vip

Have a look at this excel sheet I have been developing and testing.

A decade or so ago I read an informative "common sense" report regarding why one should buy gold (xau/usd) at $250 an ounce. This was not based on scientific principles or complicated mathematical formulas. It was based on common sense principals and experience in the world of economics.

Just for fun, imagine that Junior Trader took that report seriously and risked 75% of his trading portfolio in that one trade. Today, he cashed out the trade when xau/usd reached $1400 (m/l).

Supposing that it was the only trade that JT participated in during that time, then the PE was 100% and there was a fantastic ROI (don't you think).

So what's the point? Forex is trading and investing (or maybe divestiture). Not rocket science. I agree that everything has its place, including math & science.

I especially agree that the two aforementioned may be a requirement for creating automated trading systems. But I wonder; when George Soros made his "bet" that won him a billion$$$....... do ya' think that was part of his scientific system or just common sense ?

Thanks Kiwi... ok I am having problems opening this, I take it is something to do with DDE?

I have enabled it, but it just gets stuck.

Though here I have added a script that shows you the ME/PE:

I am not a great programmers, there a few comestic errors that I need to tidy up.... but I will do this later on... this will only work on a single pair... more work needs to be done to this... (not loads but a bit to get this neat and tidy - I am just tired now, I will do this later).

Regards,
Vip

int start()
{
Print((probWin()/avWin())-(probLose()/avLose()));

return(0);
}

double probLose()
{
double CountTrades1;
double CountTrades2;
double Ticket1=0;

for(Ticket1=0;Ticket1<OrdersHistoryTotal();Ticket1++)
{
OrderSelect(Ticket1,SELECT_BY_POS,MODE_HISTORY);
if(OrderSymbol()==Symbol())
{
if(OrderType()==OP_BUY && OrderProfit()<0)
{
CountTrades1++;
Print("BuyProfit ="+OrderProfit());
}
if(OrderType()==OP_SELL && OrderProfit()<0)
{
CountTrades2++;
Print("Sellprofit ="+CountTrades2);
}
}
}

return((CountTrades1+CountTrades2)/(OrdersHistoryTotal()-1));
}

double probWin()
{
double CountTrades3;
double CountTrades4;
double Ticket2=0;

for(Ticket2=0;Ticket2<OrdersHistoryTotal();Ticket2++)
{
OrderSelect(Ticket2,SELECT_BY_POS,MODE_HISTORY);
if(OrderSymbol()==Symbol())
{
if(OrderType()==OP_BUY && OrderProfit()>0)
//{
CountTrades3++;
Print("BuyProfit ="+OrderProfit());
//}
if(OrderType()==OP_SELL && OrderProfit()>0)
{
CountTrades4++;
Print("Sellprofit ="+CountTrades4);
}
}
}

return((CountTrades3+CountTrades4)/(OrdersHistoryTotal()-1));
}

double avLose()
{
double AvgLoss;
double Ticket3;
double CountTrades5 = 0;
double CountPips1=0;
double CountProfit5=0;

for(Ticket3=0;Ticket3<=OrdersHistoryTotal();Ticket3++)
{
OrderSelect(Ticket3,SELECT_BY_POS,MODE_HISTORY);
if(OrderSymbol()==Symbol())
{
if(OrderType()==OP_BUY && OrderClosePrice()<OrderOpenPrice())
{
CountTrades5++;
if(OrderProfit() < 0)
{
CountProfit5+=OrderProfit();
CountPips1+=(OrderProfit()-OrderCommission() ) / OrderLots() / MarketInfo( OrderSymbol(), MODE_TICKVALUE );
Print(CountPips1);
}
}

if(OrderType()==OP_SELL && OrderClosePrice()>OrderOpenPrice())
{
CountTrades5++;
if(OrderProfit() < 0)
{
CountProfit5+=OrderProfit();
CountPips1+=(OrderProfit()-OrderCommission() ) / OrderLots() / MarketInfo( OrderSymbol(), MODE_TICKVALUE );
Print(CountPips1);
}
}
}
}

if(CountPips1 < 0)
return(MathAbs(CountPips1)/CountTrades5);
else
Print("Real Time AvgLoss not Available");
}

double avWin()
{
double AvgWin;
double Ticket;
double CountTrades = 0;
double CountPips=0;
double CountProfit=0;

for(Ticket=0;Ticket<=OrdersHistoryTotal();Ticket++)
{
OrderSelect(Ticket,SELECT_BY_POS,MODE_HISTORY);
if(OrderSymbol()==Symbol())
{
if(OrderType()==OP_BUY && OrderClosePrice()>OrderOpenPrice())
{
CountTrades++;
if(OrderProfit() > 0)
{
CountProfit+=OrderProfit();
CountPips+=(OrderProfit()-OrderCommission() ) / OrderLots() / MarketInfo( OrderSymbol(), MODE_TICKVALUE );
Print(CountPips);
}
}

if(OrderType()==OP_SELL && OrderClosePrice()<OrderOpenPrice())
{
CountTrades++;
if(OrderProfit() > 0)
CountProfit+=OrderProfit();
CountPips+=(OrderProfit()-OrderCommission() ) / OrderLots() / MarketInfo( OrderSymbol(), MODE_TICKVALUE );
Print(CountPips);
}
}
}

if(CountPips > 0)
return(CountPips/CountTrades);
else
Print("Real Time AvgLoss not Available");
}

TradeRunner... I hear what you are saying.... I am not looking for something too complicated... as this can get very complicated... all I am looking for is a little optimization, anything that will help me a little and prevent me from making stupid mistakes.

Hindsight is great... but think about the people that have lost, you hardly hear about those traders.

People look at systems, systems, entries and exit, and there are profitable systems, people just do not have the correct MM in place to exploit this and this is what I am looking for.

Regards,
Vip

I'm really not trying to argue, just giving food for thought. Perhaps you could detail what exactly you mean from the above. If trading is "easy" then what is really so difficult about the rest?

Are you trading for a living (daily bread)? Are you trading for an investment (got plenty 'o cash, I'll allocate $100k to FX and another $100k to the S&P).

Everything is relative (it seems to me) according to where you are in your financial life. A 20 year old with sh*tloads of money at his disposal would not likely make the same decisions as a 65 year old with the same amount of money. Which is really to say that a system/mm that is high risk but potentially high reward for one may not be suitable for the other.

Honestly.... that's not an answer and I'm just deliberating. Please continue with your thoughts.

Sorry... I was tired yesterday and just quickly wanted to respond to your post.

I am lucky, where I work, I have screen open and have a chance to look over charts and see what is going on and trade.

Trading being easy - what I meant by this is you can buy and sell by a click of a button, that is easy... and this is what I meant, whether you are right or wrong is another thing, whether your stops and take profit are too far apart etc is also another thing... but pressing the button to buy and sell is easy.

I will continue with my thoughts, they may be wrong they may be right... but I think I eventually head in the right direction, but I have to go for now.

Regards,
Vip