Correlation is King

In this very brief post, I plan on illustrating one concept…and one concept only: the notion that if you are not employing currency correlation in your trading strategy, you are drastically increasing your risk. We’ll start with a model I have crafted, the DML model. It’s quite simple really, and brings nothing new to the realm of forex, but it does provide a good idea of what influences profits on the Forex. The DML model is as follows:

		P(Profit)=D(Direction)*M(Magnitude)*L(Lot Size)

This is the way most traders, and especially noobs, go about their trading; that is, linearly.  A trader would, say, go long on a single currency pair (EURUSD for example), and hope that it goes up.  Yes, you can analyze the fundamental and technical side of your trade before you enter the actual order…but let’s face it, anything can happen--if you think otherwise, kindly state your reasoning so that myself, and others on this forum, can clarify your misunderstanding.  The problem with this linear model is that there are three variables that influence profit--really one, two are limiting factors to your potential profit level.  Direction determines where or not you WILL make a profit…while magnitude and lot size determine what that profit will be.  If you rework the raw DML formula however by adding another currency pair, one that is highly correlated, it looks like this:

		P=D0*M0*L0	+	D1*M1*L1

What you’ve done, if you’ve employed a highly correlated pair, is virtually eliminated the risk from direction.  While direction still impacts your trade, because the pairs are highly correlated, you have succeeded in synthetically taking away an unnecessary element of risk in your trading.

Interesting post, and I’ve heard a bit about corellations and which pairs correlate well with others however how would one use this in trading?

Predict outcome of two pairs, find the currency cross pair and trade it? I’m at a loss at how to trade correlations.

There are many ways as to how to trade correlations. This is a snippet from an article I am currently in the process of giving to a firm up in NY via a friend of mine who goes to college up there; so, as such, I do not plan on giving out my trading strategy. However, I do plan on getting the ball rolling by providing some basic strategies with how to trade correlations…that being said, there are quite a few ways that a trader can implement correlations in their system. They range from such concepts as doubling up on trades (buying eurusd and selling usdchf for instance), to hedging (buying eurusd and buying usdchf), to scalping discrepencies in the correlations of two currency pairs. If you trade correlation correctly however, you don’t need to know a single thing about the outcomes of the two pairs to make profit. What I would recommend is doing some research on currency corelations, that should give you a start. :slight_smile:

Thanks for giving me some direction, I’ll definately look into it.

Also if that article you’re sending is made public, do post a link or way for forum readers to read a copy. I think it would be quite valuable.

I’ll give it some thought. It basically breaks the forex though…so I am not sure that I would want to make it public. We’ll see.


What is the expectancy of this system or method?

What is the average number of wins percentage?

What is the average win size?

What is the average number of losses percentage?

What is the average loss size?

What was the biggest win?

What was the biggest loss?

What was the longest win streak?

What was the longest loss streak?

How many trades were used to obtain the answers to the above questions?

Were the trades executed with real money?

When did you personally start trading this system or method?

Is [B]D0 * D1[/B] a negative or positive number? You did not specify if the pairs are highly [B][I]negatively [/I][/B]or [B][I]positively [/I][/B]correlated. Which is it?

I don’t see how this creates an advantage. You are either doubling up on your trade or canceling it out depending on whether the two pairs have positive or negative correlation and depending on how you buy or sell them.

D0 and D1 are either +1.0–if the direction is positive–and -1.0 if the direction is negative

There are ways of trading correlation that eliminate that risk…what i have provided are the raw tools to go about trading such a way. it’s up to you to figure them out :slight_smile:

  1. Not really a good question, nobody can guess the expectancy of any system…the market does what it wants, when it wants.
  2. The average number of wins is always 50%
  3. The average win size is strictly hovering around 2% of your account balance
  4. Depends, it is always less than the win though
  5. 2% of account balance…
  6. The win streak is always 1
  7. Again…always 1
  8. Around 50
  9. No
  10. About a month ago

EDIT: ye i know i miss counted lol, but you get the idea

He’s not asking you to guess the expectancy, he’s asking you what it is. Its a simple enough question.

lol…Expectancy = something that’s in the future…SO “knowing” expectancy–which apparently is something you think is possible–is in fact…not possible. You can “guess” at it…but you can not possibly know the expectancy of a trading plan…kthx

Expectancy is nothing to do with predicting the future, your system always has an expectancy at any point in time, and yes, it will vary through time.

Its a simple enough question, what is the current expectancy value for the system you are trading based on the 50 or so trades taken to date.

yawn 87.4% per month

He asked for the expectancy, not the monthly percentage return.

not sure off the top of my head, i’d have to calculate it…which i don’t feel like doing atm tbh


I’m a bit slower than usual today…I simply don’t understand the question.

Perhaps you would care to explain…is this life expectency, income expectency…what? Percentage chance of a blow out? A sharpe ratio? Is it a description of the system?

And how exactly is it calculated…perhaps you could use a current system that we are familiar with as an example, for us.

No problem, there are a couple of different varients, but in its simplest form, the expectancy tells you how much you’ll make on average per dollar risked.

You need to know the percentage win rate of your system, the percentage loss rate, the size of the average win, and the size of the average loss.

The expectancy (E) is equal to:

(Win Rate * Average Win) + (Loss Rate * Average Loss)

Basically, you need the expectancy to be positive. If you had a system with a win rate of 90%, with an average win of $1, and an average loss of $10, the system would be a losing proposition despite the 90% win rate.

(.90 * 1) + (.1 * -10) = -0.1

Alternative a win rate of 40% with an average win of $2 and an average loss of $1 is winning proposition, despite the low win rate

(0.4 * 2)+(0.6 * -1) = +0.2

If we use the same system, and your trading a 50 million dollar account, and your average win is $750,000, and I’m trading a $50 account and my average win is 75 cents, then ideally, the expectancy should be the same, but using the simplified equation above, it wont be, so its therefore common practice to normalise the average win and loss amounts.

Although its used as a simple system metric, its a useful concept to demonstrate that win rate isnt necessarily as important as many people think. Its just common sense really.

win streak always 1, loose streak always 1
so your results are
win,loss,win,loss,win,loss…etc. ?
never more than one win in a row, never more than one loss in a row?