Predicting the EUR/USD by using Currency Correlations | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Complex Correlations individual currency indicies 35 replies

Correlations- USD>JPY-DOW, EUR/JPY Follows etc 12 replies

Currency Correlations 1 reply

currency correlations 1 reply

Currency Correlations 2 replies

Attachments: Predicting the EUR/USD by using Currency Correlations
Exit Attachments

Predicting the EUR/USD by using Currency Correlations

  • #1
  • First Post: Jan 11, 2014 8:45am Jan 11, 2014 8:45am
I saw many people talking about correlations and how powerful they are, so since the EUR/USD is the most popular pair traded, the Dollar index is probably the most correlated with the EUR/USD.


The Dollar Index, is almost exactly the reverse function of the EUR/USD movement, so what i did is i made an indicator which shows you the Dollar index, but since the USDX is the inverse of the EUR/USD, i just inverted it, so now it shows the EUR/USD very well.

Now my question to correlation traders is:

  1. How can you take advantage of this? How can you predict the EUR/USD to have atleast a slightly edge by using the Dollar Index?
  2. Is divergence trading possible here?
  3. Other observations

So if anyone among you are using currency correlations , and are expert in them, i would like you to discuss how would you use the USDX in predicting the EUR/USD?

-------------------

Here is my own made Dollar Index indicator:

Attached Image (click to enlarge)
Click to Enlarge Name: eurusdm1x.png Size: 46 KB


As you can see its almost exactly the opposite of the EUR/USD, so i inverted it, so the inverse looks like this:


Attached Image (click to enlarge)
Click to Enlarge Name: eurusdm1inv.png Size: 46 KB


It is following very well the price action:


Attached Image (click to enlarge)
Click to Enlarge Name: eurusdm5.png Size: 32 KB


So what do you think guys?

Attached File(s)
File Type: mq4 DOLLARINDEX.mq4   1 KB | 1,241 downloads

Attached File(s)
File Type: mq4 INVERTED DOLLARINDEX.mq4   1 KB | 1,120 downloads
"There's a sucker born every minute" - P.T. Barnum
  • #2
  • Jan 11, 2014 10:05am Jan 11, 2014 10:05am
Quoting Proximus
the Dollar index is probably the most correlated with the EUR/USD
US dollar index is made for 57% out of E/U. You're merely correlating E/U with itself.
No greed. No fear. Just maths.
  • #3
  • Jan 11, 2014 12:00pm Jan 11, 2014 12:00pm
yap!!
I'm with u proximus

it gives a vision of ratio between the usd and the other currencies.
it goes up, it means usd get stronger
it goes down, weaker usd
I just simplify the complicated things
at least it can gives me more and better consideration before opening position

but don't forget some currencies can get a lot of steam due to its fundie
when it happens, usd index will not affect much

that's just my visions, cmiiw
  • #4
  • Jan 11, 2014 12:23pm Jan 11, 2014 12:23pm
Quoting Proximus
So what do you think guys?
I think, and just confirmed on your profile, you're not into trading for very long. Everything seems like a big novelty/discovery to you. Nothing wrong with that in the sense that you're learning, but your time surely could be spent more productively if you ask me.
"To hold, you must first open your hand. Let go." - Lao Tzu
  • #5
  • Jan 11, 2014 12:40pm Jan 11, 2014 12:40pm
Unfortunately you're confusing correlation with causation. Since the DX is made up of a large amount of Eur the two 'co-vary' (though due to the way they are set up when the Eur goes up the DX will go down). One is not 'causing' the other. For example drownings are correlated with Ice Cream sales. But one does't 'cause' the other. They're correlated because in summer more people drown by swimming and more people buy Ice Cream. In Winter both go down. So it's the season of the year that underlies the correlation. With the Eur/USD and the DX it's the Euro that is accounting for the majority of the correlation. If you could find a change in one that precedes the other - than you'd have a predictor!
  • #6
  • Jan 11, 2014 12:58pm Jan 11, 2014 12:58pm
Here is a simple example. In EURUSD it is not clear why price reversed right there, but if you look at the USDINX it starts to make ''some'' sense.
Attached Image (click to enlarge)
Click to Enlarge Name: 2014-01-11- 22-44-23.png Size: 44 KB


This type of analysis can be used to your benefit, but tbh, it is not easy to successfully implement this into real trading for practical reasons.... unless you have great understanding about price movement and can multitask successfully.
  • #7
  • Jan 11, 2014 1:08pm Jan 11, 2014 1:08pm
Quoting PipMeUp
{quote} US dollar index is made for 57% out of E/U. You're merely correlating E/U with itself.
Good point, i will substract the Euro, from it and leave only USD containing pairs, so we shall compare any strong USD denominated pair which has any cause to the EUR.Also i should use a more smooth average, like more weigth on the euroland pairs like the GBP & CHF & SEK, and less on the asian pairs

The dollarindex is really a bad idea.

Quoting lugy
yap!! I'm with u proximus it gives a vision of ratio between the usd and the other currencies. it goes up, it means usd get stronger it goes down, weaker usd I just simplify the complicated things at least it can gives me more and better consideration before opening position but don't forget some currencies can get a lot of steam due to its fundie when it happens, usd index will not affect much that's just my visions, cmiiw
Yep i need a better composition of pairs, i will work on that

Quoting vox dei
{quote} I think, and just confirmed on your profile, you're not into trading for very long. Everything seems like a big novelty/discovery to you. Nothing wrong with that in the sense that you're learning, but your time surely could be spent more productively if you ask me.
My profile is just a profile, what counts is my experience, and that i have much, i have 2 year experience from the technical-fundamental to the quantitative side.

Quoting schpsyk
Unfortunately you're confusing correlation with causation. Since the DX is made up of a large amount of Eur the two 'co-vary' (though due to the way they are set up when the Eur goes up the DX will go down). One is not 'causing' the other. For example drownings are correlated with Ice Cream sales. But one does't 'cause' the other. They're correlated because in summer more people drown by swimming and more people buy Ice Cream. In Winter both go down. So it's the season of the year that underlies the correlation. With the Eur/USD and the DX it's...
You have a point there, but correlation can also signal hidden momentum changes which are barely visible in the current currency and more visible in any other.

So i think its a valid point to check on them.If we fail to gain any edge, atleast we tried

But i think this thread is dedicated for such research and observations, so its a good attempt to check the correlations and find momentum shifts in any correlating currencies maybe we can predict the market with a better accuracy or atleast gain a little bit of edge there
"There's a sucker born every minute" - P.T. Barnum
  • #8
  • Jan 11, 2014 1:36pm Jan 11, 2014 1:36pm
Most correlating or rather inverse correlating pairs with eu are usd/chf and (what many traders may be surprise) usd/dkk. On every tf correlations is about -0,98.
  • #9
  • Jan 11, 2014 2:01pm Jan 11, 2014 2:01pm
Quoting M15
Most correlating or rather inverse correlating pairs with eu are usd/chf and (what many traders may be surprise) usd/dkk. On every tf correlations is about -0,98.
Strange that traders are surprised by what they should already know. EUR/CHF is pegged by SNB. EUR/DKK is soft pegged by the Danish centrak bank. What a surprise they correlate 98%!

http://www.purefx.co.uk/foreign-curr...d-to-the-euro/
No greed. No fear. Just maths.
  • #10
  • Edited 2:39pm Jan 11, 2014 2:05pm | Edited 2:39pm
Quoting PipMeUp
{quote} Strange that traders are surprised by what they should already know. EUR/CHF is pegged by SNB. EUR/DKK is soft pegged by the Danish centrak bank. What a surprise they correlate 98%! http://www.purefx.co.uk/foreign-curr...d-to-the-euro/
If its pegged then it wont help us much, because no fluctuation = no information carrying capacity in the EUR/DKK about the future of the EUR/USD.

So i will not include the EUR/DKK in the next version of the indicator...

PS: Although its fixed rate, its strange, the volume fluctuation suggests that people do trade this:

Attached Image (click to enlarge)
Click to Enlarge Name: eurdkkh1.png Size: 33 KB


If there would be no spread, this would be the easiest pair to trade, since the variance of the price is so small,just put 2 opposite pending orders outside the min/max of the channel and wait for huge spike breakouts = 100% winrate

Grid systems would work on fixed rate currencies if there would be a small spread, but unfortunately brokers are clever and restrict such things
"There's a sucker born every minute" - P.T. Barnum
  • #11
  • Edited 3:03pm Jan 11, 2014 2:47pm | Edited 3:03pm
The best way to trade EUR/USD or any pair for that matter is to LOOK AT WHAT THE CHART IS TELLING YOU!

I don't see where the dollar index fits into this one bit. You're looking for correlations that may or may not exist. Focus on the charts and you'll do much better.
Please check out my blog at https://from-zero-to-hero . net
  • #12
  • Edited 4:54pm Jan 11, 2014 4:13pm | Edited 4:54pm
Ok i made a few improvements, so i collected all major currencies that are worth something, from wikipedia:

http://en.wikipedia.org/wiki/Currency_pair

  1. USD ($)
  2. EUR (€)
  3. JPY (¥)
  4. GBP (£)
  5. AUD ($)
  6. CHF (Fr)
  7. CAD ($)
  8. MXN ($)
  9. CNY (¥)
  10. NZD ($)
  11. SEK (kr)
  12. RUB (RUB)
  13. HKD ($)
  14. SGD ($)
  15. TRY (TRY)

Now i removed the EUR, because this will be our base currency to predict, and the quote will be the USD, also removed the chinese yuan because my broker doesnt have it, and the CHF because its pegged so its a "circular causation" so its like which came first the chicken or the egg?.
Also removed the RUB and the HKD because i could not find any correlation table with the EUR USD,but anyway probably not very related.

So the new correlation indicator will now use these pairs to determine the EUR/USD:

  1. USDJPY correlation: -78.7%
  2. GBPUSD correlation: 58.7%
  3. AUDUSD correlation: 40.9%
  4. USDCAD correlation: -61.6%
  5. USDMXN correlation: -16.2%
  6. NZDUSD correlation: 25.4%
  7. USDSEK correlation: -58.5%
  8. USDSGD correlation: -29.9%
  9. USDTRY correlation: -11.5%

So if you want to use my tool, make sure your broker supports them & that you have recent historical data downloaded from each pair (as much candles as you want the indicator to draw),otherwise the indicator will show a flat line.Final work was to find the weights, this was easy i just added the absolute values of the correlations together and find the proportions/weight of each pair:

USDJPY weight: 20.63450446%
GBPUSD weight: 15.39066597%
AUDUSD weight: 10.72364971%
USDCAD weight: 16.15102255%
USDMXN weight: 4.247509177%
NZDUSD weight: 6.659674882%
USDSEK weight: 15.33822758%
USDSGD weight: 7.839538542%
USDTRY weight: 3.015207132%

Divided each pair's weight by 100 to fit the indicator screen, and there you go, i present you the Correlator V1.0 indicator

Attached Image (click to enlarge)
Click to Enlarge Name: eurusdm1.png Size: 31 KB

The MT4 code is simple:

Inserted Code
  for (int i=0; i < limit; i++) 
  {
 CORRELATOR[i]=
 //BASE
 MathPow(iClose("GBPUSD",0,i), 0.15390666)*
 MathPow(iClose("AUDUSD",0,i), 0.107236497)*
 MathPow(iClose("NZDUSD",0,i), 0.066596749)*
 //QUOTE
 MathPow(iClose("USDJPY",0,i),-0.206345045)*
 MathPow(iClose("USDCAD",0,i),-0.161510225)*
 MathPow(iClose("USDMXN",0,i),-0.042475092)* 
 MathPow(iClose("USDSEK",0,i),-0.153382276)*   
 MathPow(iClose("USDSGD",0,i),-0.078395385)*
 MathPow(iClose("USDTRY",0,i),-0.030152071);
  }

No circular or self correlation used anymore, the EUR is now calculated from the pairs listed above, and yet its still very accurate, and now you can probably check for momentum shifts and/or divergences or find other ineficciencies to predict the EUR/USD, so i`m curious about your opinion guys, please post your thoughts about this indicator

Attached File(s)
File Type: mq4 CORRELATOR v1.0.mq4   2 KB | 1,200 downloads
"There's a sucker born every minute" - P.T. Barnum
  • #13
  • Jan 11, 2014 10:00pm Jan 11, 2014 10:00pm
I am already aware that the research you are posting will have influence on my own. I follow your postings with interest. Keep up the good work.
  • #14
  • Jan 12, 2014 12:55am Jan 12, 2014 12:55am
Quoting PipMeUp
{quote} US dollar index is made for 57% out of E/U. You're merely correlating E/U with itself.
Maybe i'm dumb, but if this is true, shouldn't you trade the US dollar index based on E/U? Not the other way around?
  • #15
  • Jan 12, 2014 2:47am Jan 12, 2014 2:47am
Quoting Proximus
PS: Although its fixed rate, its strange, the volume fluctuation suggests that people do trade this
Of course it is. Danish bank doesn't peg the DKK for fun. There is an economy behind. Speculation is what makes the market efficient. It is

actually the primary role of a market: killing speculations arbitrage.

Quoting Proximus
but unfortunately brokers are clever and restrict such things Smile
Brokers just don't care. The spread IS what restrict such a thing. The spread naturally comes from the lack of opportunity which reduces

the liquidity. LP make money on big spread here and on a lot of small spread on the majors. They cannot make money on a lot of small spread

on E/DK because they would make no money. They cannot make money on big spread on E/U because they compete against each other.

Quoting Proximus
Also removed the RUB and the HKD because i could not find any correlation table with the EUR USD,but anyway

probably not very related.
HKD is pegged to USD between ~7.75 and ~7.80. I know nothing about RUB except I wouldn't try to trade that...

Quoting Proximus
now you can probably check for momentum shifts and/or divergences or find other ineficciencies to predict the

EUR/USD, so i`m curious about your opinion guys, please post your thoughts about this indicator
You can also build the same kind of index for each of the 7 majors. Deduce the strongest curreny and the weakest one and look at the chart of their cross ;-)
=> http://www.forexfactory.com/showthread.php?t=163158
I'm also doing something similar, using a different approach: http://www.forexfactory.com/showthread.php?t=435706
No greed. No fear. Just maths.
  • #16
  • Edited 5:45am Jan 12, 2014 5:30am | Edited 5:45am
How about saying the market is kinda efficient, and kinda random and kinda predictable as well? Kinda...

The dollar index, however you derive it, is meant to be a measure of the strength of the dollar, and for my interest in correlated pairs, (The implied relationships of complementary triplets) probably a better tool in understanding those. THE USD rising against one currency is only a possible sign that it is rising in general, but the dollar index rising is a certainty that it is, in general.

In my brief start at studying correlated triplets, I have observed that certain bowl-shaped patterns appear to be more effective at penetrating the filter provided by the intermediate pair.

The pattern has been described as hand-shaped, and been made fun of (or maybe not) as "boob" shaped. I have perused several systems that make much use of this pattern, and think it should be taken seriously.

Yes, I have observed that the triplet thing is not iron, but I believe an edge can be obtained if I do not commit myself prematurely.

It is almost impossible for a chance in the real world to be absolutely even; you try to get it biased, however slightly, in your favor, and ensure most of all that it is not biased against.

One thing I have never taken a look at that might be worthwhile noticing: When the spreads widen during a news event, do they ever widen more for a buy than for a sell or vice-verse? Any body ever notice?

Of course the quoted spreads are equal by definition. But practically speaking, a pair that is accelerating into the wild blue yonder moves up some before it gets filled, and as far as you are concerned, this increases the spread. On the other hand, if you surmise things have gone too far and are due to come back down ( a frequent occurrence) your sell order gets filled instantly by brokers still desperate to find sellers. So the defacto spread is wider going one way than the other, isn't it?

Just another little wrinkle you get when trading the news...

Sooner or later everybody trades the news, if only just once....
  • #17
  • Jan 12, 2014 7:14am Jan 12, 2014 7:14am
Quoting TheGrind
{quote} Maybe i'm dumb, but if this is true, shouldn't you trade the US dollar index based on E/U? Not the other way around?
It doesnt matter, its correlating the currency with itself which is not rational, because they cause eaother, its not good to compare 2 things where the 2nd thing is part of the 1st thing.

Quoting Rap Skallion
How about saying the market is kinda efficient, and kinda random and kinda predictable as well? Kinda... The dollar index, however you derive it, is meant to be a measure of the strength of the dollar, and for my interest in correlated pairs, (The implied relationships of complementary triplets) probably a better tool in understanding those. THE USD rising against one currency is only a possible sign that it is rising in general, but the dollar index rising is a certainty that it is, in general. In my brief start at studying correlated triplets,...
The only way to be sure, if someone would quantitatively analyze those inneficiencies which arise from correlations.Otherwise its just wishful thinking.
"There's a sucker born every minute" - P.T. Barnum
  • #18
  • Jan 12, 2014 10:05am Jan 12, 2014 10:05am
(from proximus) "The only way to be sure, if someone would quantitatively analyze those inneficiencies which arise from correlations.Otherwise its just wishful thinking."

What percentage accuracy would you look for in one of these quantifications? +- 20%? I would think that would be the limit of something remotely useful...
The closer you come to to 50% the closer you are to (random )., I think

Would you use the root mean squared method, the sma method?

You'd need a methodology to determine when you have an inefficiency, Which would require a precise definition of what and inefficiency is.

If this sounds like I'm taking this on, maybe I am, kinda, sorta.

If I don't find something more profitable to run with in my spare time.

All this thinking makes my head hurt.

So I break, and come back to it from a different direction.

Thanks for your input. I know I'm putting a lot of foolishness on display, which can lead to nothing at all. No Guru here.
  • #19
  • Jan 12, 2014 10:33am Jan 12, 2014 10:33am
The screenshot shows four things about complimentary triplets.
1. They only work some of the time. That fits with my predilection for pending trades.
2. The pattern continues for a while. You have time to get in if you see it has just started.
3. The "Hand" formation ("U" formation) very effectively filtered through the "Screen" set up by the intermediary pair.
4. You don't seem to get a whole lot of signals, true or false, so that you are not encouraged to overtrade.

I don't have enough to compute with yet, but what I have seen has made me keep my interest.
Sorry if I'm over-posting.
Attached Image (click to enlarge)
http://www.forexfactory.com/attachment.php?attachmentid=1345580&stc=1&thumb=1&d=1389550811




Similar Threads
Card without bank account accepted by brokers6 replies
Incorporating Slippage Assumption in backtesting4 replies
Market Predictability - Technical Signals25 replies
Contesting the Accepted View of Bernanke0 replies
  • #20
  • Jan 12, 2014 1:50pm Jan 12, 2014 1:50pm
Quoting Rap Skallion
(from proximus) "The only way to be sure, if someone would quantitatively analyze those inneficiencies which arise from correlations.Otherwise its just wishful thinking." What percentage accuracy would you look for in one of these quantifications? +- 20%? I would think that would be the limit of something remotely useful... The closer you come to to 50% the closer you are to (random )., I think Would you use the root mean squared method, the sma method? You'd need a methodology to determine when you have an inefficiency, Which would require a precise...
+20% would be godly, i dont think any trader in the world has such predicting capability, eve n if they do they must hide it very well.But 6-10% could be acceptable, if the sample size is big enough i think 6% can still be an edge even if some of you thinks its random.

The best method to check would be to check for autocorrelations in the price itself, to see weather the newly composed indicator of mine has any information carrying capacity about future prices on the EUR/USD.Or we can correlate the 2 prices but shift the real EUR/USD feed into the future (on backtest), to simulate if the EUR/USD candle come after it was predicted by my indicator, or not.If there would exist a little bit of correlation of this kind in the future then it could mean that we can predict the price to a certain degree.

Now we only need to put a 1:1 RR position size and play the coin toss game , but with an unfair coin, where the advantage is on our side
And make infinite money...

Quote
One thing I have never taken a look at that might be worthwhile noticing: When the spreads widen during a news event, do they ever widen more for a buy than for a sell or vice-verse? Any body ever notice?

Thats just a coincidence probably.

Quote
The screenshot shows four things about complimentary triplets.
1. They only work some of the time. That fits with my predilection for pending trades.
2. The pattern continues for a while. You have time to get in if you see it has just started.
3. The "Hand" formation ("U" formation) very effectively filtered through the "Screen" set up by the intermediary pair.
4. You don't seem to get a whole lot of signals, true or false, so that you are not encouraged to overtrade.

I dont think that particular pattern there is my u shaped one, my pattern is a volatility breakout/pullback pattern, in your charts i cant see volatility increase before the breakout.The fact that the market takes a U shape can happen, but those charts in particular, i dont think it resembles the real U pattern of which i was takling about in other threads.
"There's a sucker born every minute" - P.T. Barnum
0 traders viewing now