Correlation + Momentum Strategy?

Hi all,

I am aware that many people attempted to do such a thing on this forum but I can't seem to find anyone in threads who actually did extensive testing of this.

Consider two anti-correlated pairs such as EURUSD and USDCHF. Using their respective momentum indicators (RSI, STOCHASTIC, etc.), we can build a synthetic indicator like the difference of both (e.g., RSI(EURUSD) - RSI(STOCHASTIC)) to find times where EURUSD is showing overbuying signs, while USDCHF confirms it with overselling signs. The anti-correlation would make it so that if the reversal has more chance to happen on both, and if it does not happen, there could be possibilities where one pair actually reverts while the other does not.

I made some small testing using M1 data in Python using this system:
- Buy EURUSD and sell USDCHF whenever STOCH(USDCHF) - STOCH(EURUSD) > 60 and the other way around
- Close the order when the differential reaches 0 again
...but it seems like it's not really working out at that timeframe. Did anyone else actually already try this in demo/live conditions, and can share some knowledge on this? (e.g., which pairs, which timeframe, which trading sessions, which indicator, which exit condition?)

This thread would be more like an open discussion, and I would be very happy if you guys could share your experience on such ideas, and how to improve.
Thanks!

I traded for a while something similar with those 2 pairs, since they are VERY negative correlated they almost always move in opposite directions. I just waited for a good signal and picked the strongest of the 2 pairs at that moment and once it started to go up/down i just picked the highest/lowest point to place a trade on the other pair (since the direction was already "confirmed")and adjust my overall profit/loss based on the running trades

If you add eur/chf to the mix you can have a another layer of data to this correlation as guidance; just watch out with your "exposure" since you are basically betting double on the "same" currency

NOW to make any type of indicator like you mention, you WILL need more accuracy than a regular time chart could give you (unless you have no problem reading a time chart in the first place)

Not a bad idea overall but needs more "detailing" like a car hehe, good luck =)

my idea is to trade negativ correlated pairs not via math models but via pic
https://www.forexfactory.com/thread/...4#post13800834
EURGBP EURUSD GBPUSD

How are you guys implementing this in practice? Had it worked out for you?

I have been thinking a lot about correlations as they relate to hedging. The correlation co-efficient essentially being how close to perfect the pair acts as a hedge vs another. A perfectly hedged instrument would be a correlation of 1.

What intrigues me is how this can be taken advantage of with high-risk, low-reward setups (2:1, 4:1, 9:1, etc) with super high correlations or essentially trading the same pair in both directions.

Say you have 2 independent events.

The 4 possible outcomes are

WW
LW
WL
LL

With correlations and a high-risk, low-reward setup, you can eliminate the LL outcome altogether, and the remaining outcomes become

WW
L(W * n)
(W *n)L

n is dependent on how volatile the market is during that initial setup.

Can you please elaborate more on this? I'm not sure I understand that correctly...

https://www.forexfactory.com/thread/...hedging?page=5

https://www.forexfactory.com/thread/...ategy?page=100

Maybe something along those lines? I would like to test this kind of strategy, did anyone already do some tests on this?

https://www.forexfactory.com/attachm...1&d=1365674857
https://www.forexfactory.com/attachm...1&d=1365673411

The idea is to wait for overbought + oversold situations and trade accordingly, by trying to make trades as robust using correlated pairs.
Would be happy to hear from you guys on this!