Forums  > General  > Correlation and causation in markets  
     
Page 1 of 1
Display using:  

yoneda


Total Posts: 14
Joined: Jun 2019
 
Posted: 2020-08-14 22:00
Hi everyone, I have been pondering about exploiting correlation in markets. Say I detect correlation (via some test) in two price series at say the daily timeframe. Then I zoom in to the intraday and observe that the correlation persists. Then i get down to the minute and see that

1. correlation persists at this level, but happen too fast that I cannot exploit.

2. correlation persists with a delta t delay, I can exploit but not enough volume and often i have to market buy/sell

3. correlation disappears.

In case 1, how do i know which causes which? or I cant? In case 3, how can correlation build up at larger timeframe? In case 1, have you succeeded in front run such parties (that is trying to find a third source that is faster than the causing series)?

Lastly, your overall thoughts? Am I insane?

gaj


Total Posts: 113
Joined: Apr 2018
 
Posted: 2020-08-15 01:29
3 is the typical case when you zoom in to a very short timeframe, but it's not really about speed. It's because of the bid ask spread and transaction costs. Even if two instruments are 99% correlated at a daily timeframe, they won't move together tick by tick. When one instrument ticks up, the other may stay at the same price because it's not enough to go up a full tick yet. The move may be reflected by change in book pressure or weighted mid, but it usually doesn't capture the full correlation.

nikol


Total Posts: 1176
Joined: Jun 2005
 
Posted: 2020-08-15 11:45
@yoneda

On 3. That's normal. Look into Epps effect:
"is the phenomenon that the empirical correlation between the returns of two different stocks decreases with the length of the interval for which the price changes are measured." (Wiki)
Previous Thread :: Next Thread 
Page 1 of 1