Successful Automation?

[mahras2]

Quote from 2gtt:

one reason is that high stress events like the April 2000 Nasdaq tank or 9/11 cause higher degrees of correlation because of margin calls. an extreme event that hammers a portfolio (especially in futures) will force selling to cover the call. usually people are overleveraged in futures accounts, and they will have to sell everything even if only one of their positions get hit.

Edward Kim
2GTT,LLC

That could be a factor. However, I also believe that this is due to the fact that markets nowadays are closely connected. You have fund managers holding positions in markets ranging from US equities, to East Asian currencies, to Scandinavian bonds. Thus, once any high stress events leads a considerable number of these "super macro" players you get either mass selling/buying of securities leading to increased correlation.

Thats one hypothesis anyways.

 

[stephencrowley]

I am 100% automated.. high frequency, never hold overnight positions.. I look for 'synthetic pairs' or baskets of securities that move in some relation to another basket or single security. The average profit per trade is extremely small but makes up for it by being high volume.

Quote from 2gtt:

i don't look at charts either: it's only an example of what i see some other traders doing.

you are not purely mechanical, right? i like to hear more about your trading style. i trade mechanically, and my basket is from most futures exchanges around the world. the instruments i trade are highly uncorrelated (especially during extreme events).

Edward Kim
2GTT, LLC

 

[mahras2]

Interesting Steven, so you took pair trading another step by creating synthetic pairs? Were they selected due to fundamental relationships such as being the same sector etc?

 

[stephencrowley]

Essentially, yeah, I found it pretty hard to find single pairs.. easier for etfs/indicies, but still not exact. Many times they are in the same sector but it's not necessary.. it's a pretty painstaking process to find them: collection of high frequency data, testing for lead-lag relationships.. making sure they are relatively stable over time.

Quote from mahras2:

Interesting Steven, so you took pair trading another step by creating synthetic pairs? Were they selected due to fundamental relationships such as being the same sector etc?

 

[mahras2]

Yea. I dont think its really possible to automatically trade obvious pairs any longer which everyone and their sister is following.

I will soon do some research on potential "pair trading" situation in FX during high stress events.

 

[Prevail]

Interesting discussion. Diversification is great but like some of you stated, when a high level sigma event occurs correlation is frequently absolute value 1.

 

[newbunch]

Quote from mahras2:

My faith in diversification is also reduced due to the recent increase in correlation of markets during high stress events. Unrelated instruments tend to display a high correlation during high stress events than traditional theory would otherwise suggest. Currently I am doing some research on the implications of short term prices in markets during high stress news events and this phenomenon is displayed based upon preliminary data.

Make sure you've read When Genius Failed. He writes quite a lot about how all the world markets became correlated in 1998.

 

[mahras2]

Yes I have read it. But it really can't be missed. The terror attacks in London being a prime example when nearly every currency pair was affected by the event.

 

[stephencrowley]

How exactly were exchange rates affected? Is it a zero-sum game where some were up and some down, all down?

Quote from mahras2:

Yes I have read it. But it really can't be missed. The terror attacks in London being a prime example when nearly every currency pair was affected by the event.

 

[mahras2]

It differed. For example naturally the GBP tanked however the CHF showed strength. This is usually expected because Switzerland is known to be a neutral country and a "safe haven" for capital in times of crisis.