Stock Index Spread Trading Opportunities

bone

bone

I have clients trading my models with weighted spread combinations amongst the Russell, S&P, NASDAQ, FTSE, DAX, NIKKEI, DJ EuroStoxx, and CAC futures - as well as about 50 ETFs. I think I'm up to about 175 highly correlated combinations. I have clients complaining about running out of monitor space on quad monitors with four 25" monitors, and having to set audible alarm levels to keep track of it all.

You don't have to just take directional risk with these products - they are excellent relative value tools. I know that alot of guys will spread the mini-Dow vs. S&P mini, the DAX vs. the DJ EuroStoxx, etc. etc..

But listen, there's so much more to it than that. Learn to build your own markets - be creative! Tilt the risk/reward skew in your favor with statistics and good correlation studies. You can trade 750 XLF versus 100 SPY. You can trade 1 Kospi 200 future versus 1 FTSE future. Sky's the limit. You don't have to day trade or scalp the flat price directional ES like a gazillion other humans and try to make sense of candle formations, wave counts, Gann lines, chart patterns and all that stuff.

Try to Zig when everybody else is Zagging, for God's sake. I have been trading full-time and been positive for 17 straight years. Most of the successful traders I know who have banked some decent coin are trading some sort spread combination, and it's usually not a terribly obvious one. Judging by the content of most of these posts on ET, too many of you are chasing a poor risk/reward skew. Most of you guys are going about this the wrong way.
 
Please be careful for the rest of the week regarding technical trading signals and thin or passive/aggressive holiday markets.

Many independent/proprietary traders suffer during the U.S. Thanksgiving and Christmas/New Years holiday timeframes. Even when traders start off trading smaller than normal, they tend to get caught up in the moment and add to losing positions. Sometimes the best way to 'make' money is to sit on your hands or go find a bar.
 
I just signed a client today who has been a long time ET member, and who will use my strategy to trade equity pairs remotely and clearing Bright. I have a couple more clients who want to move into that space, so it should be interesting.

We've been running into hardware limitations - seems we need at least a Duo 2 Core processor running a 64 bit operating system for the trading model. I just ordered a Duo Quad Core processor Workstation box today. One of my clients runs 56 different spread combination models simultaneously using real-time tic T&S data.
 
To Bone...can you share with us your recommended contract vs contract ratios when spread trading any combination of the following:

ES vs YM
YM vs NQ
NQ vs ES

meaning...to make it even...would it be 1 to 1 per each of the above or 1 to 2 etc.
 
ES vs. YM; traded 1:1 (blended currency and volatility) with a 2 year correlation of +99.3%

NQ vs. YM; traded 1.5:1 (blended currency and volatility) with a 2 year correlation of +92.7%

NQ vs. ES; traded 1.5:1 on a currency-adjusted basis; traded 2:1 on a volatility-adjusted basis with a 2-year correlation of +95.3%

Although we trade them using these ratios, the charting in CQG is performed with proprietary expressions for the new 8.2 release version. The actual trading models are backtested for statistical fitness against cleaned one minute tic data using MatLab and Statistica toolbox packages.
 
Quote from bone:

ES vs. YM; traded 1:1 (blended currency and volatility) with a 2 year correlation of +99.3%

NQ vs. YM; traded 1.5:1 (blended currency and volatility) with a 2 year correlation of +92.7%

NQ vs. ES; traded 1.5:1 on a currency-adjusted basis; traded 2:1 on a volatility-adjusted basis with a 2-year correlation of +95.3%
More...

Bone, what do you mean by "blended currency and volatility" and why do you use a 2 year correlation?

Based on what I have read about correlation, the shorter the correlation the better i.e. 3-6 months would be more robust.
 
Jefferey:

You raise a perfectly valid point. Before I release a spread combination to clients, I also test it for intraday statistical fitness. I also want what I determine to be acceptable statistical Z-scores, so that usually brings me into the twelve, eighteen, twenty-four month sampling timeframes. I'm quite sensitive about what I show clients in terms of validated performance, so I tend to choose stronger performers.
 
When I refer to the term "blended", I mean that the currency-adjusted hedge ratio and the volatility-adjusted hedge ratio are close enough to average.

If those ratios are more than let's say a 25% differential, then a choice needs to be made regarding which one to bias.

Given the movement in the equity markets, hedge ratios have to be checked quite frequently.
 
Quote from bone:

ES vs. YM; traded 1:1 (blended currency and volatility) with a 2 year correlation of +99.3%

NQ vs. YM; traded 1.5:1 (blended currency and volatility) with a 2 year correlation of +92.7%

NQ vs. ES; traded 1.5:1 on a currency-adjusted basis; traded 2:1 on a volatility-adjusted basis with a 2-year correlation of +95.3%

Although we trade them using these ratios, the charting in CQG is performed with proprietary expressions for the new 8.2 release version. The actual trading models are backtested for statistical fitness against cleaned one minute tic data using MatLab and Statistica toolbox packages.
More...
thanks Bone...question though, how would one in all reality trade 1.5 contracts?...can that be done and how?
 
You must log in or register to reply here.
Back
Top