SpreadProfessor Clients - Thanks !

I distributed to clients a white paper I wrote with respect to what I consider to be a minimum acceptable amount of exchange data required in order to take an entry on a particular spread combination. The purpose of this guidance was to establish a minimal baseline of data points and a threshold historical timeline in order for the proprietary studies and our rules set to function with an optimal degree of confidence.

This guidance affects a small population of less liquid intra market spreads, and this issue has been a rather rare phenomenon to date - it manifests itself largely with newer clients who are just getting started paper trading. My 3K minimum contract expiry OI rule, which has been in place for years, has been and continues to be a good rule.
 
Clients who are "greener" and have just started to paper trade typically err on the side of being too early with respect to trade entries.

More experienced clients who have been paper trading for four to six months will show more patience and allow a trade set-up to more completely show itself in terms of the indicator package and our entry rules; therefore, the spread combinations ( especially intra market combinations further back in the curve ) they choose will almost certainly have adequate historical data.
 
bone said:
Tony,

Great questions, very constructive and much appreciated. Spread trading is a heavily used strategy by IB's, big prop traders and fund traders. As such, my answers reflect my own experiences - there are literally hundreds of thousands of arbitrage and relative value trading strategies and my answers are not by any means all encompassing. If you have properly done your homework in terms of modeling your spread trade, the trade should be largely immune from broader market delta directionality. As such, if indeed your spreading strategy models and trades with minimal immunity from broad market delta directionality you are generating pure portable alpha - and that's a good thing. Spread traders are essentially betting to profit from either the convergence or divergence between at least two ( but sometimes many ) highly correlated products. To accomplish such a feat, you have to essentially choose from a statistical standpoint highly correlated products. Could be 10 Yr. Cash Treasury Note versus the ZN future, could be a basket of the top 30 stocks listed in the S&P versus the ES future, could be the GC future versus a Gold Mining ETF, could be Jan16 NG vs Mar16NG vs May16NG... sky's the limit. You are essentially trading a market within a market. There is an incredible amount of dimensionality to this. A very good friend of mine makes markets at a major bank in OTC interest rate swaps - and he immediately hedges them with a specific Eurodollar month/year future dependent upon the swap duration he either bought or sold for an institutional client. These guys are essentially grinding out income; they are not betting the farm. Many of these traders work at a desk and they wish to remain employed - they want to get paid, and their managers want to get paid. No one is interested in taking what they would consider to be huge directional bets or crazy risk.

I digress. Your questions:

1. STIRS are short term interest rates. The most commonly traded cash products are Treasury Bills and the most commonly traded OTC products are duration proxy plain vanilla swaps. The most commonly traded STIR futures are the US Eurodollars and the ECB Euribor. There are also lower liquidity national instruments in Liffe and the SFE. These are basically zero coupon money market instruments. For the longer termed STIRS, there is an elevated level of convexity risk.

2. Properly constructed spread combinations do NOT require a benchmark per se. Having said that, many spread traders incorporate "benchmark" products into their spread trading strategies purely for liquidity reasons. FDX vs UPS could statistically be just as viable as SPY vs a basket including AAPL, MSFT, XOM, JNJ, GE, BRK-B. There is often more opportunity in the road less traveled.
More...

Hi Bone,
Thanks for the thorough reply. I have an unrelated follow up question.

I've heard you and others talk about lead-lag relationships in spreads.
Would you please explain what's meant by lead-lag in the context of spread trading.
Please explain like I'm the proverbial five year old. Maybe with an example, how's its traded and why lead-lag is a good thing (assuming it is).

Thanks
 
tonysoprano said:
Hi Bone,
I've heard you and others talk about lead-lag relationships in spreads.
Would you please explain what's meant by lead-lag in the context of spread trading.

Thanks
More...

This is my personal opinion, and there are certainly other valid opinions about this out there. "Lead-lag" to me at least is defined as the use of automation or very good manual point-and-click skills to "pick off" one of the products in a highly correlated intra commodity or intra product spread pair ( or one leg vs several or many in a highly correlated basket ) for a quick tic or so. The strategies that I hear of working these days are automated.

In fact, many years ago, there were times when I was manually legging a spread on a very short term time horizon where I just took a tic or two profit on that one leg without ever executing the other leg. It's happened in the pit and its' happened on the screen.

I'll give you a couple of examples that were relevant a few years ago. There was a period of time where the Canadian Dollar and the Crude Oil Futures contracts moved virtually tic-for-tic. I had some clients who chose to scalp the "trailing" product against the first mover or "leader" for quick profits. Around the same time period, the same relationship existed with the Australian Dollar and the SFE T-Bills. You could never say which product would "lead", and the relationships as I recalled lasted for a few to several months before they fell apart. In terms of the two clients that I know did this for several months - while my understanding is that is was to various degrees profitable for them, they were so engaged and hyper focused on one or two spread relationships that they ultimately ignored literally thousands of other spread combinations in the marketplace. Furthermore, the cointegration lags and disparities between these products could be very hostile. Also, your statistical sampling and modeling is always a look backwards, and when these once highly correlated intra market pairs decoupled, you could suffer a very bad streak of losses before the sampling caught up or you'd bled enough.

It would be, in my opinion, very difficult to do this with the obvious examples like SPY vs ES, or ZF vs ZN, or CL vs HO...

When these relationships fall apart, it can be brutal. The Bund vs ZN and quite frequently Brent vs WTI come to mind.

While I teach my clients to perform the statistical correlation analyses to ferret out unique intra market spread combinations, I do not show them how to scalp "lead-lag" relationships per se. For most clients, it would be a distraction from managing a portfolio of swing trades, and without a substantial investment for the ECN infrastructure and the requisite specialized programming required to properly automate such a strategy it is a "horse for another course" as it were.

Ultimately, I have to teach material that gives the best chance of success to the largest cross-section of clients with varied trading backgrounds. Which is why we swing trade a portfolio of spread combinations which are diversified across all electronically available market sectors ( energy, interest rates, softs, grains, etc. etc. ).
 
Last edited:
bone said:
This is my personal opinion, and there are certainly other valid opinions about this out there. "Lead-lag" to me at least is defined as the use of automation or very good manual point-and-click skills to "pick off" one of the products in a highly correlated intra commodity or intra product spread pair ( or one leg vs several or many in a highly correlated basket ) for a quick tic or so. The strategies that I hear of working these days are automated.

In fact, many years ago, there were times when I was manually legging a spread on a very short term time horizon where I just took a tic or two profit on that one leg without ever executing the other leg. It's happened in the pit and its' happened on the screen.

I'll give you a couple of examples that were relevant a few years ago. There was a period of time where the Canadian Dollar and the Crude Oil Futures contracts moved virtually tic-for-tic. I had some clients who chose to scalp the "trailing" product against the first mover or "leader" for quick profits. Around the same time period, the same relationship existed with the Australian Dollar and the SFE T-Bills. You could never say which product would "lead", and the relationships as I recalled lasted for a few to several months before they fell apart. In terms of the two clients that I know did this for several months - while my understanding is that is was to various degrees profitable for them, they were so engaged and hyper focused on one or two spread relationships that they ultimately ignored literally thousands of other spread combinations in the marketplace. Furthermore, the cointegration lags and disparities between these products could be very hostile. Also, your statistical sampling and modeling is always a look backwards, and when these once highly correlated intra market pairs decoupled, you could suffer a very bad streak of losses before the sampling caught up or you'd bled enough.

It would be, in my opinion, very difficult to do this with the obvious examples like SPY vs ES, or ZF vs ZN, or CL vs HO...

When these relationships fall apart, it can be brutal. The Bund vs ZN and quite frequently Brent vs WTI come to mind.

While I teach my clients to perform the statistical correlation analyses to ferret out unique intra market spread combinations, I do not show them how to scalp "lead-lag" relationships per se. For most clients, it would be a distraction from managing a portfolio of swing trades, and without a substantial investment for the ECN infrastructure and the requisite specialized programming required to properly automate such a strategy it is a "horse for another course" as it were.
More...
thanks man, we learn so much when you speak to yourselves.
 
If people can't tell that he's using sockpuppet accounts, they deserve what they get. Bone/tonysoprano has his history plastered all over this site. Let the customers do their due diligence and make their bets...
 
Pure unadulterated horseshit. If Barron can find that I have ever posted under another account or pseudonym or IP address than "Bone" in the 13 years that I have been an ET member, then I would encourage Baron to say so publicly here on ET.
 
bone said:
When these relationships fall apart, it can be brutal. The Bund vs ZN and quite frequently Brent vs WTI come to mind.
More...

I find that it can even vary by session as well. I notice that in the Asian and European sessions, CL is more tightly correlated with DX and 6E than it is during the day session. This of course probably has to do with greater volume in FX futures during those times.
 
The longest-lasting intra market, "not terribly obvious" lead-lag positively correlated spread relationship that I can recall seeing was the Japanese Yen and the third contract month Eurodollar in the 1990's. Lasted for a few years.
 
You must log in or register to reply here.
Back
Top