Hi JonMartin, the short answer is YES – we use our framework in Bloomberg – an example link is here – http://www.zerohedge.com/sites/default/files/images/user3303/imageroot/2011/10/20111104_context.png – but it needs to be updated every day with our recalibrated factors and exposures and so becomes burdensome for most. That is why we created the site to enable traders to use the framework with no reliance on their own data provider.
The CIX for .CONTEXT is unfortunately not straightforward and nests a number of other custom indices. We have worked with some institutional clients to create the framework and custom indices for them in house but in general (unless higher frequency – not HFT – perspective is needed) it is unnecessary for most traders/hedgers.
Hi, just from a high level point of view, how’s the context calculated?
In the sense, do we look at the closing price of yesterday, say all prices at 16:00 NY and if the return of FX basket is Z, HY index is X, IG index is Y and then claims that ES should be at this level now? Or do we continuously roll, ie, we look at the last 30 minute of the return of FX baskets, HY index, etc and claim ES should be at this level now?
ie, if we fixed the previous time, then I expect the divergence will increase by the end of the day as the differences are accumulating (ie, the sd of the sum of each observations has to be higher) . If we rolled the previous time, then it’s difficult to look at the difference curve because a difference from -2 to 0 doesn’t mean we capture 2 points. In this particular case, if we had long ES and short index at -2, when the spread is 0, it means we make nothing.
Consider CONTEXT as being driven by a set of non-linearly dependent (i.e. not linearly correlated) factors including FX Carry, credit, commodities, rates/curves, and swap spreads. Our framework learns individual relationships among all of these factors and ES in order that we can create a quick-and-dirty perspective on how a broad basket of global risk-drivers is behaving relative to US equities.
The chart (difference curve as you describe it) is always self-consistent – i.e. there is no running recalibration within the chart (or over-fitting in real-time) as, just as you mention, this would make the framework very hard to use practically (which is our simple goal for our own trading and hopefully for yours). Instead the effort is made on creating a process that uses our experience with these relationships and the market’s empirical behavior, along with an understanding on when these relationships change regime, in order to know when a re-calibration is necessary.
The key for us is a framework that has an end-result that is practically useful (identifying market disruptions and pin-pointing specific risk drivers), actionable (whether for flow comprehension, risk support, or in our institutional client’s case convergence execution), and easy-to-use.
Hi, when are you guys making a real-time suscription available? Will the suscription reveal the ratios of future contracts to make this an actionable strategy?
We will be releasing the real-time subscription service very soon as we are close to finalizing the infrastructure testing. The current framework provides a 30-minute delayed version to non-subscribers.
We will be revealing the deltas for SPY Arbitrage (and several of our other ETF-based arbitrage strategies) – these will be updated every day as we recalibrate our models. CONTEXT is based on a blend of cash and futures market drivers and we work with our institutional client-base to create an internal infrastructure for execution of the compression trades. We will provide support on the use of CONTEXT in biasing trades and improving hedges – as well as perspective on what the main drivers of risk are on any specific day – this should help with creating executable trades and invaluable color on what signals to watch for intraday.
We have had several requests for a pure futures-based version of CONTEXT and will be working on creating that framework but in order to manage expectations, this is not a short-term project. We are also working on partnerships with several brokers to provide execution services based on the strategy itself directly – with the goal of reducing transaction costs, increasing liquidity, and broadening applicability.
Thanks for your patience while we work towards release. regards Tim
We will unveil the real-time subscription service soon. We are currently providing all of the frameworks on a 30-minute delayed basis with periodic auto-refresh for non-subscribers as we test the limits of our infrastructure.
How exactly does one use this? Is the assumption that the model and ES will come together? Can one tell if ES will move down (if it is higher than Context) or Context move up? Or is it just long one short the other? If the latter, what are the main components that one should short?
CONTEXT has a number of uses. Some of our institutional clients do indeed arbitrage the ‘spread’ by trading ES against the basket (as they have access to the calibrated weights and deltas). Some clients will use CONTEXT to bias their swing or intraday trading in one direction or another (i.e. if ES is excessively ‘high’ relative to CONTEXT then they will prefer to take short positions). A disconnect is also useful as it ‘highlights’ when one or more risk drivers are moving in an unexpected manner (i.e. the recent shift in relative USD-EUR-ES relationships as EUR becomes an increasingly preferred carry currency).
The main components include FX carry pairs, commodities (such as Gold and Oil specifically), TSY yields (and curve shape changes), and swap and credit levels/curves – this is why the ‘arbitrageurs will tend to be institutional in nature since the components are somewhat OTC as opposed to exchange-traded. We have designed purely ETF-based alternative for some unlevered funds/individuals (non-institutional) which can be traded by more investors.
Our non-institutional clients will also combine CONTEXT dislocations with their own trading styles (or with our credit-equity SPY Arb model) to increase win probabilities.
Several options-trading clients will bias their hedging programs (and rebalances) in favor of CONTEXT, preferring to hedge less/more if CONTEXT is biased in a direction favorable/unfavorable to their position.
The CONTEXT model is recalibrated (approximately daily to two-daily) dependent on the dynamics of the overall model or individual drivers but is usually static for the US day session and pre- and post-market (dependent on other markets – drivers – being open at the time).
Could you tell us what the carry/risk drivers are? I have looking at the context model at 5pm GMT and noticed that ES was 14 points above model. This morning it appears that at the same time last night model is 2 points above ES,
Drivers/Factors include FX carry pairs, commodities (we often use Gold and Oil specifically), Treasury yields (and often more importantly from a variance perspective – curve shape changes), swap and credit spread levels and curve shapes. The factors themselves change infrequently but the weightings and cross-dependencies will be recalibrated regularly – as you will see on the site (and note in your question).
The CONTEXT model is recalibrated based on co-dependence changes (one of the reasons we exposed the rolling correlation on the site). You will see this regularly and it is part of the model’s infrastructure. Our institutional clients will run longer-term (calibrated) models based on the frameworks we build for them (if they are holding/trading arbitrages for longer periods) but CONTEXT overall is designed to be a proxy for what risk-assets in general are doing relative to the US equity market. The difference you note is simply a recalibration (as correlations had broken down to a trigger point) of our short-term model (which is the CONTEXT model that we expose on the site).
The CONTEXT-ES relationship is not perfect. However, it has been used by us and our institutional and non-institutional clients for some time for arbitrages (more institutional), biasing trades (institutional and non-institutional), biasing hedging programs (more options institutional), and finally the simplest and cleanest way to use CONTEXT – to get a quick-and-dirty view of whether equity moves are well supported by broad risk-assets or not – which we use ourselves to spot divergences or more simply to reduce size / manage risk.
I hope this helps answer your question. Cheers Tim
Hello,
Is it possible to create the Context Model using Bloomberg ?
Thanks
Hi JonMartin,
the short answer is YES – we use our framework in Bloomberg – an example link is here – http://www.zerohedge.com/sites/default/files/images/user3303/imageroot/2011/10/20111104_context.png – but it needs to be updated every day with our recalibrated factors and exposures and so becomes burdensome for most. That is why we created the site to enable traders to use the framework with no reliance on their own data provider.
The CIX for .CONTEXT is unfortunately not straightforward and nests a number of other custom indices. We have worked with some institutional clients to create the framework and custom indices for them in house but in general (unless higher frequency – not HFT – perspective is needed) it is unnecessary for most traders/hedgers.
Hope this helps.
Cheers
Tim
Alright then. Will just stick to this site for intraday updates.
Thanks and appreciate your time.
Jon
Hi, just from a high level point of view, how’s the context calculated?
In the sense, do we look at the closing price of yesterday, say all prices at 16:00 NY and if the return of FX basket is Z, HY index is X, IG index is Y and then claims that ES should be at this level now? Or do we continuously roll, ie, we look at the last 30 minute of the return of FX baskets, HY index, etc and claim ES should be at this level now?
ie, if we fixed the previous time, then I expect the divergence will increase by the end of the day as the differences are accumulating (ie, the sd of the sum of each observations has to be higher) . If we rolled the previous time, then it’s difficult to look at the difference curve because a difference from -2 to 0 doesn’t mean we capture 2 points. In this particular case, if we had long ES and short index at -2, when the spread is 0, it means we make nothing.
Dear Magnoyu,
Consider CONTEXT as being driven by a set of non-linearly dependent (i.e. not linearly correlated) factors including FX Carry, credit, commodities, rates/curves, and swap spreads. Our framework learns individual relationships among all of these factors and ES in order that we can create a quick-and-dirty perspective on how a broad basket of global risk-drivers is behaving relative to US equities.
The chart (difference curve as you describe it) is always self-consistent – i.e. there is no running recalibration within the chart (or over-fitting in real-time) as, just as you mention, this would make the framework very hard to use practically (which is our simple goal for our own trading and hopefully for yours). Instead the effort is made on creating a process that uses our experience with these relationships and the market’s empirical behavior, along with an understanding on when these relationships change regime, in order to know when a re-calibration is necessary.
The key for us is a framework that has an end-result that is practically useful (identifying market disruptions and pin-pointing specific risk drivers), actionable (whether for flow comprehension, risk support, or in our institutional client’s case convergence execution), and easy-to-use.
We hope this helped.
Cheers
Tim
Hi,
when are you guys making a real-time suscription available? Will the suscription reveal the ratios of future contracts to make this an actionable strategy?
Thank you very much.
Dear Enguerrando,
We will be releasing the real-time subscription service very soon as we are close to finalizing the infrastructure testing. The current framework provides a 30-minute delayed version to non-subscribers.
We will be revealing the deltas for SPY Arbitrage (and several of our other ETF-based arbitrage strategies) – these will be updated every day as we recalibrate our models. CONTEXT is based on a blend of cash and futures market drivers and we work with our institutional client-base to create an internal infrastructure for execution of the compression trades. We will provide support on the use of CONTEXT in biasing trades and improving hedges – as well as perspective on what the main drivers of risk are on any specific day – this should help with creating executable trades and invaluable color on what signals to watch for intraday.
We have had several requests for a pure futures-based version of CONTEXT and will be working on creating that framework but in order to manage expectations, this is not a short-term project. We are also working on partnerships with several brokers to provide execution services based on the strategy itself directly – with the goal of reducing transaction costs, increasing liquidity, and broadening applicability.
Thanks for your patience while we work towards release.
regards
Tim
is this model updated in real time on this site, when I refresh the site. It seems its about 30 mins delayed.
Dear KGB0077,
We will unveil the real-time subscription service soon. We are currently providing all of the frameworks on a 30-minute delayed basis with periodic auto-refresh for non-subscribers as we test the limits of our infrastructure.
Thank you for your patience.
Regards
Tim
How exactly does one use this? Is the assumption that the model and ES will come together? Can one tell if ES will move down (if it is higher than Context) or Context move up? Or is it just long one short the other? If the latter, what are the main components that one should short?
Thanks.
Dear GMAK,
CONTEXT has a number of uses. Some of our institutional clients do indeed arbitrage the ‘spread’ by trading ES against the basket (as they have access to the calibrated weights and deltas). Some clients will use CONTEXT to bias their swing or intraday trading in one direction or another (i.e. if ES is excessively ‘high’ relative to CONTEXT then they will prefer to take short positions). A disconnect is also useful as it ‘highlights’ when one or more risk drivers are moving in an unexpected manner (i.e. the recent shift in relative USD-EUR-ES relationships as EUR becomes an increasingly preferred carry currency).
The main components include FX carry pairs, commodities (such as Gold and Oil specifically), TSY yields (and curve shape changes), and swap and credit levels/curves – this is why the ‘arbitrageurs will tend to be institutional in nature since the components are somewhat OTC as opposed to exchange-traded. We have designed purely ETF-based alternative for some unlevered funds/individuals (non-institutional) which can be traded by more investors.
Our non-institutional clients will also combine CONTEXT dislocations with their own trading styles (or with our credit-equity SPY Arb model) to increase win probabilities.
Several options-trading clients will bias their hedging programs (and rebalances) in favor of CONTEXT, preferring to hedge less/more if CONTEXT is biased in a direction favorable/unfavorable to their position.
The CONTEXT model is recalibrated (approximately daily to two-daily) dependent on the dynamics of the overall model or individual drivers but is usually static for the US day session and pre- and post-market (dependent on other markets – drivers – being open at the time).
Cheers
Tim
Hi,
Could you tell us what the carry/risk drivers are? I have looking at the context model at 5pm GMT and noticed that ES was 14 points above model. This morning it appears that at the same time last night model is 2 points above ES,
Regards,
Scott
Dear Scoted,
Drivers/Factors include FX carry pairs, commodities (we often use Gold and Oil specifically), Treasury yields (and often more importantly from a variance perspective – curve shape changes), swap and credit spread levels and curve shapes. The factors themselves change infrequently but the weightings and cross-dependencies will be recalibrated regularly – as you will see on the site (and note in your question).
The CONTEXT model is recalibrated based on co-dependence changes (one of the reasons we exposed the rolling correlation on the site). You will see this regularly and it is part of the model’s infrastructure. Our institutional clients will run longer-term (calibrated) models based on the frameworks we build for them (if they are holding/trading arbitrages for longer periods) but CONTEXT overall is designed to be a proxy for what risk-assets in general are doing relative to the US equity market. The difference you note is simply a recalibration (as correlations had broken down to a trigger point) of our short-term model (which is the CONTEXT model that we expose on the site).
The CONTEXT-ES relationship is not perfect. However, it has been used by us and our institutional and non-institutional clients for some time for arbitrages (more institutional), biasing trades (institutional and non-institutional), biasing hedging programs (more options institutional), and finally the simplest and cleanest way to use CONTEXT – to get a quick-and-dirty view of whether equity moves are well supported by broad risk-assets or not – which we use ourselves to spot divergences or more simply to reduce size / manage risk.
I hope this helps answer your question.
Cheers
Tim
Hi Tim,
how can I become your client?
Thanks.
Hi Tim,
how is the suscription service going_ i am very interested in becoming a client.
Thanks.