Fully automated futures trading

[globalarbtrader]

Hi Rob, I have been trying to wrap my head around changing skew over different measurement frequencies (prior discussion of it in your journal on page 51). I was wondering if this has anything to do with returns volatility hammering any +ve drift in the short term but over longer measurement periods the drift dominating (since the drift goes up faster than sdev)? Is this what the equation here is sort of saying?

....happy holidays to all!

Er no, not at all. The equation is just the definition of skew, regardless of what causes it. The earlier discussion was about trend following systems exhibiting momentum only at a time period which is greater than their typical holding period. At short horizons the measured skew will be dominated by the skew of the underlying assets, since a slow system won't really be changing positions very much.

GAT

 

[canicola]

Hi Rob,

Thanks for sharing all this interesting stuff. It's really fun to follow your trading activities. I've purchased both of your books and have a question regarding your second book Smart Portfolios. For ETF selection do you usually prefer to buy on an UK exchange or on country specific exchanges? One often reads that for getting US equity exposure it makes sense to buy an ETF domiciled in Ireland. However, doing some research I found that it depends heavily on the tax treaty of your home country and the US (in this example). For Ireland the withholding tax is reduced from 30% to 15%. However, the 15% are lost since they are on the fund level and if you (probably) correctly declare it on your tax statement you get double taxed. On the other hand, if the tax is on your level you might get back the full amount or some form of a tax credit. Do you consider this for every ETF? I guess this would be important for equity where you go down to country exposure but it looks like quite some effort. On top the commission vary a lot across region for IB.

thanks for your insight

cheers

 

[Elder]

Er no, not at all. The equation is just the definition of skew, regardless of what causes it. The earlier discussion was about trend following systems exhibiting momentum only at a time period which is greater than their typical holding period. At short horizons the measured skew will be dominated by the skew of the underlying assets, since a slow system won't really be changing positions very much.

GAT

Of course, that makes much more sense. This time I blame the Xmas booze . There might be a few further attempts to drink and think in the upcoming year, so apologies in advance. In the meantime, I wish you a very successful 2018!

 

[AvantGarde]

Hi GAT,
The recent flattening of the US yield curve got me thinking that it might be worth trend following on the slope of the yield curve. To work out the ratio of say 2yr and 10yr, would you use a duration adjusted spread or look at the daily dollar standard deviation, or something else to work out the correct spread? I imagine the ratio of contracts is dynamic, and not fixed, over time and that the results likely don’t vary too much between spread methods, but are important nonetheless...

 

[globalarbtrader]

Hi GAT,
The recent flattening of the US yield curve got me thinking that it might be worth trend following on the slope of the yield curve. To work out the ratio of say 2yr and 10yr, would you use a duration adjusted spread or look at the daily dollar standard deviation, or something else to work out the correct spread? I imagine the ratio of contracts is dynamic, and not fixed, over time and that the results likely don’t vary too much between spread methods, but are important nonetheless...

I've done this in the past (trend followed 2/10) and it works pretty well.

There are broadly 3 RV methods; (if your ignore cash neutral which is stupid in this context): duration neutral, vol neutral and one you didn't mention that is commonly used in equity space which is Beta neutral (where I guess Beta would be versus something like the fed funds rate, or the first PCA of the interest rate covariance matrix). However as you say the first two (duration and vol) make the most sense. I've looked at both in the past and performance wise there is no statistical difference, but also:

Advantages of vol neutral:

- in periods like we're just coming out of where the 2 year leg was pinned makes more sense
- simpler to calculate (don't need to find, or calculate, durations for each leg)

Advantages of duration neutral:

- being interest rate neutral makes the most 'economic sense'
- more commonly used by other RV traders (or is this a disadvantage?)
- leverage doesn't vary so much over time

Personally if I was to run this strategy right now I'd probably go with vol neutral.

GAT

 

[djames]

Hi all, after reading Systematic Trading and getting exposed to futures via a similar system, and now reading Smart Portfolios - I wonder what would be the disadvantage of structuring my "smart portfolio" within the bounds of my automated trading system? Meaning that a would express the diversification of assets exposed in Smart Portfolios through instruments incorporated into the trend following daily data system.

 

[globalarbtrader]

Hi Rob,

Thanks for sharing all this interesting stuff. It's really fun to follow your trading activities. I've purchased both of your books and have a question regarding your second book Smart Portfolios. For ETF selection do you usually prefer to buy on an UK exchange or on country specific exchanges? One often reads that for getting US equity exposure it makes sense to buy an ETF domiciled in Ireland. However, doing some research I found that it depends heavily on the tax treaty of your home country and the US (in this example). For Ireland the withholding tax is reduced from 30% to 15%. However, the 15% are lost since they are on the fund level and if you (probably) correctly declare it on your tax statement you get double taxed. On the other hand, if the tax is on your level you might get back the full amount or some form of a tax credit. Do you consider this for every ETF? I guess this would be important for equity where you go down to country exposure but it looks like quite some effort. On top the commission vary a lot across region for IB.

thanks for your insight

cheers

I stick to buying UK listed ETFs in the UK. However the vast majority of these are domiciled in Ireland or Luxembourg. Neither of these places charges any withholding tax, at least on the ETFs I own. This makes them tax wise foreign funds on which there is zero tax, so I just declare them on my tax form and then have to pay dividends at the UK rate (Capital gains is the same, regardless).

I'm not an expert on the tax situation for US investors, however AFAIK there is no withholding tax on Irish ETFs at source so I would imagine it would be the same.

This was top of google:

https://www2.deloitte.com/ie/en/pages/financial-services/articles/taxation-of-irish-etfs.html

GAT

 

[globalarbtrader]

Hi all, after reading Systematic Trading and getting exposed to futures via a similar system, and now reading Smart Portfolios - I wonder what would be the disadvantage of structuring my "smart portfolio" within the bounds of my automated trading system? Meaning that a would express the diversification of assets exposed in Smart Portfolios through instruments incorporated into the trend following daily data system.

Yes you could do this easily by adding +10 to all your bond / equity instrument forecasts to create a long bias to the risk premia in each asset. Also for things like precious metals, and perhaps also commodity futures if you trade them (it doesn't make sense for FX). You may also want to consider changing your instrument weights so they better reflect the typical portfolio weights in the book.

I don't do this myself as I have a long only portfolio that is substantially larger than my trading account.

There may be cost / tax implications of course.

GAT

 

[Gambit]

I've done this in the past (trend followed 2/10) and it works pretty well.

There are broadly 3 RV methods; (if your ignore cash neutral which is stupid in this context): duration neutral, vol neutral and one you didn't mention that is commonly used in equity space which is Beta neutral (where I guess Beta would be versus something like the fed funds rate, or the first PCA of the interest rate covariance matrix). However as you say the first two (duration and vol) make the most sense. I've looked at both in the past and performance wise there is no statistical difference, but also:

Advantages of vol neutral:

- in periods like we're just coming out of where the 2 year leg was pinned makes more sense
- simpler to calculate (don't need to find, or calculate, durations for each leg)

Personally if I was to run this strategy right now I'd probably go with vol neutral.

GAT

I'm not that familiar with fixed income so I had a question. Since interest rates are 0 or near zero in most Western countries, would it be beneficial to have bias to the downside ie rising rate environment leading to lower bond prices? I'd appreciate anybody's input on this.

 

[globalarbtrader]

I'm not that familiar with fixed income so I had a question. Since interest rates are 0 or near zero in most Western countries, would it be beneficial to have bias to the downside ie rising rate environment leading to lower bond prices?

Let's test this properly. If you regress vol normalise total returns (y axis) against current interest rate level (x axis) you will find there is a positive relationship (low rates, low returns) for long holding periods (1 year plus) BUT:

• the relationship is much weaker at shorter holding periods that are more relevant for traders
• the paucity of data for very low interest rates means that the statistical evidence is weaker here

The above argues for a small bias, however I'd prefer to incorporate this as another trading rule (which could be extremely simple) and let my portfolio allocation decide automatically how good it is.

GAT