Man - this is really detailed. I never trade this way, but boy what a great roadmap to consider if you ever wanted to get involved in it. Good writing!
Good article, and very good caveat on high kurtosis risk on spreads. Food for thought… other spreads like WTI/Brent, Soy oil/ Soybeans, corn/ethanol, wti/refined products, etc. Similar, high kurtosis, traders tend to upsize due to lower margins and loser vol, then comes the hammer. A really interesting thing I spent some time on (at work some time ago) … calendar spread vs calendar spread. Some data caveats, some contracts have undergone specification changes over the years which makes stitching spreads challenging. For example Brent expiry dates got changed big time, would mess up any automated system. In energy markets you can roll forward up to a day before notice, the last day shows a big uptick in vol. Might have to roll sooner for other stuff. There a paper out there on combining trend (on outright) with contango/backwardation which is a calendar spread. Decent results. You can get a sense of the “drift “ by looking at the forward curve. You can capture the “theta drift” as time moves your contracts up/down the curve. In this context the bad events are when your contango spread jumps to backwardated or vice versa. Nasty stuff, but it does happen. Note: the difference between contango and backwardation doesn’t have to be the zero line (flat calendar spread).
Thanks Vic for the great insights! Operationally combining spread systems with outright systems is tricky, as you mention. I had to switch back to manually QC'ing every order from the portfolio of systems until I to manage improve the execution tool to handle the nuances "gracefully". If you do spread orders, then the contracts show as a "Combo" in the broker. If you trade as outrights, it shows as "Futures". But calculated margin for 1 combo is the same as 1 long and 1 short outright, as expected. It can get confusing when you have many positions, so manually executing the trades is error prone.
Can you recall the name of the paper? Sounds interesting.
Man - this is really detailed. I never trade this way, but boy what a great roadmap to consider if you ever wanted to get involved in it. Good writing!
Great work!
Thanks Jason!
Good article, and very good caveat on high kurtosis risk on spreads. Food for thought… other spreads like WTI/Brent, Soy oil/ Soybeans, corn/ethanol, wti/refined products, etc. Similar, high kurtosis, traders tend to upsize due to lower margins and loser vol, then comes the hammer. A really interesting thing I spent some time on (at work some time ago) … calendar spread vs calendar spread. Some data caveats, some contracts have undergone specification changes over the years which makes stitching spreads challenging. For example Brent expiry dates got changed big time, would mess up any automated system. In energy markets you can roll forward up to a day before notice, the last day shows a big uptick in vol. Might have to roll sooner for other stuff. There a paper out there on combining trend (on outright) with contango/backwardation which is a calendar spread. Decent results. You can get a sense of the “drift “ by looking at the forward curve. You can capture the “theta drift” as time moves your contracts up/down the curve. In this context the bad events are when your contango spread jumps to backwardated or vice versa. Nasty stuff, but it does happen. Note: the difference between contango and backwardation doesn’t have to be the zero line (flat calendar spread).
Thanks Vic for the great insights! Operationally combining spread systems with outright systems is tricky, as you mention. I had to switch back to manually QC'ing every order from the portfolio of systems until I to manage improve the execution tool to handle the nuances "gracefully". If you do spread orders, then the contracts show as a "Combo" in the broker. If you trade as outrights, it shows as "Futures". But calculated margin for 1 combo is the same as 1 long and 1 short outright, as expected. It can get confusing when you have many positions, so manually executing the trades is error prone.
Can you recall the name of the paper? Sounds interesting.
Think this might be the paper I was thinking of: https://www.sciencedirect.com/science/article/abs/pii/S0378426610001354
Awesome, thanks!. There's a paywall, but luckily the paper is also in SSRN: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1127213