Well, my position in VIX was -10 contracts, and about -40 contracts on VSTOXX. Taking VIX at 14, $1000 per point, that's $10,000 per point. My account size is about £550k - totally conceivable that it could have a one day jump of 20-30 points, which combined with likely falling equity indices (I trade NQ, HSI, SMI, Eurostoxx 50), could easily put me in margin call territory or worse.
Fully automated futures trading
- Thread starter globalarbtrader
- Start date
Well, my position in VIX was -10 contracts, and about -40 contracts on VSTOXX. Taking VIX at 14, $1000 per point, that's $10,000 per point. My account size is about £550k - totally conceivable that it could have a one day jump of 20-30 points, which combined with likely falling equity indices (I trade NQ, HSI, SMI, Eurostoxx 50), could easily put me in margin call territory or worse.
Yeah, I wouldn't be sleeping real well either at those levels. Still reasonable to do a contract or two though, no?
A jump of 20-30 points would put you out of business, if you were a public-facing money manager.
You're probably right. Well I'm glad I'm out today, otherwise I'd be another few £k down. From a purely academic argument, it doesn't add much diversification to my already ~40 futures, but yep, could probably have a couple of contracts & be OK! Having said that, I've been running it at full strength for nearly a year and banked cash all the way, don't hugely want my luck to run out.
For those who are doing live / simu trading,
do you run a EOD process to calculate positions ? If so, what do you choose as EOD time given that different futures markets might end at different times? (Say SP500 vs KR10)
If there is a delay between your execution and position calculation, do you run some sort of price staleness check like GAT?
Thanks a lot
do you run a EOD process to calculate positions ? If so, what do you choose as EOD time given that different futures markets might end at different times? (Say SP500 vs KR10)
If there is a delay between your execution and position calculation, do you run some sort of price staleness check like GAT?
Thanks a lot
This is what I do. For each instrument have I defined what I call EOD. The instrument is considered for trading at that time. How did I define what is EOD? I am using IB as source of data and use daily OHLC bars for the historical data. These daily bars have a close time (the C of OHLC is the price at this time). This close time is defined by IB for each instrument. I review the instrument approximately 30 minutes before this close time. I stitch the historical daily OHLC (up to and including the previous trading day) together with the current market price at review time. This gives me the total price history for forecast calculation and value volatility calculation. Followed by a calculation whether trading is required. If trading is required is it immediately executed.
Some examples: 3KTB (Korea) 15:15 Seoul time zone, V2TX & BOBL 17:30 German time, LE & HE 12:30 Chicago time, etc.
Note that this system may not work properly if you are e.g. using historical data from Quandl as they use for certain instruments the settle price instead of a close price. That settle price is defined in a different way.
Many thanks for your reply HobbyTrading
I'm wondering if your system does a full bootstrap (instr weights, forecast weights etc) or it uses calculated values from some time ago. If it does use a full bootstrap does it require all futures to have a price for a certain time of day together before calculation.
I'm sorry, I don't know what you mean by "bootstrap". Each instrument is judged on its own merits. I don't combine calculation results (forecast, volatility) of several instruments together.
Based on the account value and the set target account volatility (25% in my case) is the available risk capital calculated. Each instrument has a weight factor (%) defined, which basically indicates what amount of that risk capital is available to this instrument. This is compared to the instrument's value volatility per contract. And then multiplied by the normalised forecast to result in the (what I call) desired position size. If this desired size is different from the actual size is an order sent and executed.
Does this answer your question?