Fully automated futures trading

isotope1 · Jan 30, 2018

truetype said:
In 1987 equities fell by ¼. Had XIV existed, it would have fallen to zero (because its loss is truncated). Homemade short vol would have gone to -∞.

that sounds like we shouldn't trade it? Nothing's worth betting the house on.

Napoleon@Dynamite · Jan 30, 2018

isotope1 said:
that sounds like we shouldn't trade it? Nothing's worth betting the house on.

That's what it seems to me that truetype is saying. GAT, do you have an opinion on this? Seems like we may want to remove Vix from a trend following strategy, in case we get another 1987.

truetype · Jan 30, 2018

It's a free country. But when I see a post "lost 12% in two days" trendfollowing, I think, (1) way too much VIX and/or (2) way too high target vol.

tradrjoe · Jan 30, 2018

isotope1 said:
Yes. Although it's worth pointing out that @globalarbtrader has both a -10 and ewmac running, which is probably the optimal solution.

Sadly about 60% of today's losses were attributable to my short position on vix & vstoxx, so now down 12% on HWM. HWM was up 40% on 12 months though so it's not unexpected.

How did you manage to be up 40% over the past year doing trend following? Are you invested in some exotic stuff?

traider · Jan 31, 2018

Hi GAT,

what do you think of this fitting process proposed by Dr Ernest Chan?
http://epchan.blogspot.sg/2017/11/optimizing-trading-strategies-without.html

isotope1 · Jan 31, 2018

tradrjoe said:
How did you manage to be up 40% over the past year doing trend following? Are you invested in some exotic stuff?

Sorry, it was ~30%, my bad. In addition, a chunk of that came when I less diversified & more lucky (outcome of French election & CAC40 gapped in my favour). January was still great though.

isotope1 · Feb 2, 2018

Following everyone's advice, and realising how quickly vol can spike on bad news, I realised that a '2008' style spike might result in my account being bust. I'm not sure if it would have affected me (I trade the contract three months out), but some risks aren't worth taking.

I exited all my volatility positions. Effect on backtest is significantly worse results, both in terms of drawdown & sharpe. If anyone has any other thoughts on it, please contribute.

HobbyTrading · Feb 2, 2018

isotope1 said:
Following everyone's advice, and realising how quickly vol can spike on bad news, I realised that a '2008' style spike might result in my account being bust. I'm not sure if it would have affected me (I trade the contract three months out), but some risks aren't worth taking.

I exited all my volatility positions. Effect on backtest is significantly worse results, both in terms of drawdown & sharpe. If anyone has any other thoughts on it, please contribute.

If it makes you feel uncomfortable, and you can't sleep well due to it, then it might indeed be better to remove those instruments.
In my case: I do use V2TX (not the USA VIX), but assign a very low weight percentage to it.

truetype · Feb 2, 2018

This week will separate the sheep from the goats -- the truly diversified trendies from those who are just riding the 10-year bull market.

jtrader33 · Feb 2, 2018

isotope1 said:
Following everyone's advice, and realising how quickly vol can spike on bad news, I realised that a '2008' style spike might result in my account being bust. I'm not sure if it would have affected me (I trade the contract three months out), but some risks aren't worth taking.

I exited all my volatility positions. Effect on backtest is significantly worse results, both in terms of drawdown & sharpe. If anyone has any other thoughts on it, please contribute.

I don't trade quite the way you guys do in this thread, but for cases like this, I overlay some common sense position sizing constraints using cave man math. The thought process may sound silly but it has saved a lot of jobs. Ideally, you want a baseline layer of risk management that is crude and model free. From there, you can fine tune without violating your baselines in order to maximize mean/variance or whatever you prefer. In your particular case, I would do something like the following:

-During the '87 crash VIX was said to spike to >150, so let's assume a worst case of 200
-Third month futures are very unlikely to get anywhere near that high, so call a rough max of 100 (although it's probably considerably lower than that)
-That's ~600% higher than where things are trading now
-Select a survivable loss number from the position, say -30% on the account
-Then back into your max position size which would be contract value equal to a ~5% allocation in this example

Something like that should allow you to sleep at night without completely missing out.