pthomas215
1,251 posts
msg #132766
- Ignore pthomas215 |
11/24/2016 9:26:42 AM
It appears that any correlation is coincidental. Although on August 21 2015 there was a simultaneous breakout of the VIX and DUST in a big way, and more mild simultaneous escalations on March 9th 2015 and Dec 12 2014. But it might be just coincidence. I think historically it has been the opposite...so goes the market, people go to bonds and gold to hide.
|
Kevin_in_GA
4,599 posts
msg #132768
- Ignore Kevin_in_GA |
11/24/2016 10:09:36 AM
Determining the correlation between DUST and the VIX is easy:
As you can see, the correlations are low (10 day at -0.27, 50 day at -0.22, and 100 day at 0.29). These are basically UNCORRELATED asset groups. Valuable in portfolio design, but looking at these correlations I would not use DUST to trade off of ^VIX signals. IMHO.
|
jackmack
334 posts
msg #132770
- Ignore jackmack |
11/24/2016 11:22:05 AM
Kevin_in_GA
I was thinking about the position sizing and what would help in keeping with the 5.
Then I started to think about the entry and exit signals.
Maybe this was covered already in the thread but I didn't see it so... is the following feasible?
If any of the long signals fire ultimately that would mean get long - and if the next signal to get long fires you would again get long.
Then you would hold no more than 5 long positions if you got the signal to do so from 5 different filters UNTIL a short signal fires then you exit your longs and go short 1 position in the same manner as above until you would have a max of 5 short positions then hold until a long signal fires.
How would that perform?
Much worse I suspect but until I get SS I don't have the means to see how it did in back testing.
I know the thought is moving away from your signals but if ANY one of them is signaling long or short ultimately it would mean just that - no?
Thank you
|
jackmack
334 posts
msg #132771
- Ignore jackmack |
11/24/2016 12:00:58 PM
Never mind
That is a bad idea I just listed above
|
dtatu
143 posts
msg #132773
- Ignore dtatu |
11/24/2016 12:40:32 PM
... and Corr with GDX, which, I think, is more reasonable , than a 3x reverse splitted N times DUST, is:
^vix 10 day 0.42
^vix 50 day 0.12
^vix 100 day -0.41
Indeed: only dust
|
nibor100
1,097 posts
msg #132782
- Ignore nibor100 |
11/25/2016 11:33:42 AM
Using the past 6 months of ^VIX data downloaded from Yahoo Finance into Excel, I determined that if one went short each day, then 85.8% of the time there would be a profitable exit possible at an Open, within the next 10 days.
If one went long each day then 82.5% of the time there would be profitable exit possible at an Open, over the following 10 days.
However, if one went both long and short each day then just 68% of the time would both trades have a possible profitable exit within the next 10 days.
Possibly shows how regularly volatile the ^VIX is when Open prices are used for entries and exits.
Ed S.
|
ferndave
65 posts
msg #132904
- Ignore ferndave |
11/30/2016 9:19:15 AM
I scripted a backtest on the signals from 3-31 to 11-16, using TVIX and XIV, bought or sold at 5 minutes after open, with $10k. 111% return. -13.3% drawdown. Sharpe of 2.83. Ordering at 30 minutes after open dropped the returns around 20%.
Buy signal triggered for 11/30.
|
shillllihs
6,101 posts
msg #132921
- Ignore shillllihs |
11/30/2016 1:48:00 PM
Is this thing long short or what now? Knock knock.
Looks like everyone made out great and moved to the Hamptons.
|
mahkoh
1,065 posts
msg #132937
- Ignore mahkoh
modified |
11/30/2016 5:56:43 PM
Currently 3 long VIX signals on my end. Those from 11/17 and 11/21 are slightly down and up resp.
The last from 11/30 up some 6 %.
I've done some digging on NUGT and DUST and their assumed decay rate. I figured if you had gone short on both and rebalance every night at the close you could have made a small riskfree fortune?
The result actually surprised me a bit. A "pair trade" short of $ 5000 both sides from 1/1/2013 until 11/25/2016 (984 trading days) results in a total profit of $ 2962.85, or about $ 3 a day. If you take into account that rebalancing every night would take $ 2 a day ($ 1 a leg at IB's rate) you're left with 994.85. And that's not taking cost to borrow into account.
Unless of course I messed up my excel formulas..
|
dtatu
143 posts
msg #132938
- Ignore dtatu
modified |
11/30/2016 6:13:43 PM
Re . ferndave post
1.111% down to 20% for a 25 min delay in a buy/sell , using the same instruments? Anybody can explain it?
2. Do you, guys, think it's a good idea to use a 2X for buys( TVIX) and a 1X for sells( XIV) ?
3.what about backtesting a 2X , wasting , vehicle, reverse splitted N times?
- if you buy 5 units, in the backtest, at , example 1000$, you risk immensely more % of your account, no? Can one apply this kind of data to forward-test the system, while controlling the money management side of it, too?
|