A friendly warning, if you have a short attention span...skip to the conclusion! :-D
A long time follower pointed me to a website called vixcentral.com. On this page I was able to find the current and historical VIX futures structures. (Many thanks, btw!) With this data, I was able to identify periods of Contango, Strong Contango, Backwardation, and Strong Backwardation.
But first, we have to understand contango and backwardation and how they can affect XIV (and VXX). You can simply Google research contango and backwardation, but to shed some light on the interaction with XIV, Rob Hanna, founder of Quantifiable Edges and the newly introduced Overnight Edges, had this to say back in July...
“...but there is a 2nd factor that strongly influences XIV prices, and that is the VIX futures term structure. This is because XIV looks to maintain a constant-maturity short position. To do so it must buy to cover a certain number of front-month VIX futures each day and then short an offsetting amount of the 2nd month. When the 2nd month is priced above the front month that is referred to as “contango”. The stronger the contango, the more beneficial the term structure for XIV. To understand why this is beneficial consider tonight’s closing futures prices. July closed at $19.25 and August closed at $21.20. So XIV is buying at $19.25 today and selling at $21.10. Buying low and selling high like this can be quite beneficial to its value over the course of an entire month... ...the 10.1% spread between the front 2 months will have a positive influence on XIV’s price.”
Sounds good, but can a 1x leveraged XIV really compete with a 3x leveraged TNA? With contango on its side, you bet it can! Here are some tested results that I ran. First, during periods of strong contango.
Test 1: Buy TNA and XIV when VIX futures move to Strong Contango, sell when it exits to a different category. Starting amount: $10,000 each.
Final amount TNA: $12,023
Final amount XIV: $29,853
Results speak for themselves, wow!! XIV absolutely crushes TNA. (FYI: we are currently in a strong contango environment!)
Test 2: Ok, how about periods of (weak) contango. Again, starting at $10,000.
Final amount TNA: $11,966
Final amount XIV: $14,088
XIV wins again, but in a not so dramatic fashion. Still respectable!
Test 3: Backwardation (weak). Starting with $10,000.
Final amount TNA: $7331
Final amount XIV: $5754
XIV loses, as one might expect. However, let's pit TZA up against VXX during Backwardation...
Final amount TZA: $11,500
Final amount VXX: $14,122
And for fun...Let's let UVXY (2x VIX) in on the action.
Final amount UVXY: $10,456 (decay kept it from beating its 1x counterpart)
So during backwardation, it appears VXX beats TZA and UVXY. I should mention that UVXY hasn't been around a long time so it's not a very fair comparison. I am still going to avoid using it... Too dangerous for me.
Test 4: Strong Backwardation. This environment doesn't happen very often. And often after a prolonged period of strong contango, if we dip into strong backwardation, it's a great BUY (TNA) signal!
Final amount TNA: $9196
Final amount XIV: $6003
Final amount VXX: $12,935
Final amount TZA: $6354
UVXY - Not enough data.
With strong backwardation (which usually accompanies a downtrending stock market) we see VXX destroyed the decay-happy TZA.
Conclusion
When we are in a contango environment, I will favor XIV over TNA for any long signals. Short signals will continue to utilize TZA.
During backwardation, I will favor VXX over TZA for any short signals. But will continue to use TNA for buy signals. I will not touch XIV during backwardation, nor will I touch VXX during contango. And, I will never touch UVXY during either backwardation or contango. :)
I will update the webpage to list which VIX futures structure we are in (currently "strong contango") and will also display the 2 ETF's I will be favoring for system trades. Currently XIV (long signals) and TZA (short signals).
There is more work to be done, although I won't get to it anytime in the near future. Someday I will plug in this new configuration into my system and run some tests using TNA/TZA/XIV/VXX rather than just the TNA/TZA. I shudder to think how much work this will be! Lol. Additionally, I can see myself updating the webpage to track stats using all 4 ETF's but for now will just keep it simple tracking just TNA.
Here is what the current strong contango VIX futures structure chart looks like. (From vixcentral.com)
And backwardation from last August. VXX would benefit substantially over TZA. Likewise, TNA beats XIV for long/buy signals.
Thanks for reading.... have a great Labor Day!
Ok, I am a simple person... Are you/VIX saying that we are in store for a multi-month rally starting from here?
ReplyDeleteuh, wow...
Not at all. In fact, I believe we are close to topping either this coming week or next.
DeleteWhat I'm saying is that for my buy signals, I will be using XIV rather than TNA as long as we remain in contango. In theory, XIV should outperform TNA right now. Simple as that.
Incredible research, J. You've certainly done your labor for Labor Day. Thank you for this wonderful addition to the J-Trader system.
ReplyDeleteThanks, A!
DeleteGood article. About "While I can't predict what the VIX will do on any given day..." if anyone is interested knowing how and when VIX volatility will change, Volatility Research has come up with a pretty good way to predict VIX "fear guage" volatility using NYSE VXX:
ReplyDeletewww.sites.google.com/site/VolatilityResearch/
We're in the lull before the Perfect Storm.
Thanks for link. I will say some of the stuff on that site literally makes my head hurt! I'll have to down some ibuprofen in preparation of my next visit. :)
DeleteWhat is Volatility Research? I ask because their website looks rather amateurish and sensationalist.
ReplyDeleteHow do you define strong versus weak contango and backwardation?
ReplyDeleteGood question. I used a little leeway... in a prolonged period of strong contango, I required a percent contango value (the difference between the front 2 months) drop below 3% to move to "weak" contango. But during a stretch of weak contango, I required a move above 5% to get back to strong contango. This eliminated a lot of the categorical whipsaws. The same was true for backwardation.
DeleteHowever, for the purposes of my system, none of the above matters as I am using XIV regardless if the contango is weak or strong. I will post the strength though just as an FYI. Some may prefer the contango or backwardation be strong before dipping into volatility ETF's and I fully support that.
J, thanks for that analysis. One thing struck me as surprising, and there’s something else I’d like to point out to you in your analysis.
ReplyDeleteSurprising: If I understand your analysis correctly, then your system predicts volatility increases/decreases as well as index direction moves. I found this odd, since when I overlay the VIX on the SP500, they do not overlap all that much. But I can’t argue w/your numbers.
One thing to be aware of: Remember, it’s not just how much money you make, but how much you keep that matters. That is, you need to consider after-tax profits, too. And, for that matter, TNA trades can be done (and with less bleed) using Russell e-Mini futures. Besides ~50 TIMES more liquidity, 24hour trading, lowest commissions, and capital-loss-carryback (none of which are available to ETF traders), you get Section 1256 tax treatment. Now, for the poor people that may not matter much, but once you hit the higher tax brackets (which happens too fast these days), it does matter – a lot. I just crunched the numbers and, in your analysis below, two of your scenarios (when you consider what you have left after taxes) become non issues or even better (than the tested VIX-based ETF) using Russell futures instead of the TNA ETF product. For those who don’t know, doing the identical trade with an eMini product instead of an ETF will result in about 18% more left in your pocket at the end of the year. For example, $100k in yearly profits trading ETFs means, after just Federal taxes, you keep 65k (assuming 35% tax bracket), but you keep $77k if you did the identical trades using an eMini product. So choose your trading vehicle wisely.
I would add one more caution to anyone trafficking in XIV... it's only been around since Xmas 2010. That's not enough time to do sufficient back-testing, esp considering the mkt action since Xmas 2010 has been highly controlled and range-bound... and that back-testing window avoids any serious bear market activity.
ReplyDeleteA strategy recently put forth by TradingMarkets.com was recently heavily criticized for claiming such great returns, and, when XIV was back-calculated to include data since 2004 (as I recommend J-trader do w/his system, if he has not already), their impressive returns dwindled to only 9%/year returns and suffered massive drawdowns during not just the 2008 crisis, but also in 2006 and 2007.
Here's the link:
http://marketsci.wordpress.com/2012/08/02/a-critique-of-tradingmarkets-coms-vxxxiv-strategy/
Caveat Emptor.
JKH,
ReplyDeleteConcerning the correlation among SPY, XIV, and VXX, you might have missed this post I made a earlier:
"Actually, for the past year, SPY has a positive .88 correlation with XIV and a negative .93 correlation with VXX."
Also, are there E-minis for the Russell (I should know this, but I never trade any E-minis except for ES, NQ, and GC) and for volatility?
J, if you want to backtest further than XIV and VXX have existed, I have simulated values that I could email you.
J, tx for the correlation info. Yes, I missed that prior post. Thanks. To get a better feel for the correlation value and significance and if the last year or four are an anomaly or not, it'd be good to see the correlation back through at least the last 10 years (year by year). How far back does your simulated data go? The further the better. But anything that does not go back to at least 2006 is inadequate, honestly. We are likely to enter another (extended) bear phase of the mkts in the next year or so, so, trading a system that was only back-tested through a range-bound, up-trending market, during a bear mkt, will probably not turn out so well.
DeleteYes, there's an e-mini Russell... symbol is TF. Here's some info http://www.optionsxpress.com/our_products/futures/russell_2000_emini.aspx. It's a $100/pt contract so it's a bit hefty compared to the other eMini's. That makes it tough on small accounts.
I do not believe there are eMini's for volatility. Sorry.
I appreciate the offer of more historical data on XIV, but I'm not yet sold on the idea of trading volatility. I'd rather see your analysis go back 10 years or more, and see if the better returns hold up (esp after taxes).
Thanks, JKH, for you excellent points.
ReplyDeleteI think that there are simulated data for XIV and VXX on the Marketsci.com website since 2004.
What you say about backtesting is sound. But, if it applies to volatility, it should also apply to trading J's system using any ETF. So, do I understand you correctly that, in the absence of data that go back ten years, or at least to 2006, you're just using J's signals as part of the information you use to trade, not as actual mechanical system trading signals?
Trading something w/out backtesting it sufficiently is suicide. You need to run it through bull and bear market cycles lasting many years. Synthetic historical data is a start, but, still, not the real McCoy and makes the risk particularly high.
ReplyDeleteBut it's simpler than that. Even with the backtesting, I will (and have) use J's signals as supplemental to what I am doing. You've seen me posting what I'm doing, so that should be no surprise. I've commented on this before, too. Remember, always keep a human at the helm. Heck, even J does not trade his purely mechanically: from time to time, we see him take larger or smaller positions. That's smart. You might want to consider where the inventor is putting his wallet.
The fact is that J's systems are evolving. Since I started following this site (just a few months ago), I've seen first the J-trader system... then the CCI system... then the Composite system (with TNA)... now the Composite system (via "excess volatiilty" plays). And that is all in just the past few months -- where will this place be in a year or two or five? So, clearly, all of these should be taken in context, with the real probability that in the next few years you will see several of these systems stop being tracked/used and other new ones come on board. That's the normal evolution of website/blogs. And that's another reason to take the signals w/a grain of salt and be careful.
Keep in mind, also, that this whole "excess return" trading thing is one of the latest fad trading product being pushed on the mkts. Most ideas like this will fail (bad deaths), and, for those that survive, there will be 5x as many casualties along the way. Heck, I found this with a simple Google search: The tracking errors for the XIV are increasing as of late, too: http://sixfigureinvesting.com/wp-content/uploads/2011/08/VXX-XIVtracking-error.jpg So, these products are (in my mind) not ready... and definitely not time-tested vehicles I can sink some serious cash into, unlike the futs or even TNA.
Thanks for your thoughts, JKH.
ReplyDeleteI'd love to hear J's views on these issues.
Very interesting thread. I'd like to invite the interested reader to have a look at my blog post from lsat year "Can the VIX Really Predict the Market?" here: http://nightowltrader.blogspot.com/2011/02/can-vix-really-predict-market.html where I did some experiments in correlating the VIX and the SPX.
ReplyDeleteI think the VIX is important enough that I have a separate section devoted to it in my blog every night, and I look at VXX every night too.
I have to thank the great Dr. Brett Steenbarger for turning me on to the importance of the VIX. He often wrote about it in his Treder Feed blog, like this entry: http://traderfeed.blogspot.com/2008/12/volatility-of-vix-from-bear-panic-to.html.
Tx, Michelle. Would be nice if you did more than ~1 year's worth of data, though. How about running it back 10 years, since you have all the data? That way you cover a bear mkt, or two, and some nice up-trends.
ReplyDeleteAlso, remember, the XIV is based on SPVXSP, not VIX. Though they are closely correlated, they are slightly different.
As for predicting the mkts directions, I have always used the VIX in my daily analysis, but mostly to tell me when I should be net shorting or net longing (calling topping/bottoming areas), as some of your commentators replied to your post are doing. I've just never used it (as is the topic here, in this Blog) to predict day-to-day movements.
A larger data set sounds like a good idea. I'm not sure though that I want to include the period of the Great Recession though, as that could be considered an outlier that would skew the results. But I'll definitely revisit this. It would be interesting just to see how the period between the first analysis and today compares to the original results.
Delete