Alternative take on VIX action

While it appears that short term realized vol has doubled while we were asleep, it has not really shown to be sticky enough to filter through into the 20d realized vol in a significant manner (which is still hovering around the 17 level, up a couple of vols from week ago) so I am siding with Adam that it is more likely a short term blip until proven otherwise. The VIX premium over realized though has closed a bit from a very expensive level of over 11 vols to a slightly less rich figure of about 8. Is volatility set to return with a vengeance? Some big trades by smart money might seem to indicate as such as reported by Option Monster:

We have seen trades involving the selling 20,000 of the November 25-27.50 "strangle" and buying 80,000 November 40 calls for $0.79 last week and again for $0.75 this week. We have seen one trade that involved selling 20,000 of the November 47.50 calls and buying 20,000 each of the 55 and 60 calls.

The risk in the former trade is roughly $25 million--just in the VIX options. So we are not just talking about smart money, we are talking about THE smart money. And it is likely that their VIX book tied it to other things and that there are complex hedges going on.

But make no mistake, they have a lot of money on the line and don't put on these trades with the intention of losing. The VIX may come down, and the VIX futures may drop, but that is not what that smart money is betting on.

There is certainly a lot of money on the line so I thought it would be interesting to look at what might be going on. While the trade would certainly benefit from a volatility lift as the ratio leaves the trader long upside convexity, it leaves him 1. open to the gap risk between the 27.5 and 40 strikes at settlement, and 2. risk of sideways to down VIX action. The first point is unlikely to be of concern as a VIX spike to say, a 30 level would put the package solidly in the green which the trader would likely unwind at that point before theta grind eats in. It is point #2 which makes this interesting. While VIX at 25 might seem cheap in the recent historical context, it is actually closer to the rich side of things if we consider its lifetime history (since 1990) when the average is closer to 20, and the 1mth realized vol of the SPX is about 15 during that same period. At that size, even a settlement a couple of points under 25 or a sideways chop where it hangs around here would cause quite a bit of pain. Therefore it seems unlikely that he will be taking the mark to market hit and possibly a real loss at expiration, should volatility drops, without another leg to the trade that would benefit. What I am thinking is that the trade is actually a hedge for a short correlation or dispersion position, since correlation has looked very rich lately. To put simply, he is betting that stock correlation is overpriced by being long individual stock volatility and short index volatility. The trade would benefit if individual stocks do not move in unison as has been priced into index volatility recently, and it is a good bet as if the market continues to rally, it should eventually do so by picking winners over losers rather than in a continued meltup manner. The risk to the trade is if stocks do move in a highly correlated manner, as in a correction downwards, but that is where the VIX trade that Jamie explained kicks in.

If thats the case, it looks to a smart bet with convexity on both ends: if stocks rally, correlation should drop and he walks away with a nice profit on the dispersion and maybe a flat to small loss on the VIX trade and the notional on the correlation is likely much larger than on the options. Or, the general market corrects and he scores big on the VIX calls but loses on the dispersion leg. However, given that correlation is priced so rich already, the downside on that would be limited relative to the delta expansion on his long VIX calls. The only catch it would look like is if nothing happens where correlation and volatility stays expensive, but people are not likely to keep overpaying for options forever. There is big money indeed on the table here, but I am just not convinced that it is a one way bet on higher volatility, so I probably won't be rushing out to grab some vol yet.

Probability of Downside Gaps Getting Filled

Here is some statistics on how long it takes for downside gaps of > 0.75% on SPY to get filled:


Total # of occurrences = 100 (excluding Monday), average % change on the day is -2.37%.

As one can expect, responsive buyers comes in on the first subsequent day, giving it the highest probability of closing such gaps. The average closing price on that day is 99% of the pre-gap closing price, in other words, the performance today was slightly average (Friday close of $100.79 x 99% = $99.78) since SPY closed at $99.28. For the instances where the gap did not close on the first day, the average performance on the second day is negative, with the close price averaging 98.6% of the pre-gap price. Similarly, on the 3rd and 4th days, the readings are respectively 97.9% and 97.6%. Therefore, history suggests that each day that passes where we do not reclaim Friday's closing level, the odds of a search for new lows increases.

Look at relative Volatility between FXI and SPY

With the US Stock markets down more than 2% today following a sell off in Asia, the spot VIX has spiked more than 23% to 27.5 as of 2pm EST. As I have noted in my previous post, and also by other fellow options bloggers, the VIX is trading at a large premium over the realized volatility of the SPX itself and therefore it seems that rather than the market being very complacent, fear could actually be overpriced. If we close at the current levels today, the 1m realized volatility will still only be at 16%, making that premium even higher than last week at over 11. So in this market is there a better way to hedge the downside should it materialize?

To answer that, I looked to what the pundits are considering the culprit of today's action, the Chinese stock market, and whether the options there are offering better hedging opportunities than the S&P 500 options. As a proxy, I am looking at the U.S. traded ETF, the iShares FTSE/Xinhua China 25 Index (FXI). In the chart below, I have plotted the 3mth realized volatility of the FXI relative to the SPY (blue line). The lowest recent reading is just a bit shy of 1.5 earlier this year, meaning the FXI was 1.5x as volatile as the SPY on a 3mth basis. Prior to that, one would have to go all the way back to 2005, to find a reading below 1.5 (the absolute low was 1.26 in 7/2005), so it seems that 1.5 is a good rule of thumb for the lower bound of this relationship.




Looking at the November options for the FXI, with approx 3mths to expiry, the ATM volatility is quoted at about the 42% level. While there is no November options for the SPY on the board yet, the volatility term structure between the Oct and Dec ATM lines is quite flat at about 25.5%, which we will interpret as the Nov implied vol. The ratio between the two at 1.65 (Red line in graph), is the current market's expectation of the relative volatility between the two markets, and compared to the historical figures, is actually very much towards the low end. The current reading is in the 16% percentile of all historical readings, meaning there has been a 84% probability of the actual realized ratio being higher at expiration.

Although it may well be true that the turn in the Chinese market is foretelling a correction in the US one, the timing might not be coincident. It might also be that the Chinese market is more overvalued, having rallied 100% off the lows, compared to only 50% for the S&P 500. Regardless, by looking at the historical ratio of the realized volatility between the two, it appears that risk premium in the FXI is priced relatively cheap compared with the SPY, and a ratio adjusted trade might therefore offer a better risk reward alternative to purchasing outright protection in the US markets.

Expensive VIX

Despite all the VIX break out calls lately every time it ticks up, realized volatility in the SPX has not followed which makes buying options a very expensive proposition at this point. As of the close today, the VIX is quoted at a full 10 vol premium over the 1mth realized SPX volatility which is down at 14.7. As the VIX is supposed to be a forward estimate of the volatility to be realized in the next month, I took a look at historically what it meant when the premium is this large.

I extracted the sample of days when the VIX closed at > 10 vol premium over the 1mth realized volatility, the condition we are in currently. Since 1990 when the data for the new VIX methodology became available, there were 311 such days (6.3% occurrence rate).

It turns out that the subsequent realized volatility of the SPX over the next month is actually below the VIX at inception 88% of the time, by an average of 7.3 vols. The result is quite one sided and really argues against being long volatility when the premium is this large, as on average you are expected to lose 6.4% (7.3%*88%) to vega even if you do call the direction correctly. The size of the loss is 78% explained by the VIX over realized premium at inception. Furthermore, the spot VIX at the close of the period is also below the starting level in 73% of the sample, averaging 2.3 vols below. Therefore, the positive term structure of the VIX futures at the moment could be an interesting play.

On the other hand, the returns on the SPX over the next month is quite positive, averaging 1.9%, and represents a gain 64% of the time. This compares to a 0.55% return and 60% win rate over all 20 day periods during that time. The larger magnitude positive returns do tend to be skewed towards those with a higher initial starting premium, albeit not particularly well explained with a simple linear regression which only explains 33% of the correlation between the size of the return and the initial premium amount.

Favoring a short term bounce in XAU

The XAU has declined about 7% since it peaked last Tuesday at 156 and is currently indicated at about the 145 level. All 16 stocks within the index are setting new weekly lows and although this by itself does not necessarily indicate a reversal is at hand,historical data strongly favors at least a short term bounce into tomorrow's session. Since 6/2005 using the current constituents, on days when at least 15 of the 16 stocks are making new weekly lows, the average high print on the XAU on the next day is 2.5% above that day's closing price. On the 55 days that this has happened, the index traded higher on 48 of those occasions (87% prob of at least breaking even). On the best occurrence the return was 7.4% and on the worst day the high print was 0.7% below today's close. (Note: this is tomorrow's high to today's close, NOT close to close) The odds look fairly compelling for a very short term (day) trade at these levels, but beware that it does not say much about any timeframe longer than intraday, as the average close to close return for tomorrow is actually slightly negative at 0.22%.

See chart for historical distribution of XAU's next day high to today's close after at least 15 stocks in index making simultaneous weekly lows:

NYSE Volume

Quick note on NYSE volume divergence that appeared yesterday intraday, as well as today at the moment. As the SPX was hitting lows around 10:45 yesterday, the NYSE up volume comprised about 58% of total volume, the divergence eventually resolved itself by prices moving higher. The same divergence is noted today, as I am writing, the up volume currently running at 67%, while the A/D line is at -600 (yesterday was about -1200). A look back in history since my data begain in 1996 shows that when up volume is greater than 65%, the probability of a higher close is 96.6%, median gain is 1.05%.

Another Day, Another Gap

A strong start to the month of August in the SPX as buyers seek to dispel the 'sell in May go away' myth. The strong action today began with another >1% opening gap in the SPY ETF, the second such gap since only 2 sessions ago, probably as more bears are forced to cover. I suppose at least the adage never sell short a dull market is proven true. Such large gaps happening so close in proximity is in fact a fairly rare occurrence. In fact, from my SPY data going back to 2/1993, I can only find 8 instances of the SPY having 2 opening gaps of > 1% within a span of 3 sessions (two of them were rounded up to 1% but whats 0.05% between friends..). So, is this a sign of further strength? or have all the shorts already been forced to cover at the top and its down we go?

One would be quite right to read this as at least a sign of short term exhaustion in buying. Indeed out of the 8 instances, only once did the SPY managed to make a new high on the day after the second gap up day (tomorrow), suggestive of at least a pause and consolidation for the coming sessions. Thereafter, things look up for the bullish case again. The probability of the SPY making a new high, after the second gap up day, in the 5 sessions following are:

Day	Prob of new high
T+1 (Tomorrow)	1/8
T+2	3/8
T+3	3/8
T+4	5/8
T+5	7/8

On T+5, the average new high above the current high, is 1.03%. ($101.56 using today's high of $100.53) Not monstrous profits, but the odds seem pretty good. It also seems we can enter the long trade at better levels during the first 3 days, as from the 7 successful instances, the average pullback from the highs before the next upswing began is 2.6% (the min pullback was 1.9%, max 3.6%). The day on which the low print occurs on during the next 3 days is evenly split.

On the last note, its not all despair for the bears though, because for the one case where this did not work out, the SPY rolled over and dropped 5% after a week.

Quick peak under the hood

Following the SPX strength yesterday, a scan of the Russell 3000 universe of stocks shows that only 149 of them (5%) participated in making their own new highs since the broad market bottomed on Mar. 6. The chart below shows that the 5 day avg of the number of stocks making new cycle highs since that day (using the weekly avg to smooth daily fluctuations). It is evident that participation is thinning each time the SPX swung to new highs in May, June and July. By itself it does not mean that we are looking at an immediate drop, as the divergence can also be resolved with a broadening of breadth, it is something worth keeping an eye on.

Island Reversal ?

Today's action in US equities started strongly with a >1% gap in the SPY, and looked set for a trend day until sellers came in near the close and drove it to close within the bottom 10% of the day's range. Bearish chartists are certainly rubbing their hands with glee with thoughts of island reversals in their minds, especially after the vertical run up that we have experienced in the past weeks. Let see if we can gain some clue from the past action on days that start with an open gap of >1%.

Using data going back to 1993, we had163 instances of days with a SPY open gap of >1%, 53 of which (35.5%) were immediately closed on the same day (price trading below the previous close or pre-opening gap price). As we now know that was not the case today, we can narrow our focus on the rest of the sample and see what tomorrow may bring. As it turns out, the probability of the gap being closed on the second day is only about 11% (12 of 105 cases, with the average loss being 1.7%), so the odds of the formation of an 'island reversal' signaling buying exhaustion just does not seem high. If we further narrow the sample down to those appearing after a period of strength, as was the case today which marked a monthly high, we find that the bearish case diminishes even more. Within our sample, 22 of those days also marked monthly highs, out of which only 1 experienced a gap fill on the second day. In summary, historical evidence makes a bear case very hard to build, at least for the immediate trade going into Friday.

Having said that, one should also keep an eye on the breadth of the market, which might be showing signs of tiredness as evidenced by only 4 sectors managing to close above the monthly range despite the broader index managing to do so.

Low VIX ?

A common theme that has been much discussed is the persistent decline in the commonly referred to fear gauge in the market the VIX and whether it is too low and artificially manipulated given its relentless decline from the peak level of over 80 during the height of the crunch in 2008 to a current level of about 25. To recap, VIX is a measure of the implied volatility for a constant 30-day to maturity option chain based on the S&P500. If we accept that volatility is a mean reverting process, then the VIX level should be related to the realized volatility of the SPX plus some value of risk premium. As a fear gauge, the premium should also intuitively be higher in times of market stress. To test out this hypothesis, we shall take a look at a simple regression between the VIX premium (VIX - 1m realized SPX vol) vs. the 1m SPX realized vol and the weekly % change of the SPX level since 1990 with the following results:

The regression on these 2 variables alone explains 63% of the variation in the VIX premium, and both are statistically significant. The signs of the coefficients also corresponds to intuition. A higher level of realized volatility shrinks the premium (vol mean reverts), as does a positive return on the SPX. Applied to the current realized volatility of 21.3% and -0.1% SPX return (as of close 7/29) gives us a predicted VIX of 24.6% compared to the current spot reading of about 25%. Therefore, it looks like the VIX is fairly priced and not artificially depressed.

Welcome!

Welcome to my first attempt at blogging. My intention of this blog is to share my market analysis and insights, rather than directly providing buy or sell signals and hopefully build a community dedicated to sharing their ideas and studies. The focus will be macro and cross market: including major indexes, volatility and credit. My main approach to trading is to use statistics to prove - or disprove - the validity of historical patterns and setups to hopefully gain a glimpse into which side the edge is on going ahead.