Tuesday, June 30, 2015

Time for a market bounce?

I have written about my Trifecta bottom spotting model before (as an example see Sell Rosh Hashanah?). It consists of the following three conditions, which must occur closely (within a week) of each other:
  1. VIX term structure (VIX/VXV ratio) inversion
  2. NYSE TRIN more than 2
  3. My favorite intermediate term overbought-oversold model, which consists of the ratio of % of SPX stocks over their 50 dma/% of SPX stocks over their 150 dma less than 0.5
Each of these conditions, by themselves, are indicators of market fear. When used conjunction, however, they have been uncanny at calling short-term market bottoms.

The chart below shows the record of this Trifecta model in the last three years. I have marked in blue vertical lines where conditions 1 and 2 were satisfied, but not 3; and in red where all three were satisfied. In every one of the cases in the last three years, we have seen a short-term market bottom.

Trifecta Model: 100% accuracy in the last three years


We saw conditions 1 and 2 triggered yesterday, on Monday June 29, 2014, though the overbought-oversold model was not.

While the three year track record of this model has been remarkable, we need to put the results into context as the stock market has been rising steadily since the 2011 bottom. Such an environment is highly conducive to a class of models that spot oversold conditions, without worry that the market will get even more oversold.

I stress tested the Trifecta model in 2010 and 2011, when the market suffered major corrections and was far more choppy than it had been in the last three years. Though the results were still good, it was not perfect. The chart below shows the corrective period in 2010. In one case, the model signaled a minor rebound that petered out in two days. In another, it was too early as the market got even more oversold as the market fell to its final bottom about a week later.


We can see a similar pattern in 2011. The Trifecta model was too early in August 2011 as the market got even more oversold as it fell to its first bottom. September saw a number of multiple signals that were somewhat confusing. Though these multiple signals were consistent at flashing short-term bottoms, the stock market chopped around for several weeks before it began its ultimate rise.


So we can see that the Trifecta model is very good at flashing an oversold extreme, but does this represent THE BOTTOM for the current downdraft? That depends on a judgement call on the kind of market environment we are in. Are we still in the steadily rising market environment that we have seen in the last three years, or are we seeing the start of more serious market correction?


VIX spike bottom signal
Bill Luby at VIX and More had a slightly different take in a post where he wrote about stock market reactions to VIX spikes. Monday's 34% VIX spike was highly unusual and he found that it was generally an indicator of a short-term market bottom, but stocks tended to fall after a 3-5 day bounce (emphasis added)
Note that based on the data for the 23 VIX spikes in excess of 30%, the SPX has a tendency to outperform its long-term average over the course of the 1, 3 and 5-day periods following the VIX spike. Also worth noting that that 10 and 20 days following the VIX spike, the SPX has a tendency not only to underperform, but decline. Further, while the huge decline following 9/29/2008 VIX spike tends to dwarf the other data points, even when you remove the 9/29/2008 VIX spike it turns out that the SPX still loses money in the 10 and 20-day period following a VIX spike. When the analysis is extended out 50 trading days, the SPX is back to being profitable, but performing below its long-term average. On the other hand, when the analysis includes 100 days following the VIX spike, the SPX is back to outperforming its long-term average.

With the caveat that this is a limited data set, it is still worth flagging the pattern in which following a 30% one-day VIX spike, there appears to generally be a tradable oversold condition in stocks that lasts approximately one week, followed by a period of another month or so in which the markets typically has difficulty coming to terms with the threat to stocks. One quarter later, however, all fears are generally in the rear view mirror and stocks are likely to have tacked on significant gains.

Bill`s work with VIX spikes, whose history goes all the way back the the 1990`s, is suggestive that we are seeing a 2010 or 2011. If that`s so, it is consistent with the hypothesis that the market rallies into the July 4th long weekend and retreats afterwards.


The Greek referendum wildcard
We will see the results of the Greek referendum this weekend. The latest real-time signal from Lakdbrokes, indicate that the Yes side is leading with a probability of 63% compared to 37% for the No side. Such expectations are likely to create bullish tailwinds into the referendum on the weekend.

Even if the Yes side were to prevail, it is unclear how the markets might react to such a vote. Will it rally further because Greek tail-risk is off the table? Or will it retreat because the Syriza led government will likely collapse, which leaves Greece rudderless in the face a July 20 ECB repayment deadline which would implode its banking system (for more details see my post Time to buy GREK?).


Watching for the O'Neill follow-through
If the market were to see a bounce into the weekend, one way to decide if we have seen a durable bottom is to watch for a William O'Neill follow-through day, which is believed to be the sign of an intermediate term market bottom:
Summarizing the rules:

Day 1: Once a low has been established (after a correction), Day 1 occurs if the close is near the high of that day or a higher close occurs on the day after the low.
Day 2: The price must remain above the established low. If the price moves below the Day 1 low, then the pattern has been invalidated.
Day 3: The price must remain above the established low. If the price moves below the Day 1 low, then the pattern has been invalidated.
Days 4 - 7: Follow-through day must occur, with a gain greater than 1.7%, heavier volume than the previous day and heavier volume than average.
Traders often make the mistake believing that they can accurately forecast the future. A better approach might be to create a trading plan by keeping several scenarios. Define your risk parameters and pain thresholds, the potential reaction to developments under each scenario and act accordingly to the trading plan.

These markets are potentially treacherous. Under these conditions it especially pays to have a plan ahead of time.



Disclosure: Long TNA

7 comments:

Anonymous said...

Very interesting, although the early trifecta triggers would have been costly. Where does this stand as of today's close?

Cam Hui, CFA said...

As of Wednesday's close, the indicators have retreated off the deeply oversold readings seen on Monday. That's typical of what happens when the markets comes off an extreme oversold condition.

NFP Thursday morning - stay tuned!

Harapa said...

I will not go in till the Breadth and Volume improves

RAS said...

I was wondering if you could please provide more background on the interpretation of the VIX/VXV inversion? Thanks for posting this, recreating your chart gave me new insight on how to use stockcharts.com. I've now added a new capability to my analysis bag of tricks!
-RAS

Cam Hui, CFA said...

RAS - For more details on the VIX/VXV ratio please see http://humblestudentofthemarkets.blogspot.com/2014/05/worried-about-low-vix.html

RAS said...

Thanks, I reviewed the post and it was helpful. For purposes of your 3 indicator oversold condition finder, do you consider an inversion to be any VIX/VXV reading greater than 1, or is there a higher threshold level? Also, have you looked at the three indicators as a top finder? If so, and if you are willing to speak to it, what conditions would appertain?
Thanks again,
-RAS

Cam Hui, CFA said...

RAS - In answer to your questions:

1) Any VIX/VXV ratio of more than 1 is regarded as an inversion
2) I've found that these indicators have not been very good at spotting tops, only panic bottoms

Cheers,
Cam