Thanks to everyone for the amazing feedback for the previous post.
The question that triggered off this analysis was “Has the NIFTY moved too far, too soon and is a major correction imminent?” If you’ve heard any of the analysts on TV, you would know that there is a rare fund manager or analyst who believes in this rally.
This post is an attempt to look back at NIFTY’s history to see what it can tell us. This post is not about Technical Analysis, but pure and simple data analysis. For the record, you know of my views of the move from my previous post. 😊
The Tools for the Analysis
- 20WSMA (20 Weekly Simple Moving Average) is something I’ve used for a while now.
- Upmove duration is defined as time spent by NIFTY for any duration >= 10 weeks above the 20 WSMA. It is not time from bottom to top; but rather the time duration between retests of 20WSMA. The filter is a move that lasts for >= 10 weeks.
- Downmove duration is defined as time spent by NIFTY for any duration >= 10 weeks below the 20 WSMA. It is not time from top to bottom; but rather the time duration between retests of 20WSMA. The filter is a move that lasts for >= 10 weeks.
- % move is the difference between the last point at which the 20WSMA was crossed and the following top or bottom before a retest of the 20WSMA.
- One bull or bear market move could encompass multiple tests of the 20WSMA, but for the purposes of this analysis, each of the tests or retests is considered a separate data point.
- All the selections e.g. 20WSMA and 10 weeks are arbitrary and my choices, you can do the same analysis on any moving average or any duration.
- Numbers are close approximations, because I took the data from the charts rather than any database.
- Data is up to the 30th April 2017.
Back to the Analysis
Since 1991, there have been 28 instances where the NIFTY has spent >= 10 weeks above the 20WSMA and 24 instances where the NIFTY has spent >= 10 weeks below the 20WSMA. Some simple data points,
- In total amongst these 52 instances (including the current one), NIFTY has spent 978+ weeks in trend i.e. 18.8 years of 26+ years. Who said the markets trend only 30% of the time?
- The averages are (skewed by outliers of course)
|Parameter||1991 – 2017||2007 – 2017|
|Number of Upmoves||28||10|
|Average Upmove %||39.71%||29.73%|
|Average Upmove duration||21.67 Weeks||23.44 weeks|
|Price move per week – Upmove||1.83%||1.23%|
|Number of Downmoves||24||8|
|Average downmove %||-21.39%||-19.24%|
|Average downmove duration||15.43 Weeks||12.87 weeks|
|Price move per week – downmove||-1.46%||-1.38%|
The crux of the analysis
Now that we have some idea of the overall picture; let’s get down to the key elements of the analysis. Some of the questions that we’ll attempt to answer are:
What kind of rally is this?
- This is the 4th slowest rally since 2007 and the 6th slowest overall with a change of 0.88% per week. Why does it matter? Slow rallies are, generally, followed by less damaging corrections. As a side point, this is the slowest rally, ever, that began between December – February.
- Upmoves starting in December – February were comparatively common i.e. 9/28 overall; but this is only the second since 2007. They’ve become a rare-breed since 2007!! Also, such moves don’t last long i.e. average duration is 15.4 weeks; but most data is prior to 2007, so probably not that important an indicator.
When do the probabilities of this rally ending, begin to rise?
- NIFTY, very rarely, makes an intermediate top in Q2. The occurrences are only 3/28
- There is a good chance that one of April – August (at 1 in the above chart) will change. The question is which one? This is a tough one and your guess is as good as mine.
- 40% of all Upmoves have ended in 17 / 21 – 23 weeks but the Modi moves have lasted 32 (2014) and 27 (2016) weeks respectively. The average upmove duration since 2007 is 23.45 weeks.
- The mean and median duration all Upmoves which created an ATH are between 22 – 24 weeks.
- The mean and median duration of all the Upmoves is 21 – 22 weeks.
- So, here goes the educated guess
- If we hit the upper-end of the 22- 24-week mark (Modi rally tendencies), there could be a top in another 6 – 8 weeks. That means a top in late-June or more probably July.
- That would also mean 5-6 consecutive positive monthly candles; which is something that the NIFTY has tended to do in the recent past.
In the meantime, how much further can this rally go?
- Assuming we carry on with ~0.85%-0.9% move per week for another 6 – 8 weeks, we should have another 6-7% upside, for a rally of 17%+; which is also the median since 2009 i.e. the last bottom.
- Need to remember that of the 16 Upmoves (since 1991) which made an ATH, only 3 have ended with a rally < 20%.
What kind of correction we’re looking at?
- A substantial one. 2/9 corrections, following similar moves, ended at the 20WSMA. Others just stopped there to take a bit of rest, before driving further down to test the lower Bollinger Band. These numbers are dynamic, but as of today that is a 12-15% kind of correction.
- Is there anything good about this? These were all slow corrections with an average fall of <= 1% per week; hence the correction could last upto 3 – 4 months.
- The slower the upmove and the smaller it’s quantum, the lesser will be the magnitude and speed of the correction.
The Alternate Scenarios
In the above analysis, I just assumed that the NIFTY will hit the median upmove % and duration? What if that doesn’t happen before we retest the 20WSMA? That gives rise to 2 scenarios
- Scenario A: The move is almost done (Lower probability, IMO, than what we’ve discussed above)
- In this scenario, I would expect more ongoing consolidation with a smallish fall and a decent chance for a reversal from or just below the 20WSMA.
- The slowness of the upmove would preclude a major and steep fall, unless the world goes to hell.
- Scenario B: The move has a long-way to go (Lowest probability, IMO) before the retest of 20WSMA say 10.5K or upwards
- I can’t figure why this would happen and how this would happen but this would mean a steepening and speeding up of the move at some point in-time with some interim consolidations.
- There have been only 3 NIFTY rallies longer than 30 weeks; so such a move can’t be a grinding upward move. Of-course once the 20WSMA or the 50WSMA is retested, then another rally is always possible to take out all possible levels.
What does the VIX Say?
I started looking at VIX monthly data, looking for cyclicality and I found some cyclicality; just not as strong or as much as expected. For the record, there is only weakish correlation between a rising VIX and a falling market. With that in mind, the VIX patterns that I’ve seen are
- Dies post-budget into somewhere between April – July.
- Rises upto make a high somewhere between July – October.
- Dies into December – January.
There is almost no guarantee that this will hold; but the VIX is bound to rise sometime during H2. The current fall has been unprecedented e.g. this is the first time ever, IndiaVIX hasn’t touched 15 for two consecutive months.
I’ve listed down a few scenarios based upon how I’ve interpreted the historical data. It is possible that none of these turn out to be true or that Trump does something or Le Pen wins or Merkel loses or there is a war with North Korea or another black swan event happens and the markets just crack.
Do keep in mind that as the data changes over the next few weeks and months, the probabilities in favor of and against these scenarios will also change.
I am reachable on twitter @jdsfinance
This is simple data analysis, done to satisfy curiosity and answer some questions and should NOT be treated as any kind of investment advice.