Testing a product to destruction, to the outer limits of tolerance and beyond, is a recognised quality control technique. It's more unusual to test a service paradigm to the snapping point. But, such a test of Indian trading systems occurred naturally on Monday. Things nearly fell apart but finally the system held -- just about. However a lot of anomalies came into sharp focus.
The first and most obvious imperfection is the non-aligned nature of trading on the NSE and BSE to name only the two major exchanges. The NSE settles on Tuesday, and begins a new account on Wednesday while BSE settlements close on Friday. Then again, BSE traders can cheerfully carry forward using badla as a mechanism while NSE traders have to settle. The circuit-breakers are different for the same scrips on the two exchange because of different closing days. Also trades cannot be switched with ease from one exchange to the other. Plus, there are often days when one or the other exchange is closed but not both.
All this means that on key swing days, the two exchanges can show radically different trading patterns and levels. This is despite the fact that the longterm correlation between BSE and NSE is very high (around .86) which means they rarely move in different directions. However, the beta or the coefficient of correlation is a different matter. The NSE Nifty has a beta of 0.80 which shows distinctly lower volatility than the Sensex and, in general, the Sens-ex has magnified moodswings.
The combined charts of Nifty and Sensex daily trading normalised and superimposed over the last two years to a base of 100 points in June 1996 shows both the general correlation fit and the greater Sensex volatility. Both charts especially the chart of intra-day volatility clearly shows the Sensex often moves through a greater trading range.
In statistical terms, the Sensex has an average intra-day range of 1.92 per cent of its opening price while the Nifty averages 1.90 per cent. The standard deviation of the Sensex intra-day range is also higher at 0.997 while the Nifty daily range has a standard deviation of 0.967. Those little second decimal differences arise from large changes on certain days.
Superficially, that should lead to huge arbitrage opportunities, but the arbitrage process is inefficient because a trade made on one exchange cannot be switched on the other and also because of the different settlement periods. So, arbitrageurs cannot gain and increase market efficiency simultaneously.
Days like Monday are rare however. A combination of several circumstances led to spectacular differences in intra-day price and closing levels. Trading volatility on NSE and BSE both is restricted by circuit breakers which cut in at +/- 10 per cent daily and allow a total of 25 per cent movement per week on previous week's close. But BSE will take Friday's close and NSE will take Tuesday's close for circuit filters making the difference.
On Monday, NSE trading was further restricted by settlement considerations and the Nifty swung through a range of 66 points during the day (6.7 per cent of Friday's closing price) and the circuit breaker was clamped on 208 stocks. The Sensex in contrast swung through an amazing 236 points (6.83 per cent) but the circuit breaker clamped down on only 124 stocks. In the end, the Sensex however recovered to come back to within a single point of its previous closing levels, while the Nifty ended down by minus 2.51 per cent. Settlement considerations again prevailed -- bulls were unprepared to take single-day open position on NSE, but could speculate for four sessions on BSE.
On Wednesday, the pattern of volatility was further reinforced by rumours of default of several brokers on the BSE. Many stocks such as Pentafour, BPL, Satyam, Sterlite slid down to the bottom end of their circuit breakers. BPL, Videocon and Sterlite lost 25 per cent of their value in three sessions and further selling would be prohibited this week. The BSE had 38 scrips htting the daily filters while 15 hit weekly filters.
On the NSE, in contrast, 126 scrips hit the daily filters because Tuesday was subdued. This restricted Nifty losses to 3.23 per cent while Sensex saw much larger losses at 4.51 per cent. The volatility was once again lower on NSE.
Given this trend and the limitations on efficient arbitrage, what strategies exist for speculators and investors to exploit? One obvious backroom check is to always compare prices on both exchan-ges before making a trade. That a surprising number of people don't can be inferred from the fact that volumes don't usually transfer from one exchange to the other when there is a large price differential!
Second is the simplistic arbitrage opportunity of delivery-backed buying on BSE on Friday and selling on NSE on Tuesday. A similar three-session opportunity can be created by taking delivery on NSE on Tuesday and selling on BSE on Friday. In the midst of a strong uptrend, this can be a near-risk free arbitrage with cash locked in for a very short period. Unfort-unately a reverse short-sale strategy is tough because of lack of carry-forward on NSE and that trades cannot be switched.
A third point is that delivery-backed purchases in midcaps and smallcaps should be done on NSE if possible while profit-taking should be indulged in on BSE. Greater volatility on BSE usually means that stocks rise more there during an uptrend.
When it comes to more speculative methods, the first problems occur with identifying stocks in real time. Very few online trading/technical analysis packages are available and they are quite expensive. Bridge is one of the best packages and Metastock also has an online version. The tools of intra-day technical analysis are quite often borrowed from forex trading. However, chaos theory suggests that the standard tools would also work very short term except that time-periods would need much finetuning.
Another possible angle that needs exploration is a long term statistical analysis of individual stocks to establish differences in trading range averages on NSE and BSE and also to establish relative betas. If a group of such scrips can be isolated then both trader and investor can establish profitable strategies until such time as the current systems are smoothed out. Once index futures get established on the Nifty, there will also be other rich arbitrage opportunities.
We looked at two scrips listed on both BSE and NSE and see lots of action. One is Reliance, a big boy on both exchanges. The other, Krishna Filaments, has seen action since Harshad Mehta extolled its virtues. The patterns are interesting.
Reliance: The stock has a .994 correlation between its trading patterns on NSE and BSE. However NSE trading volumes are higher and the volatility is somewhat more on NSE. The NSE Reliance scrip has a beta of 1.02 versus BSE over the last two years -- the stock moves slightly faster in the same direction.
The average trading range on NSE is also slightly higher at 3.34 per cent daily versus 3.29 at the BSE. The scrip has also gained slightly more in the last 2 years on the NSE returning 26 per cent versus 125 per cent. This implies that the investor ought to be doing his profit-taking on NSE and buying on BSE during a perceived uptrend and vice versa during a downtrend.
Krishna Filaments: This is a scrip with much smaller liquidity and much more volatility than Reliance. It had returned close to 140 per cent on both exchan-ges over the last two years. In the last year, it has returned around 212 per cent (as of June 9, 1998). The BSE-NSE correlation is very strong -- around .98. The NSE registers higher volumes but BSE registers more volatility. The NSE intra-day range is around 2.45 per cent of opening price while BSE intra-day range is around 2.60 per cent. The NSE has a 0.98 beta which implies slightly slower moves. An investor should probably be buying on NSE whereas a speculator would get better returns trading it on the BSE.
These studies are merely indicative and in no way exhaustive. Much imaginative data-basing could be done to identify and target high-volatility stocks which show significant beta differences and large differentials in intra-day range. Many interesting strategies could be worked out to exploit those differences. The patterns appears to be highly susceptible to programmed trading.
