To play teenpatti (three-card poker) or five-card western poker well, a gambler needs numerical skills, nerves and an understanding of psychology. All these attributes are also useful for investors.

Different groups of teenpatti players have different styles. Some games are very aggressive, with players 'blind' and bluffing. Other games are conservative -- everyone 'sees' early and packs without good cards.

An aggressive player in a conservative game gets killed because he insists on bluffing when his opponents have good cards.

A conservative in an aggressive game makes less money than he should because the others know he plays only with good cards. Successful teenpatti players adapt to blend in and thus maximise returns.

Blending in

A stock market investor requires similar skills to maximise returns. An optimistic investor gets killed in a conservative market because he buys high. A conservative investor underperforms in an optimistic market because he sells early.

Using theoretical valuations without reference to the characteristics of a specific market also leads to underperformance.

There are many textbook definitions of valuation -- most discount future cash-flows to arrive at current fair price. If the answer does not 'fit' a specific stock market, investors using this valuation model will not get optimal returns.

In the following, we try to judge the valuation characteristics of the Indian stock market. What is the Indian market's consensus estimates? At what level is there high buying interest (that is, a consensus estimate of under-valuation), and what level is there high selling pressure (a consensus estimate of over-valuation)?

Such a study is not idle navel-watching. A model built around answers to these subjective questions offers a reasonable idea of the returns an investor can expect at any given price. There are other interesting ways in which this data can be used to tweak standard models.

P/E - a useful benchmark

Stock markets are manic depressive and prices swing wildly. What we've done is to express Sensex movements in terms of average price-earnings discounts and then look for patterns in the manic depressive behaviour.

Does buying appear below certain P/E multiples and is there a key P/E level above which sellers dominate? If true, we have useful benchmarks.

Other ratios could be used instead of P/E -- price to book-value or price to cash earnings may be more 'accurate'. But P/E is an universally understood metric; it's easily available and widely tracked.

So it made sense to use P/E in terms of practical convenience. The data was accessed from the BSE website (www.bseindia.com) and the P/E used there is calculated on a weighted average of trailing Sensex-30 full-year earnings per share.

Higher prices, lower P/Es

'Annual highs' is a chart of the annual highs of the Sensex between 1995 and 2005 (till date). 'Annual lows' displays Sensex lows between 1995 and 2005. These two charts have prices and associated P/E ratios across the same timeframes. The third one 'high-low' charts the annual highs and lows only in terms of P/Es.

Some interesting trends are apparent. If we look at annual highs, we see that, although prices have moved up, P/E ratios have generally declined through this decade. Except during the Net fever of 1999-2000, the Sensex has never been valued at over 30 P/E in the last 10 years.

For example, in 1995, prices peaked at P/E 30 and 3943 Sensex. In 2004, at a high of 6617, the P/E was only 17-plus. In 2005, at an all-time high of 6954, the P/E was 16.6.

On an average, the Sensex has peaked at about 21 (ignoring calendar 2005). During the entire bull-run of 2003-05, the Sensex has never hit a high of 20-plus.

The same behaviour of lower P/E and higher prices is confirmed by annual lows. In 1995, prices bottomed at 2891 at a P/E of 22. In 2004, at 4227, the P/E was 11 and so far in 2005, at a low of 6069, the P/E has been 14.5. On average, the Sensex has bottomed at 13.7 (if we ignore 2005). It has hit single-digit levels once -- in 2003.

Looking at the third chart, the differential between the annual high P/E and annual low P/E has been greater than eight only twice, in the period after the Net boom went bust (2000 and 2001). The average annual differential has been 7.93 (ignoring 2005). Most of the time, the high-low differential has been six-eight.

The Indian market has definitely become more conservative since the early days of liberalisation (in 1992 P/Es hit 60-plus!) given the trend of lower P/Es. So it is prudent for investors to become more conservative in their approach than a decade ago. Otherwise they would buy at consensus estimates of 'high' prices.

Consensus estimates

The data suggests that the Indian market's consensus estimate is that it's undervalued at P/Es below 14, and it also 'thinks' it's overvalued above 21. Since the market hasn't climbed above 18 since 2001, our margin of safety for peaks would be around 17-plus. We could create trading rules such that we always buy below 14 and start selling above 17.

At the current Sensex level of 6606, the P/E ratio is about 15.8; so we are about 1.2 P/E below our 'safe' peak level. Since we are also about 2 P/E from our safe bottom level, the market seems to be in one of those very rare phases when it is fairly valued, more or less in the middle of the high-low band.

Upside in 2005-06

The upside in 2005-06 could be a lot. We are currently working on trailing 2003-04 EPS. If the annual high hits 17-plus P/E level, the market would be trading around 7105, which is just 7.5 per cent appreciation.

But there's the growth factor. The Sensex EPS will be recalculated once 2004-05 results come in. The consensus EPS growth estimates for 2004-05 are in a band of 20-30 per cent and the CAGR (compounded annual growth rate) for Sensex EPS since 2000 is 20.8 per cent.

Let's first assume 20 per cent EPS growth, as estimated in the Sensex EPS chart. This means that if the Sensex maintains its pattern of an annual high of 17+, it would peak somewhere close to 8600. At an optimistic EPS growth of 30 per cent and assumption of 17+ P/E, it would hit 9200.

Is there also a potential downside implied in terms of the 2004-05 EPS? At 13.7 P/E (the average annual low) of the 2004-05 EPS with an assumption of minimum 20 per cent EPS growth, the Sensex could hit a level of 6919 in the current fiscal.

That's well above the current level and, in fact, if one assumes an EPS growth of 20 per cent in 2004-05, the market is trading at about 13.

Astonishing, as it seems, if the market maintains its long-term trend of annual P/E highs and lows, there is a good chance that we are close to the lows for fiscal year 2005-06.

Other valuation models

Let's look at this in terms of standard valuation models. At a P/E discount rate of 14, (the low level of our suggested trading band) the implied rate of return is about 7 per cent through the rule of thumb inversion of the P/E ratio.

At a P/E discount of 17, the implied rate is just below 6 per cent. This band of a 6-7 per cent discount rate fits well with current interest rates, suggesting that the consensus market valuations are good.

But interest rate expectations are on the higher side for 2005-06. That means the market may be more conservative in its discounts in 2005-06.

Let's assume that the implied rate of return moves up to a minimum of about 6.5 and rises to a maximum of 8 for fiscal 2005-06.

This would translate into a high-low P/E band of 12.5-15.5 -- quite conservative in terms of previous behaviour.

After we've adjusted for a minimum 20 per cent EPS growth, our P/E low for 2005-06 would be 6300 at P/E 12.5. Our maximum return expectations pare down to 6.5 per cent, which is the equivalent of 7800 at P/E 15.5. This estimate still suggests considerable upside in 2005-06 at an appreciation of 18.5 per cent from current levels.

However, there are more pessimistic ways of viewing the same data. If there's a catastrophic rise in rates, the market might drop a lot. Its lowest discount in the last decade was 7.5 (2003) -- that would be 3800 in 2004-05 EPS terms or 42 per cent down. More likely, P/E would drop to 10-11 levels.

PEG

Optimists will suggest that there is room for higher long-term growth in view of the low P/E-EPS growth ratio.

In most markets, the rule of thumb is that fair valuation comes at a PEG of 1. That is, a market, which records 20 per cent EPS growth ought to have a P/E discount of 20.

The Indian PEG has usually been closer to 0.6 or 0.7, which suggests that on the PEG metric, Indian markets are consistently under-valued.

But that trend of dropping P/Es over the past decade also suggests it is better to be conservative and assume that the market will continue to be undervalued in terms of PEG. Most Indian investors keep a healthy margin for creative accounting methods!

Conclusions

• Indian investors have become more conservative through 1995-2005.
• The market appears fairly valued.
• On 2005-06 EPS growth estimates, there is room for appreciation of between 18-40 per cent (Sensex highs of 7800-9200).
• On the downside, a serious bear market could see prices dropping by between 15- 40 per cent. (Sensex 3800 -5550).