You can also have a conflict between those numbers. For example, the financial sector contains more actively traded companies than textiles, refiners and white goods taken together. So you could have a day when financial stocks move up while textiles, refiners and white goods drop. There would be a healthy advance decline ratio but the sector breadth would be negative.

Depth means both the number of liquid stocks as well as the volumes registered along with price shifts. If say, a stock (or an index) is trading 10 lakh shares at Rs 100, how many shares will trade at Rs 120 or Rs 80? Impact cost is an inverse proxy for depth - the lower the impact cost, the deeper the market.

Diversification generally means diversification across sectors. It could also mean diversification across asset classes. And it may mean diversification in size - large caps versus small caps since these often move in different directions. Ultimately any portfolio that pulls risks down without impacting returns negatively is usefully diversified.

The last three years have been the broadest bull market in Indian history, whatever the criteria you use. The advance-decline ratios have been positive and the sector counts have been positive. This is also the deepest market ever.

Again, this is true whether one is talking about high volumes alongside price shifts or about liquid stocks. The rapid expansion of the derivatives segment is a good pointer.  There are now about 225 stocks that can be hedged in F&O.

Random diversification across sectors has worked well in the past three years. Gains have come across the board. This has been reflected in institutional holdings. Both FIIs and mutual funds now hold relatively smaller scrips. And of course, the unit linked insurance plan (Ulip) corpus has also been spread out across sectors and stocks.

There is a feeling breadth will narrow over the 12-18 months. We may still have a bull market through 2008-9 but it will be increasingly driven by a few specific sectors. Power, metals, energy and construction seem the favoured areas with financial services and telecom also expected to chug along.

Telecom has already taken a hit due to the spectrum mess. Don't underestimate the government's ability to continue slowing things down. But the market is likely to continue growing at 8 million new subscribers a month even if 3G is held up and service quality deteriorates. 

The perception on financial services in fact, is mixed. Domestic rates seem to be coming down and more bank FPOs are guaranteed due to Basel II norms. But the dollar could do unpredictable things and there could be a bust or at least a dramatic slowdown in the housing market. Brokerages could also hit a consolidation phase.

The "killer app" in power is hopes that reform will continue. I wonder how far the electoral scenario will affect this? State governments have an irritating habit of offering free power whenever there's a tight election in prospect. In a 3-5 year scenario, the sector will either do well - or it will go completely bust.

Metals are driven by strong global demand and increasing scales in Indian operations. The ability to integrate up from captive mining to captive power to metals will fatten operating margins. The risk is that America will go spectacularly bust and take the commodity cycle down with it.

Energy awaits the flood of gas expected to start flowing by 2009. It also waits for reform and freeing of prices. Don't expect that before the general elections. Engineering and Construction of course, will continue to register huge orders throughout.

Of these favoured sectors, the only "perennials" seem to be engineering and telecom. Power, financial services, metals and energy could see wild fluctuations through the next 12-18 months, though all have good prospects in the longer term.

Given the obvious risks and long timeframes, the "hot" sectors all seem overvalued. Does it make sense to go contrarian and buy into "cold" sectors such as FMCG, automobiles and IT instead? In the 12-18 month time frame, these cyclicals are likely to recover. And they're relatively cheap.