The Council of Mortgage Lenders is bang on when it says it is futile to try to update its house price forecast.

Frankly it is probably sufficient to say that houses are worth less than they were a year ago and in a year’s time they will probably be worth a lot less – that is on average.

We are experiencing a market correction and no one really knows how fast it will take place and to what place it will take prices. As mentioned before the speed of any correction influences the likely depth of any fall.

But why not try to forecast what prices might be in six months or a year?

There are two good reasons I can think of why it’s not worth the effort.

One reason is pretty obvious. The market is so turbulent that you can make seemingly realistic assumptions today on the strength and direction of forces influencing house prices that could easily end up looking extremely foolish with hindsight a week or so later. So a forecast in the current climate is pretty much a wild stab in the dark.

But there is a not quite so obvious reason. Exactly what are you forecasting against? Take your pick of any of a number of indexes and you can end up with wildly different views of “actual” house price movements.

Let’s take a fairly extreme case and compare Halifax with the index produced by Communities and Local Government.

In February the average house price measured by Halifax was £193,448 and CLG measured it at £217,089. Not that far apart then.

In July the Halifax average price stood at £178,440, while CLG put the average house price at £217,171.

So, according to Halifax house prices dropped by 9.3% with the average house about £18,000 cheaper over the six months, but according to CLG house prices rose by a fraction, with the average worth an extra £82.

You could easily dismiss this as “bonkers” statisticians needing to get out into the real world. You may have a point about getting out into the real world, but I challenge you to sit in your office and devise a house price index that holds its rigour whatever the economic weather.

The builders of indexes have to come up with what they feel is a robust way to take sample data and build from that a realistic average or set of “type” or “regional” averages.

Why do they need samples? I hear you ask. Why not use the Land Registry data? Well, leaving aside its tardiness and some questions over accuracy, the answer is that it is itself just a sample, not the total population.

For obvious reasons, not all homes are on the market and sold all the same time, so the statisticians have to interpret what the price of those homes not being sold might be. And there will be a bias in the sample of homes on the market, if we compared it with the total population of all homes. One simple reasons is that those on the market always contain the entire population of “new homes”, more or less.

Anyway, relatively speaking all is fine in a normal economic climate, most of the relationships will hold.

But at present you have huge economic disturbances impacting variably on the market. So the coefficients, equations, weightings or whatever they use to create the average from the sample are thrown about as the shape of the sample radically alters.

Why should the sample radically alter? Well because the credit crunch is impacting on some groups harder or earlier than it is on others. But more than that different groups are able to buy broadly the same house at different prices.

The phase “cash buyer, lucky you” is used significantly more now than it was in the heat of the boom. You get a bargain ahead of the heavily mortgage backed buyer. Same house, different price.

And further, if your LTV (loan to value ratio) is high then expect a lower valuation on the house from the mortgage company. Why? Because they see the market dropping and want to steer well clear of having a loan on a property worth less than the property. So homes bought by those a high LTV will probably be relatively cheaper. Same house, different price.

There will be many other similar quirks impacting on particular regions, classes of buyer and seller, classes of home that have been thrown up by the craziness we are in.

How the classes of buyer, seller, home or region are weighting within the model will in large part determine the average. More importantly they will determine the direction and scale of change.

So you are the statistician plugging data from the sample into your model in the hope something meaningful will pop out at the other end, what are you thinking? Probably I hope not too many people ask too many tricky questions.

More to the point, if you can’t work out what a meaningful figure for the average house price or the scale of any price movements, how on earth are you meant to forecast what an average house price will be in at any point in the near future?

As Donald Rumsfelt said in one of his smarter moments, there are known knowns, known unknowns and unknown unknowns.

Well we live in a world filled with plenty of unknown unknowns, so forecasting – at least short term – is a mugs game. And I like forecasts.