# Unit price models based on demand

A question for the economists and statisticians out there:

Is it possible to (semi-)accurately guage the anticipated price of an asset based on the current price and demand? For instance, at the current volume of exchange trading, one peercoin is worth ~\$0.90. If \$50,000 of new investment were to be added to the market, the price would likely go up.

Are there models that can be applied that give you a range of what that new price point would be? I know that I’ve seen something similar on BitcoinWisdom’s charts that show the cost required to move the price point from X to Y, but I’m not sure if that’s the same thing that I’m describing (though it might be a component of it).

Is this a futile excerise? Or is it something that exists already for other types of assets and could potentially be adapted for cryptoassets?

I think I know why you might be asking.

The short answer is “no” – you can’t know for sure. Let’s say the price is currently \$0.90, and \$50000 of new investment is coming in. What’s not known here is, how much of the asset people are willing to sell at \$0.95, but not \$0.91. It could be none, or it could be an enormous amount. If it’s none, the price may push upward past that point, and never return. If it’s a lot, and the sell orders are not yet in the order book, the market would still look identical, so there’s no way to detect this “market sentiment”. After the buy order, these sellers would step in and push the price back.

You could estimate this, based on market depth, but there’s no exact science to it, as far as I know.

That was my assumption, but I was hoping that there was some esoteric branch of economics that dealt with these models.

Oh, well, if anyone comes across one, or develops a scheme that appears to work, I’m interested in learning more.

I think this is the crux of the issue. Some markets are thin, but inelastic, with traders stepping in as needed to maintain a trading channel. Other markets are highly liquid, but more elastic and volatile. You can’t tell this by looking solely at the order book.

well, on irc gribble responds to ;;market sell or buy commands:

;;market sell 10000 gribble: Bitstamp | A market order to sell 10000 bitcoins right now would net 4642647.0832 USD and would take the last price down to 400.0000 USD, resulting in an average price of 464.2647 USD/BTC. | Data vintage: 56.4311 seconds ;;market buy 10000 gribble: Bitstamp | A market order to buy 10000 bitcoins right now would take 5703435.3023 USD and would take the last price up to 634.7000 USD, resulting in an average price of 570.3435 USD/BTC. | Data vintage: 85.7678 seconds

idk how gribble does it exactly

That algorithm simply looks at buy and sell orders that are already in the order book. It correctly predicts “the last price” after a trade, but this ignores any correction after a large market order. Just because the last price goes to 400 USD after a whale sell doesn’t mean that the price will stay near 400 USD.

Ben, you are asking a billion dollar question. You can get some quick-and-dirty answers here in the forum. But not much more than that. Everyone has his model. Every model has some accuracy and corresponding uncertainty. If you act according to a model you have to face the risk associated with its uncertainty. Wall Street companies hire the brightest mind to come up with models with less uncertainty for themselves. Those brightest minds, like that of my college classmate who went to MIT for his physics PhD , discovers complex models that the marke department sell to customers who had not prepared to understand.

Ah, but you see, this is for the Community. It’s open source!

Think of the children!