06.06.2013 Views

STOCHASTIC

STOCHASTIC

STOCHASTIC

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

392 GORDON PYE<br />

be the vector of price differences between the actual price in period 0, 1, ••• , t and<br />

the initial price. Let n,(p) be the n for the seller in period t. Let nb(p) be the n of<br />

the buyer in period t. From the definition of n for the seller and buyer it follows that:<br />

«»(p) - Pi — MinoSTS,Ipr] = MaxoSrS, [-pT] - (-p.) = n,(-p).<br />

Since — p is the mirror image of the price differentials about zero, the stated proposition<br />

has been proved.<br />

It has now been established that the amount sold by a seller and the amount bought<br />

by a buyer for any n and t are equal, and that nj(p) = n,( —p). Initially, a buyer will<br />

want to buy the same amount of the asset as a seller. Subsequently, however, if the<br />

price rises n, will continue to equal zero while ra& will rise. Thus, a buyer will want to<br />

buy more while a seller will not want to sell more. The converse will hold if the price<br />

falls. Behavior of buyers and sellers will be symmetrical. In the example in Figure 1 the<br />

minimax behavior of a buyer will be given by the mirror image of each of the series.<br />

In other words, the amounts by the nodes when the figure is rotated 180° about the<br />

horizontal axis will give the quantities which a buyer will purchase. It is interesting to<br />

note that sequential, minimax policies will tend to have a reinforcing effect on price<br />

movements. As has been seen, price rises will make buyers want to buy more without<br />

causing any increase in sales by sellers.<br />

6. Policies under Reversibility<br />

So far it has been assumed that the seller is prohibited from repurchasing any of the<br />

asset. This assumption will now be relaxed to permit repurchases at any time before T.<br />

Margin purchases and short sales, however, will continue to be prohibited. As discussed<br />

in the introduction this has the effect of changing the flavor of the problem to one of<br />

multiperiod portfolio selection. However, it is unconventional in that the objective<br />

continues to be one of minimaxing regret. Under these conditions the problem does not<br />

seem to replicate situations in which dollar averaging has been recommended. Nevertheless,<br />

the problem is of interest because the answer is simple and forms an interesting<br />

contrast to the results obtained previously.<br />

The possibility of repurchase changes the maximum amount which can be obtained<br />

for the asset, given the price series which occurs. Without repurchase the best one can<br />

do is sell out everything at the highest price which occurs. In general, with repurchase<br />

one can do better. The best that can be done is to be fully invested during every price<br />

increase and completely disinvested during every price decrease. Thus, the amount<br />

obtained for the asset will be equal to the initial price plus the sum of all the price increases<br />

which occur. The expression for regret in (3) must be changed to read as follows:<br />

(15) R = S+ Ap

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!