Selling bundles

Is there some reason that you can't sell bundles? The "Buy As A Bundle" interface won't allow you to enter a negative number.

If you are, for example, long stock A and short stock B which make up all or most of a bundle (e.g. US.REPUBLICAN12 and US.DEMOCRATIC12) then you can convert to a position that is purely long stock A by buying bundles, but you can't convert to a position that is purely short stock B by selling bundles.

Why not?

This can be important when (as often happens) there are large orders in the book for one stock but not the other.

I can give you an example of what would happen if you were able to sell bundles; this may be the reason that the option doesn't exist.

Say there is a bundle of 2 mutually exclusive contracts, and the highest buy price for each contract is 60c. Two mutually exclusive contracts each trading at 60c gives a total odds of 120%, which is of course meaningless. Using bundles, it is possible to buy equal amounts in both stocks for a total of $1.00, and then sell them at the 60c price to the highest buyer, until each stock is driven down to 50c (or lower).

The flipside of this is that if two stocks were trading with 40c being the highest sell price for each, the total odds are 80%. This actually isn't a problem per se, it just indicates that there isn't enough information present in the market for people willing to buy up contracts. But if we allowed selling of bundles, then it would be possible to sell a bundle and then buy back the stocks at 40c each (giving a return of 60c + 60c = $1.20) until the stocks reached 50c each.

Now, this second scenario isn't a problem in itself, but it means we could never have a situation where a bundle of stocks was trading at a total below 100%. The market would become less representative of the state of known information by artificially increasing the odds of all constituent stocks until the total reached 100%.

In your second scenario you can already buy both stocks for 40c, hold on to them until the contract closes, and make a 25% return. This will also quickly drive the total price to $1.00.

The only difference is that the ability to sell bundles to the house will free up that capital immediately instead of having to hold the stocks until maturity, thus increasing liquidity in the market.

I'm unsure why you'd regard a total price of less than $1.00 as being somehow more representative of the state of known information. Especially given that the situation will not last long in any case.

I don't really see how selling bundles is any different to buying them. Allowing bundle selling would put more cash in our pockets and improve market liquidity.

maxb - neither, I just have to assume the fact that we don't already have the ability to do so was due to a deliberate decision when the site was first set up. Now it's possible that there was some implementation limitation for the feature (and it was put on a 'to do' list), but I suspect there's some other underlying reason for it.

maxb Wrote:
-------------------------------------------------------
> I don't really see how selling bundles is any
> different to buying them. Allowing bundle selling
> would put more cash in our pockets and improve
> market liquidity.

+1

With binary prediction stocks, short selling is no different than buying the opposite, i.e. if you short sell a "no change" option for the OCR, that's the same as betting that the OCR will change.

Looking at it from the mathematical point of view, selling is the same as buying a negative number of objects. Of course you can't own a negative number of physical objects like books, but you can own a negative number of abstract objects like shares in prediction stocks.

So I don't see any substantive difference between buying and short selling. You're betting one way or the other that an event will or will not occur.

To be honest I'm not really sure what's in it for ipredict in offering the "bundling" option, they don't get any trading fees out of it. Maybe they want to help people cash up so they can do more trades and generate more trading fees. Regardless, it doesn't make much sense to offer bundling for "positive" trades but not "negative" trades. I should be able "buy" a negative number of bundles.

Presumably if you buy -1 bundle you still pay $1 (since the maximum payout of shorting everything is the same as buying everything: $1).

So now I'm trying to wrap my head around the absurd situation of buying 1 bundle (cost: $1, portfolio: 1 of each stock) and then buying -1 bundle (cost: $1, portfolio: no stocks). Total cost: $2, portfolio: nothing.

You could have some exception for if you are long when you sell the bundle, but what happens if you hold a mix of long and short positions?

Perhaps it is easier to look at it as owning positive shares or owning negative shares.
Mathematically speaking, there is no difference between buying or shorting stock on iPredict. Shorting a stock is equivalent to buying stock of the opposite outcome. Either way you are holding contracts which you can make you profit or loss.

Quote
MaxTheCreator
Perhaps it is easier to look at it as owning positive shares or owning negative shares.
Mathematically speaking, there is no difference between buying or shorting stock on iPredict. Shorting a stock is equivalent to buying stock of the opposite outcome. Either way you are holding contracts which you can make you profit or loss.

This doesn't really answer how my scenario is fair or sensible, though.

If you buy a stock and sell a stock at market (or sell and then buy, no difference) you reach a zero stock position and (if you managed to buy and sell at the same price) a zero cash position. Doing the same with a bundle leaves you $2 down.

If you short a bundle of n stocks, the payout would be $n-1, not $1 (as all but one will close at $0).

Suppose there's a bundle of 5 stocks, and you have one share of one of those stocks. You then short the whole bundle for $4, except you then own +1 share and -1 share of the same stock. However this stock closes, you'll make $1 (if it closes at $0, you make $1 from the short share, and if it closes at $1, you make $1 from the long share). These can therefore be cancelled out, paying out $1 immediately, in effect shorting the bundle for $3 and ending up with 0 shares of the stock you previously had a share in, and -1 of the others.

In summary: For a bundle with n stocks, x of which you are long in, shorting a bundle should cost $n-x-1. In your scenario, the cost would be $1 to buy a bundle, and -$1 to sell it (since n=x), resulting in no change to your net worth.



Edited 1 time(s). Last edit at 11/19/2012 02:41PM by hmonkey.

hmonkey Wrote:
-------------------------------------------------------

Quote
hmonkey
If you short a bundle of n stocks, the payout
would be $n-1, not $1 (as all but one will close
at $0).

This is the brainfart I was having. I knew there had to be one. Thanks.