A Short-Sale Arbitrage in a Three-State Market
Summary
The document describes a one-period market with three states and three risky assets. Each asset costs one initially and pays two only in its corresponding state, while the setup assumes no interest. The question is whether the absence of a risk-neutral probability measure implies arbitrage and what portfolio realizes it.
The accepted answer sells one unit of each risky asset, receiving three at the start, and buys two units of the risk-free asset. At the end, the risky positions owe two in every state, exactly offset by the risk-free holding, leaving a certain initial gain of one. This illustrates the state-by-state payoff test for arbitrage. The construction relies on the stated prices, payoffs, and financing assumptions; it does not address transaction costs, short-sale constraints, or other market frictions.
Key ideas
- Each risky asset pays only in one of the three possible states.
- Shorting one unit of every risky asset yields a total initial receipt of three.
- Holding two units of the risk-free asset offsets the combined future liability in every state.
- The portfolio produces a positive initial cash flow and zero net future payoff under the stated assumptions.
Tags
Full text
# What is the arbitrage opportunity in this simple one-period market?
# What is the arbitrage opportunity in this simple one-period market?
I have a single period market, and three states, and I have 3 risky assets. I assume no interest.
So I have three states $\Omega=\{\omega_1,\omega_2,\omega_3\}$. All assets start with the value 1, and for each n the price of risky asset n, is 2 in $\omega_n$ and the other assets have value 0. That is, the price function is:
$S_1(0)=1,S_2(0)=1,S_3(0)=1$
$S_1(1,\omega_1)=2,S_1(1,\omega_2)=0,S_1(1,\omega_3)=0$
$S_2(1,\omega_1)=0,S_2(1,\omega_2)=2,S_2(1,\omega_3)=0$
$S_3(1,\omega_1)=0,S_3(1,\omega_2)=0,S_3(1,\omega_3)=2$
Now, the theory says that there is no arbitrage iff there exists a risk neutral probability measure.
From what I see there is no risk neutral probability measure, because then we would have to have:
$\begin{bmatrix}2 &0 & 0 \\0&2 & 0\\0 &0 &2 \\ 1&1&1\end{bmatrix}\begin{bmatrix}q_1\\ q_2 \\q_3\end{bmatrix}=\begin{bmatrix}1\\1\\1\\1\end{bmatrix}$.
And this set of equations have no solutions, hence there should exist an arbitrage opportunity?
However, I can't find a strategy which guarantees money, I have tried to find it numerically, but wasn't able to find one. Do you see the strategy or have I made mistake somewhere?
## Answer by Brownian3 (score 5, accepted)
https://quant.stackexchange.com/a/25114
Sell 1 unit of S1,2,3 respectively, gain 3; buy 2 units of risk-free asset, cost 2.
No matter which state appears, the future payoff/loss is 0 for sure, while you will gain 1 at the beginning.
## Answer by Steinwolfe (score 0)
https://quant.stackexchange.com/a/25101
I have not studied this, but intuitively...
If all of the states had equal probability of occurring, then the expectation of each asset $E[S_n] = \frac{2}{3}$. Imagine instead the probability of $\omega_1$ was $0.5$, with the other two $0.25$. Now $E[S_1] = 1, E[S_2]=E[S_3]=\frac{1}{3}$.
It seems to me, irrespective of the probabilities of the states, the expectation of the equal weighted market as a whole is 2/3.
From wikipedia for Risk Neutral Measure:
> A risk neutral measure is a probability measure Q on the set of states such that there exists an r (the "discount rate") with $q_i = rE_Q[D_i]$ for each $i$ where $D_i$ [...] represents the payoff of security $i$
In this case there does exist an $r$, and maybe it equals $\frac{2}{3}$.
If we set if $q_1 = q_2 = q_3 = \frac{1}{2}$ then this all appears to work.
With respect to arbitrage, well, given their -ve expectation, presumably you can short the stocks, and have a positive expectation? For example sell 1 share of each of $S_1, S_2, S_3$ and keep the proceeds ($3$) at zero interest. In the second period you can buy back (in all states) for $2$. Making free money.Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.