No-Arbitrage Conditions in a One-Period Binomial Model
Summary
The document explains the one-period binomial model’s no-arbitrage condition by comparing the stock’s up and down factors with the risk-free growth factor. If the stock’s return in both states is no better than the risk-free asset, shorting stock to buy the bond can create an arbitrage. If the stock’s return in both states is at least as high as the risk-free return, borrowing through the bond to buy stock can create an arbitrage. The strict ordering places the risk-free growth factor between the two stock outcomes.
The response also expresses a zero-cost portfolio using positions in bonds and stock, then evaluates its terminal payoff in each state. This gives a way to check when a candidate portfolio can produce nonnegative payoffs with a positive payoff in at least one state. Positive up and down factors additionally ensure positive stock prices in both outcomes. The explanation assumes a single period, two assets, and no transaction costs or other market frictions.
Key ideas
- In a one-period binomial model, no arbitrage requires the risk-free growth factor to lie strictly between the down and up factors.
- If both stock outcomes underperform the risk-free asset, shorting stock and buying the bond can create arbitrage.
- If both stock outcomes outperform the risk-free asset, borrowing to buy stock can create arbitrage.
- A zero-cost stock and bond portfolio can be checked by comparing its payoff in each state.
- Positive up and down factors ensure that modeled stock prices remain positive.
Tags
Full text
# Non-Negativity of up-factor and down-factor in Binomial No-Arbitrage Pricing Model
# Non-Negativity of up-factor and down-factor in Binomial No-Arbitrage Pricing Model
Consider a stock which is trading at $S_0$ at time $t=0$ and is expected to be trading at price $uS_0$ or $dS_0$ at time t=1 where $u$ and $d$ are up-factor and down-factor. The theory says that to rule out the arbitrage, we must assume that : $0<d<1+r<u?$ Can someone explain how does this assumption takes care of no-arbitrage?
## Answer by vanguard2k (score 3)
https://quant.stackexchange.com/a/14188
This sounds like the first chapter of Björks book am I right? It treats a single-stage model. Simply put, if $1+r \leq d$ you buy the stock and have $V_1\geq 0$ with positive probability of making a profit. If $1+r \geq u$ you want to sell the stock short and buy the bond from the proceeds. The result is the same.
Edit: To show that the condition is sufficient, we can follow Proposition 2.3 from "Arbitrage Theory in Continuous Time". In a two-asset, one-period world one can characterize all possible arbitrage portfolios (because $V_0 = 0$) by $x+yS_0 = 0$ ($x$ is the amount of money invested in bonds, $y$ in stocks) and thus write the value at time $1$ explicitly:
$V_1 = y S_0 (u- (1+r)), \text{if S goes up}$ and $V_1 = y S_0 (d- (1+r)), \text{if S goes down}$
Now, for an arbitrage portfolio with $y>0$ we need that $V_1 > 0$. That can only happen if $u>1+R$ and $d>1+r$. Similarly, for an arbitrage portfolio with $y<0$, whe get the other direction of the inequality.
## Answer by emcor (score 0)
https://quant.stackexchange.com/a/14192
If the condition $$0<d<1+r<u$$ is not satisfied, the Binomial model (with $d<u$) would have immediate arbitrage opportunity:
1) $1+r\geq u$: Then the riskfree asset would yield least as much return as the stock in any state for any probability, so you could short the stock to buy the riskfree asset and end up with some riskless profit with positive probability (arbitrage).
2) $d\geq 1+r$: Then the stock would yield least as much return as the riskfree asset in any state for any probability, so you could short the riskfree assetto buy the stock and end up with some riskless profit with positive probability (arbitrage).
The condition $d,u>0$ is just to ensure positive stockprices in all states.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.