Michael V. Klibanov,Andrey V. Kuzhuget
Abstract:A new mathematical model for the Black-Scholes equation is proposed to forecast option prices. This model includes new interval for the price of the underlying stock as well as new initial and boundary conditions. Conventional notions of maturity time and strike prices are not used. The Black-Scholes equation is solved as a parabolic equation with the reversed time, which is an ill-posed problem. Thus, a regularization method is used to solve it. This idea is verified on real market data for twenty liquid options. A trading strategy is proposed. This strategy indicates that our method is profitable on at least those twenty options. We conjecture that our method might lead to significant profits of those financial institutions which trade large amounts of options. We caution, however, that detailed further studies are necessary to verify this conjecture.
What problem does this paper attempt to address?
The main problem that this paper attempts to solve is to predict stock option prices by solving an ill - posed problem (i.e., the Black - Scholes equation in reverse time), and propose a trading strategy based on this prediction. Specifically:
1. **Prediction problem**: The paper proposes a new mathematical model for solving the ill - posed problem of the Black - Scholes equation in the positive time direction. The traditional Black - Scholes equation is solved in the reverse time direction (from the expiration time \(t = T\) to \(t = 0\)), which is a well - posed problem. However, the author hopes to solve the equation in the positive time direction (from \(t = 0\) to \(t = T\)) to predict option prices in the short term. Since this is an ill - posed problem, a regularization method is required to solve it.
2. **Trading strategy**: Based on the prediction results, the paper proposes a trading strategy, which has been verified on actual market data. The author randomly selected 20 options with good liquidity, applied their prediction model and trading strategy, and evaluated the profitability of these strategies.
### Specific problems and solutions
1. **Prediction problem**:
- **Mathematical model**: The author proposes a new mathematical model, including new stock price ranges, initial conditions, and boundary conditions. The traditional concepts of expiration time and strike price are not used.
- **Regularization method**: Since the Black - Scholes equation in the positive time direction is an ill - posed problem, the author uses a regularization method to solve this problem. Specifically, they use a functional \(J_{\alpha}(u)\) similar to Tikhonov regularization and find an approximate solution by minimizing this functional.
- **Numerical method**: The author uses the finite difference method to discretize the partial differential equation and the conjugate gradient method to minimize the regularization functional \(J_{\alpha}(u)\).
2. **Trading strategy**:
- **Buy - sell rules**: According to the predicted option prices, the author formulates specific buy - sell rules. For example, if the predicted price is higher than a certain threshold (such as 0.03 US dollars higher than the ask price), then buy the option; if the predicted price is lower than a certain threshold, then do not trade.
- **Empirical analysis**: The author applies their model and trading strategy to 20 options with good liquidity and records the profit or loss of each option. The results show that among these 20 options, 17 achieved profitability, with a total profit of 20.8 US dollars.
### Main conclusions
Although the author's method often overestimates when predicting option prices, the trading strategy proposed can still achieve profitability in most cases. The author believes that if large financial institutions adopt this technology for large - scale trading, they may obtain significant profits. However, they also point out that further research on more option data is required to verify this conjecture.
### Future work
1. **Improve prediction accuracy**: The author suggests that prediction accuracy can be improved by estimating the volatility coefficient more accurately. Volatility may depend on the stock price \(s\) and time \(t\), that is, \(\sigma=\sigma(s, t)\).
2. **Solve nonlinear inverse problems**: Calculating volatility requires solving a very difficult coefficient inverse problem, which is more challenging than the current problem. The author suggests that a globally convergent numerical method can be tried to solve this problem.
In conclusion, this paper successfully predicts short - term option prices by solving an ill - posed problem and proposes an effective trading strategy. Although there are some limitations, the performance of its method on actual market data is encouraging.