Narrow Results

For the strategy of investing in gold and Bitcoin, first collect the historical prices of two types of investment products in the market, and use the wavelet neural network model and WT-LSTM model to model and analyze to predict the future price trends of gold and Bitcoin. Second, considering the difference in price fluctuations between gold and Bitcoin, based on the GARCH-EVT model to increase the risk uncertainty of financial assets, proposes how to achieve the best trading strategy under risk characteristics. Finally, considering the influence of transaction rate on income, we use particle swarm algorithm and genetic algorithm to study what kind of transaction rate can achieve maximum income. The study found that although traders can predict future trends based on daily price changes, due to the different risk factors of gold and Bitcoin, and the different sensitivity of the two financial assets to transaction costs, trading strategies will be very different.

In this paper we deal with the optimization problem of maximizing the expected total utility from consumption under the case of partial information. By means of the martingale method and filter theory, we have acquired an explicit solution to optimal investment and consumption determined by the security prices for a special security price process. Furthermore, we establish a simple formula for valuing information, provided that the utility function is logarithmic. In the end, we extend most of the conclusions to a general situation where both the interest rate and dispersion coefficient of risk security follow some stochastic processes.

Constant proportion portfolio insurance (CPPI) is a structured product created on the basis of a trading strategy. The idea of the strategy is to have an exposure to the upside potential of a risky asset while providing a capital guarantee against downside risk with the additional feature that in case the product has since initiation performed well more risk is taken while if the product has suffered mark-to-market losses, the risk is reduced. In a standard CPPI contract, a fraction of the initial capital is guaranteed at maturity. This payment is assured by investing part of the fund in a riskless manner. The other part of the fund’s value is invested in a risky asset to offer the upside potential. We refer to the floor as the discounted guaranteed amount at maturity. The percentage allocated to the risky asset is typically defined as a constant multiplier of the fund value above the floor. The remaining part of the fund is invested in a riskless manner. In this paper, we combine conic trading in the above described CPPIs. Conic trading strategies explore particular sophisticated trading strategies founded by the conic finance theory i.e. they are valued using nonlinear conditional expectations with respect to nonadditive probabilities. The main idea of this paper is that the multiplier is taken now to be state dependent. In case the algorithm sees value in the underlying asset the multiplier is increased, whereas if the assets is situated in a state with low value or opportunities, the multiplier is reduced. In addition, the direction of the trade, i.e. going long or short the underlying asset, is also decided on the basis of the policy function derived by employing the conic finance algorithm. Since nonadditive probabilities attain conservatism by exaggerating upwards tail loss events and exaggerating downwards tail gain events, the new Conic CPPI strategies can be seen on the one hand to be more conservative and on the other hand better in exploiting trading opportunities.

To assist with the detection of bubbles and negative bubbles in financial markets, a criterion is introduced to indicate whether a market is likely to be in a superexponential regime (where growth in such a regime would correspond to an asset price bubble and decline to an negative bubble) as opposed to “normal” exponential behavior typified by a constant rate of growth or decline. The criterion is founded on the Johansen–Ledoit–Sornette model of asset dynamics in a bubble and is derived from a linear fit to observed data with a nonlinear time transformation with parameters distributed uniformly in their permitted ranges. Making use of expected values rather than the underlying distribution, the criterion is straightforward and efficient to compute and can in principle be applied in real time to intra-day markets as well as longer timescales. In some circumstances, the criterion is shown to have certain predictive qualities when applied to a portfolio of stocks, and could be used as input into algorithmic trading strategies. A simple strategy is described which is based on market reversion predictions of a portfolio of stocks and which in back-testing generates notable returns.

In this paper, we construct five systemic risk indicators and test their performances based on four different datasets. It is observed that the five indicators can accurately indicate the increment of systemic risks during the periods of sub-prime crisis and European debt crisis. Trading strategies based on the risk indicators are further designed to test the warning ability of future price drops. The backtests reveal that trading based on the five indicators provides satisfied excess returns when the trading costs are included. Our results provide insights to find new network-based risk indicators to early warn the systemic risks in financial markets.

In stock trading, a common phenomenon is that the trends of stocks in the same industry are very similar. In contrast, the movements of stocks in different industries are often different. Therefore, applying the same model to all stock trading is inappropriate without distinguishing the industries in which the stocks belong. However, recommending an optimal industry stock trading model is very challenging based on performance evaluation indicators. First, the indicators of the trading model are diverse. Second, the ranking of multiple indicators is often inconsistent. In the paper, we model the problem to be solved as a multi-criteria decision-making process. Therefore, we first divide stock dataset into nine industries according to their main business. Then, we apply several machine learning algorithms as candidate models to generate trading signals. Second, we conduct daily trading backtesting based on the trading signals to obtain multiple performance evaluation indicators. Third, we propose an optimal recommendation algorithm for the industry stock trading model with TODIM. The experimental results in the US stock market and China’s A-share market show that the proposed algorithm can get a better trading model out-of-sample industry stock. Moreover, we effectively evaluate the generalization ability of the algorithm based on the proposed metrics. Finally, the proposed long–short portfolios based on the algorithm have achieved returns exceeding the benchmark on most out-of-sample datasets.

Investing in stocks is a common choice for financial management. Technical indicators (TIs) assist investors in determining the best trading time to make a fortune. Moving average (MA) and relative strength index (RSI) are the most common TIs. The proposed hybrid technique maximizes the capabilities of these two indicators. This study utilizes the quantum-inspired algorithm to assist effectively in searching for the optimal solution in the vast solution space. The proposed trading system contains four innovative features. First, the traditional usage restriction of MA and RSI is removed to increase their potential effectiveness and identify the most profitable trading strategy. Second, this research proposes an innovative hybrid indicator (HI) that combines MA and RSI to simultaneously achieve both benefits. HI eliminates the restriction of employing a single indicator at once. Third, an efficient quantum-inspired algorithm, the Global-best-guided Quantum-inspired Tabu Search Algorithm with Quantum NOT Gate (GNQTS), effectively and efficiently explores optimized parameters. Fourth, this study proposes 60 sliding windows to determine the optimal period for training and testing. The investment targets include well-known indices: DJIA, S&P 500, NASDAQ Composite Index, and NYA, as well as high-reputation companies on the US stock market: DJIA components. By removing the restrictions imposed by these two indicators and the use of HI, the experiment results demonstrate that GNQTS can discover optimal parameters to generate higher returns than state-of-the-art methods and buy-and-hold (B&H) strategies. The proposed hybrid strategy provides for the promising prospect of quantum-inspired applications and the utilization of multiple indicators.

We introduce a new concept of the average holding price (AHP) in the stock market. We show that, under certain assumptions on the investors’ behaviors, the AHP of a stock can be estimated using the historical trading prices and volumes. In contrast to the moving average of the stock price or the volume weighted average price, the AHP serves as a more objective benchmark for estimating the average profit or loss level of the stockholders. The algorithms for estimating the AHP are developed. Simulation studies show that the true AHP can be estimated accurately using our algorithm.

As a trading method, quantitative investment has been widely used for more than 30 years, and its investment performance is stable. As the scale of the international financial market continues to expand, more and more investors have been recognized. Therefore, using quantitative decisions to trade financial products is the mainstream direction in the future. Taking gold and bitcoin, for example, we first establish a prediction model. Since there are limited historical prices to refer to at the early stage of trading, we adopt a robust regression strategy of moving averages to buy stocks. When we have commodity prices for more than 200 trading days, we use BP neural network to predict the price of the next trading day. This prevents the data from being too far back in time and affecting our current price trend. We then use polynomials to fit our predicted product prices and compare them to the actual values to evaluate the prediction model. Finally, with our quantitative decision model, our assets increased from $1,000 in September 2016 to approximately $192,922 in September 2021, which can be proven to be an excellent strategy. For Question 2, we believe that investors have the highest probability of profiting from this investment if we accurately judge future commodity price movements. We use a polynomial to fit the scatter plot of the product price predicted by the BP neural network. We perform the KS test, reliability analysis, and correlation analysis on the fitted curves and the actual prices of the products. It is found that our prediction curve is well-fitted. For sensitivity analysis, we change the transaction cost, finding that the transaction cost is negatively correlated with our income using the previously constructed prediction and decision model. When transaction costs appear, we should adjust our investment strategy in time to avoid frequent trading. An increase in transaction costs results in a small decrease in revenue; therefore, commissions are not sensitive to the final revenue. In conclusion, we provide a memorandum to help traders better understand and apply our quantitative trading strategy.

This work considers trading a mean-reverting asset. The strategy is to determine a low price to buy and a high price to sell so that the expected return is maximized. Slippage cost is imposed on each transaction. Our effort is devoted to developing a recursive stochastic approximation type algorithm to estimate the desired selling and buying prices. The advantage of this approach is that the underlying asset is model free. Only observed stock prices are required, so it can be performed on line. After briefly presenting the asymptotic results, simulations and real market data are used to demonstrate the performance of the proposed algorithm.