偶然与金融街一位专家聊天，得知“高频统计套利”这一所谓“金融大规模毁灭性武器”的术语。
这种基于3毫秒操作的系统、650纳秒的交换机数据传输、多线程并行以及复杂数学模型和统计规律等，
一系列复杂技术的系统，真能够实现几乎零风险的投资？乍一看，真有些不可思议。。
In addition to the above-mentioned facts, there are also some basic facts:
1, statistical interest rate Statistical arbitrage is a model-based arbitrage strategy that searches for patterns from historical transaction data of an asset, discovers arbitrage opportunities between two or more assets, and then sets a trade threshold using the model to fit the patterns of change in the price of the asset. Trading is done by automatically issuing trading signals based on real-time market information through computer programs.
What is the meaning of statistical interest:
Statistical arbitrage is a model-driven investment process that profits by constructing simultaneous multi-head and blank-head portfolios when the price of an asset deviates from its theoretical price or the model predicted price. When the price difference between two stocks deviates from the long-term mean, we believe there is bound to be an average response. Since the two stocks belong to the same industry or sector, there is a high probability of a peer-to-peer collision, and the price difference-response mean is likely to be because the price of the relatively overvalued stock decreases more than the price of the relatively undervalued stock decreases when both prices fall; It is also possible that when both prices rise, the relative undervaluation of the stock increases more than the relative overvaluation of the stock; or that when the price of two stocks falls at once, the price difference response is equal due to the rise of the relatively undervalued stock and the relative decline of the relatively overvalued stock.
In the meantime, I'm going to try to make a few statistics. There are two main types of statistical arbitrage strategies, the pairing strategy and the principal component analysis. Pairing, or spread trading, is the most common strategy used in statistical arbitrage, which refers to constructing a multi-headed asset while constructing a blank headed asset and simultaneously tying the two positions at some point in the future. This is a market-neutral strategy that can be immune to market risk and yield low-risk returns by capturing relatively incorrect pricing opportunities between two or more assets. Profit is gained by analyzing the difference between the actual price of an asset and the price predicted by the model. When the actual price of an asset is higher than the price predicted by the model, the asset is overvalued, and the asset is sold until the actual asset price equals the price predicted by the model. In this case, the buyer will buy back the asset to flatten the previous empty position.
3, Domestic and international development of processed transactions. Due to the rapid development of digital funds in the last 10 years, processed transactions have received increasing attention and application. According to the statistics of the New York Stock Exchange, in recent years, the proportion of processed transactions in the total number of transactions on the New York Stock Exchange has remained at about 30%. According to Goldman Sachs reports, processed transactions can bring huge returns to investors. In 2009, Goldman Sachs' total processed transactions accounted for about 50% of the total processed transactions on the New York Stock Exchange, almost double the 27% at the end of 2008. In the second quarter of 2009, Goldman Sachs had 46 trading days with a daily profit of over $100 million. In the third quarter of 2009, it had 36 trading days with a daily profit of over $100 million and a loss of only one day in the quarter.
4, In our country, processed trading is still in its infancy, in the futures market, mainly using short and ultra-short trend trading strategies to speculate on commodity futures, while in the securities market, mainstream brokers use processed trading to leverage exchange-traded funds. Most of the best work done in the country is done by a team, and Guo Taijunan is an important representative of it.
5. Statistical leverage: the main hedging strategy in the field of quantitative investments
Statistical arbitrage is a commonly used hedging strategy in the field of quantitative investing. Statistical arbitrage is the practice of statistical analysis of historical data to find certain patterns of change in the price or yield of an asset, and then using hedging to construct a portfolio to capture the temporary deviation of the asset price from its underlying value.
Statistical arbitrage technology is a multidisciplinary and integrated investment technique based on knowledge of pricing theory, statistical decision theory, game theory, statistical pattern recognition techniques, time series analysis methods, measurement economic modeling, modern computational methods, and so on.
Statistical arbitrage has four main strategies:配对交易策略(Pairs Trading)；多因子套利策略(Multi-factor Model)；均值回归策略(Mean-reverting Strategies)；协整套利策略(Cointegration)。
6th, the mystery of magic: But one thing that cannot be ignored is that this mathematical automated investment model also has some fatal flaws: The model assumes that both the assumptions and the calculations are based on historical statistics, but historical statistics can never fully cover future phenomena.