Efficient and Near-Optimal Online Portfolio Selection
提出一种新算法，在在线投资组合选择问题中达到与通用投资组合算法几乎相同的对数遗憾保证，但每轮运行时间从指数级降至多项式级，适合高频交易场景。
In the problem of online portfolio selection as formulated by Cover [Cover TM (1991) Universal portfolios. Math. Finance 1(1):1–29], the trader repeatedly distributes the trader’s capital over d assets in each of T > 1 rounds with the goal of maximizing the total return. Cover proposed an algorithm, termed “universal portfolios,” that performs nearly as well as the best (in hindsight) static assignment of a portfolio with an [Formula: see text] logarithmic regret. Without imposing any restrictions on the market, this guarantee is known to be worst case optimal, and no other algorithm attaining it has been discovered so far. Unfortunately, Cover’s algorithm crucially relies on computing a certain d-dimensional integral, which must be approximated in any implementation; this results in a prohibitive [Formula: see text] per-round runtime for the fastest known implementation. We propose an algorithm for online portfolio selection that satisfies essentially the same regret guarantee as universal portfolios—up to a constant factor and replacement of [Formula: see text] with [Formula: see text]—yet has a drastically reduced runtime of [Formula: see text] per round. The selected portfolio minimizes the observed logarithmic loss regularized with the log-determinant of its Hessian—equivalently, the hybrid logarithmic-volumetric barrier of the polytope specified by the asset return vectors. As such, our work reveals surprising connections of online portfolio selection with two classic topics in optimization theory: cutting-plane and interior-point algorithms. Funding: This work was supported by the Directorate for Computer and Information Science and Engineering [Grant CIF-1908905].