Searcharxiv⌕ Search

arXiv subjects

Yuh-Dauh Lyuu

Publications and source records attributed to Yuh-Dauh Lyuu.

4 recordsLinked to original sources
arXiv2022

Backtesting Trading Strategies with GAN To Avoid Overfitting

Many works have shown the overfitting hazard of selecting a trading strategy based only on good IS (in sample) performance. But most of them have merely shown such phenomena exist without offering ways to avoid them. We propose an approach to avoid overfitting: A good (meaning non-overfitting) trading strategy should still work well on paths generated in accordance with the distribution of the historical data. We use GAN with LSTM to learn or fit the distribution of the historical time series . Then trading strategies are backtested by the paths generated by GAN to avoid overfitting.(This paper is an tanslated English version of a thesis (10.6342/NTU201801645) which was originally written in Chinese in 2018, where some statements and claims are outdated in 2022)

cs.CE↗
arXiv2010

On irreversible dynamic monopolies in general graphs

Consider the following coloring process in a simple directed graph $G(V,E)$ with positive indegrees. Initially, a set $S$ of vertices are white, whereas all the others are black. Thereafter, a black vertex is colored white whenever more than half of its in-neighbors are white. The coloring process ends when no additional vertices can be colored white. If all vertices end up white, we call $S$ an irreversible dynamic monopoly (or dynamo for short) under the strict-majority scenario. An irreversible dynamo under the simple-majority scenario is defined similarly except that a black vertex is colored white when at least half of its in-neighbors are white. We derive upper bounds of $(2/3)\,|\,V\,|$ and $|\,V\,|/2$ on the minimum sizes of irreversible dynamos under the strict and the simple-majority scenarios, respectively. For the special case when $G$ is an undirected connected graph, we prove the existence of an irreversible dynamo with size at most $\lceil |\,V\,|/2 \rceil$ under the strict-majority scenario. Let $ε>0$ be any constant. We also show that, unless $\text{NP}\subseteq \text{TIME}(n^{O(\ln \ln n)}),$ no polynomial-time, $((1/2-ε)\ln |\,V\,|)$-approximation algorithms exist for finding the minimum irreversible dynamo under either the strict or the simple-majority scenario. The inapproximability results hold even for bipartite graphs with diameter at most 8.

cs.DM↗
arXiv2000

Optimal Buy-and-Hold Strategies for Financial Markets with Bounded Daily Returns

In the context of investment analysis, we formulate an abstract online computing problem called a planning game and develop general tools for solving such a game. We then use the tools to investigate a practical buy-and-hold trading problem faced by long-term investors in stocks. We obtain the unique optimal static online algorithm for the problem and determine its exact competitive ratio. We also compare this algorithm with the popular dollar averaging strategy using actual market data.

cs.CE↗