Unstable fluctuations in financial markets caused by the 2008 financial crisis and currently by the Covid-19 crisis have generated greater concern among investors regarding their capital protection. In view of this situation, the consideration of alternative investments has taken a relevant position to protect their wealth and obtain profits. Due to the relevance of these investments in these times, this study proposes using artificial intelligence to predict the value of alternative investments, specifically the numismatic asset the Walking Liberty Half Dollar. To achieve this objective, the use of Generalized Regression Neural Networks has been proposed over a sample 25 coins of the Walking Liberty Half Dollar with several qualities valued in the period 2000-2019. Two models were proposed, one for the entire selected sample and the other one for each type of coin depending on its year of minting. Thus, it has been found that the model proposed for the entire sample has a success rate of between 86.12% and 97% while the approach for each type of coin obtained success rates that even reach 100%. The variables that have the greatest influence within the model are the state of conservation of the coin, its age, and its exclusivity. In this way, these results provide fundamental information to investors to understand the behaviour of these assets, and to be able to formulate more profitable investment portfolios, especially in times of great economic instability.

由 2008 年金融危机和目前由 Covid-19 危机引起的金融市场不稳定波动已引起投资者对其资本保护的更大担忧。针对这种情况，另类投资的考虑已经占据了保护其财富和获取利润的相关地位。由于这些投资在这些时代的相关性，本研究建议使用人工智能来预测另类投资的价值，特别是钱币资产 Walking Liberty Half Dollar。为实现这一目标，有人提议在 25 枚步行自由半美元硬币样本上使用广义回归神经网络，这些硬币具有 2000 年至 2019 年期间的多种价值。提出了两种模式，一个用于整个选定的样本，另一个用于每种类型的硬币，具体取决于其铸造年份。因此，已经发现针对整个样本提出的模型的成功率在 86.12% 到 97% 之间，而针对每种硬币的方法获得的成功率甚至达到了 100%。在模型中影响最大的变量是硬币的保存状态、年龄和排他性。通过这种方式，这些结果为投资者提供了基本信息，以了解这些资产的行为，并能够制定更有利可图的投资组合，尤其是在经济严重不稳定的时期。12％和97％，而每种硬币的方法获得的成功率甚至达到100％。在模型中影响最大的变量是硬币的保存状态、年龄和排他性。通过这种方式，这些结果为投资者提供了基本信息，以了解这些资产的行为，并能够制定更有利可图的投资组合，尤其是在经济严重不稳定的时期。12％和97％，而每种硬币的方法获得的成功率甚至达到100％。在模型中影响最大的变量是硬币的保存状态、年龄和排他性。通过这种方式，这些结果为投资者提供了基本信息，以了解这些资产的行为，并能够制定更有利可图的投资组合，尤其是在经济严重不稳定的时期。