摘要
近年来,比特币已经成为了最流行的加密货币之一,它的价格波动经常被关注和研究,目前比特币的市场价值非常高,已经达到了历史新高,在全球范围内引起了广泛的兴趣和关注,它已经成为了数字资产领域的重要一员。本文通过研究时间序列模型在预测中的应用,提出了滚动ARIMA预测模型,使用MASE,MAE,RMSE比较了三次指数平滑模型和滚动ARIMA模型的预测效果,最后预测了2023年6月的比特币收盘价。研究表明,在短期预测方面,滚动ARIMA模型和三次指数平滑预测模型效果大致相同,但是滚动ARIMA模型具有较好的方向控制和误差控制效果,最终得出滚动ARIMA模型适合长期预测的结论。结果显示,在2023年6月,比特币收盘价呈现波动趋势,并且波动趋于平缓。
In recent years, Bitcoin has become one of the most popular cryptocurrencies, its price fluctuations are often paid attention to and studied, the current market value of Bitcoin is very high, has reached a record high, has attracted widespread interest and attention worldwide, it has become an important member of the digital asset field. By studying the application of time series model in forecasting, this paper proposes a rolling ARIMA forecasting model, compares the prediction effect of the three exponential smoothing model and the rolling ARIMA model using MASE, MAE, RMSE, DAR and finally predicts the closing price of Bitcoin in June 2023. The results show that in short-term prediction, the rolling ARIMA model and the cubic exponential smoothing prediction model have roughly the same effect, but the rolling ARIMA model has better direction control and error control effect, and finally concludes that the rolling ARIMA model is suitable for long-term prediction. The results showed that in June 2023, the Bitcoin closing price showed a volatile trend and the volatility flattened.
出处
《应用数学进展》
2023年第6期2853-2860,共8页
Advances in Applied Mathematics