The style of machine learning in the cryptocurrency circle In the cryptocurrency space, the style of machine learning strategies largely depends on the number of coins you select and how you allocate your funds. We all know that the maturity of the cryptocurrency market is far lower than that of traditional markets. So here, the fat tail effect will be more pronounced. Whether it is a positive black swan or a negative black swan, Their frequency is much higher than that of traditional markets. When you select a large number of coins, Machine learning models pay more attention to the "overall average situation" under heavy-tailed distributions. In this all-market environment, due to limited attention, The market often finds it difficult to maintain an upward trend, Therefore, the reversal effect will be more pronounced, and the style will be closer to the reversal factor. In other words, choose a multi-currency machine learning strategy, often performs steadily in the daily market, But if you encounter a few cryptocurrencies with particularly large fluctuations (such as MYX, ALPACA), The model may incur significant losses. When you select a small number of coins, Machine learning models will focus more on the strong trends of a small number of underlying assets. Focus on capturing those cryptocurrencies that are continuously rising and attracting high attention, This type of strategy is more inclined towards momentum effects, Can hedge the risks of the "select multiple coins strategy" to some extent in a highly volatile market. So, how can we reduce the losses from a multi-coin strategy in a volatile market? There are usually two methods: 1. Mixed Strategy: Integrate a part of the "select few coins" machine learning strategy, Find a balance between mean reversion and sustained trends for the model; 2. Gradient Weight Training: By using different weights for selecting coins, the model learns the mean reversion reversal style in daily market conditions. At the same time, it can capture momentum style in a highly volatile market. In this way, you can maintain a stable performance in the daily market. It can also reduce the risk of being "hit hard" by sudden events when the market changes abruptly.
Deeply cultivating proprietary quantitative trading for many years, focusing on creating a top-tier personal quantitative community & professional practical teaching plan. Cherish xiaofu method 2886
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
#量化交易
The style of machine learning in the cryptocurrency circle
In the cryptocurrency space, the style of machine learning strategies largely depends on the number of coins you select and how you allocate your funds.
We all know that the maturity of the cryptocurrency market is far lower than that of traditional markets.
So here, the fat tail effect will be more pronounced.
Whether it is a positive black swan or a negative black swan,
Their frequency is much higher than that of traditional markets.
When you select a large number of coins,
Machine learning models pay more attention to the "overall average situation" under heavy-tailed distributions.
In this all-market environment, due to limited attention,
The market often finds it difficult to maintain an upward trend,
Therefore, the reversal effect will be more pronounced, and the style will be closer to the reversal factor.
In other words, choose a multi-currency machine learning strategy,
often performs steadily in the daily market,
But if you encounter a few cryptocurrencies with particularly large fluctuations (such as MYX, ALPACA),
The model may incur significant losses.
When you select a small number of coins,
Machine learning models will focus more on the strong trends of a small number of underlying assets.
Focus on capturing those cryptocurrencies that are continuously rising and attracting high attention,
This type of strategy is more inclined towards momentum effects,
Can hedge the risks of the "select multiple coins strategy" to some extent in a highly volatile market.
So, how can we reduce the losses from a multi-coin strategy in a volatile market?
There are usually two methods:
1. Mixed Strategy:
Integrate a part of the "select few coins" machine learning strategy,
Find a balance between mean reversion and sustained trends for the model;
2. Gradient Weight Training:
By using different weights for selecting coins, the model learns the mean reversion reversal style in daily market conditions.
At the same time, it can capture momentum style in a highly volatile market.
In this way, you can maintain a stable performance in the daily market.
It can also reduce the risk of being "hit hard" by sudden events when the market changes abruptly.
Deeply cultivating proprietary quantitative trading for many years, focusing on creating a top-tier personal quantitative community & professional practical teaching plan.
Cherish xiaofu method 2886