Cryptoracle Data Analysis Team

· The distance matrix (Euclidean distance) is constructed based on the mention matrix of currency and date, and the hierarchical clustering method (ward) is used to cluster the currency;

The 10 categories are aggregated to obtain the "resonance sector" of each currency, so as to support the subsequent sector-level linkage identification and trading logic.

Let the price of a currency at the current time point be,

The price was 30 days ago,

Then the past one-month return (i.e., momentum) of that currency is

For the assets that pass the absolute momentum screening, calculate the relative volatility intensity of the ranking in the latest week, calculate the change value of the ranking in the last 7 days (rank_change_7d> 0), sort them according to M_rel from high to low, select the Top3, and ensure that the TOP3 are not in the same group.

a. Invest $1,000 in each currency

b. The momentum is recalculated every three days, and the top three currencies in terms of momentum are selected as the holding objects. All the currencies that are no longer in the top three in the current holding are sold, and the newly entered currencies in the top three are bought equally.

▶ assumed condition

With an initial capital of $3000, you can buy all the currencies (assuming M) on the market at the same time. Then each currency will receive the following funds:

Number of coins to buy per coin:

After buying, hold for a long time and do not adjust the position. On the last day (2025/6/25), calculate the value of the held coins according to the final price of each coin and add up to get the final net value.

It means that you've held it for a while, but the overall market has gone down, so buying all the coins on average is a loss