MSFTB's daily return correlation with Bitcoin was 0.16 over the previous 60 days, rising to 0.78 over the following 30 days

I pulled the daily candlestick data for Binance's MSFTBUSDT. It is a tokenized stock on Binance that tracks Microsoft. Based on UTC closing times, the latest 90 completed daily candles run from August 12 to November 9; the candle for November 10 is still in progress, so it is excluded. For the daily gains and losses mentioned below, I use both the day's open-to-close change and the change from the previous close to that day's close. For the correlations with $BTC and $ETH , I use daily returns calculated from the previous close to the current day's close.
Over 90 days, $MSFTB  rose from an opening price of 503.16 on August 12 to a closing price of 545.17 on November 9, a gain of 8.35%. Over the same period, $BTC  rose 18.50%, and $ETH  rose 13.42%. There were 47 up days and 43 down days in those 90 days, with no candlesticks that opened and closed at the same price. On August 12, the price fell 2.66% from open to close, closing at 489.76; from that closing price to the close on November 9, it rose 11.31%.
The low was 476.95 on August 19, and the high was 567.64 on October 28, 70 days later. The price rose 19.01% from the low to the high; from the August 12 open to the high, it rose 12.82%. The November 9 close was 3.96% below the high, meaning it would need to rise 4.12% to get back there; it was 14.30% above the low.
I split these 90 days into two periods using daily returns to see how closely it moved with Bitcoin. The first 60 candles run from August 12 to October 10, with 59 daily moves from the previous close to that day’s close. The correlation between $MSFTB  and $BTC  was 0.16, with a beta of 0.08; they moved in the same direction on 37 days and in opposite directions on 22, or 62.71% of those 59 days. The last 30 candles run from October 11 to November 9, with 29 daily moves. The correlation rose to 0.78 and beta to 0.90; they moved in the same direction on 21 days and in opposite directions on 8, or 72.41% of those 29 days. Across all 89 daily moves, the correlation with $BTC  was 0.39, with a beta of 0.27. Compared with $ETH , the correlation was 0.20 for the first 59 daily moves and 0.73 for the last 29. From open to close, the first 60 candles gained 6.32%, while the last 30 gained 1.95%.
Could the stronger correlation in the latter period be driven by just one or two big down days? I removed the two days with the biggest Bitcoin declines from the last 29 days: on November 5, $BTC  fell 5.49% from its previous close, while $MSFTB  fell 4.42%; on October 29, $BTC  fell 4.12%, while $MSFTB  fell 3.11%. Across the remaining 27 daily moves, the correlation was still 0.69, with a beta of 0.92. Removing each day individually and recalculating, the correlation ranged from 0.73 to 0.82. For the first 60 days, after removing the two days with Bitcoin’s biggest gains, the correlation was still 0.15.
There were also days in the last 30 when they moved differently. On November 4, $MSFTB  gained 7.19% from the previous close, rose 7.14% from open to close, and closed at 555.05. The same day, $BTC  gained 3.94% and $ETH  gained 6.59%. On November 9, $MSFTB  gained 2.78% from the previous close, while $BTC  rose just 0.69%. A higher correlation does not mean they move in lockstep every day.
The price path itself had no major swings. Over these 90 days, the biggest drawdown from a closing high was 6.74%, on November 2, when it closed at 509.97, compared with the October 28 close of 546.83. By November 4, the close had reached 555.05, a new closing high for the 90-day period. The close was below the then-current closing high on 75 days in total. Based on the highs and lows, the November 9 close was just 3.96% below the October 28 high of 567.64.
Split the 90 days into three periods. From August 12 to September 10, the price fell 2.55% from open to close; from September 11 to October 10, it rose 9.06%; and from October 11 to November 9, it rose 1.95%. The three periods had 15, 17, and 15 bullish candles, respectively, and 15, 13, and 15 bearish candles. The most recent 14 days (October 27 to November 9) gained 2.21% from open to close, while the most recent 7 days (November 3 to November 9) gained 6.98%. Average daily trading volume over these 90 days was about 330,100 USDT, compared with about 891,200 over the most recent 7 days. The average daily range (high minus low, divided by the close) was about 2.68% in the last 30 days and about 1.82% in the first 60 days.
These figures only show that, during the last 30 daily candles of this 90-day period, the daily price moves of $MSFTB  aligned significantly more closely with those of $BTC  than they did during the first 60 candles; the difference remains even after excluding the two days with the biggest Bitcoin moves in the latter period. Twenty-nine trading days is not a long time, and correlation is not causation. It cannot tell us whether Microsoft’s tokenized stock will follow Bitcoin in the future, nor can a beta of 0.90 be used to extrapolate the next daily candle. This measures what happened along this price path; it is not a prediction of future price movements.
When you see this pattern of lower correlation early and higher correlation later for $MSFTB , would you be more concerned that the latter period’s 29 daily moves are too few, or would you first check whether the 0.69 correlation still holds after removing the big down days?