Update: 6 days ago I shared a 6-month analysis of how fast BTC reacts to news. You guys said it wasn't enough data. So I just mapped 2 FULL YEARS (3,700+ events). Here is the brutal truth

Hey everyone,

Last week I posted an EDA showing the Top 5 news triggers based on a 6-month sample of CoinDesk headlines mapped to 1-minute BTC candles.

While the 15-minute volatility decay was clear, a lot of you (and my own tests) pointed out a massive flaw: 6 months is just one market regime. It’s not enough data to prove if a strategy is actually robust or just overfitting a bull run.

So, I left my pipeline running, dealt with the Binance API rate limits, and scaled the dataset to cover **2 FULL YEARS (May 2024 - May 2026)**. That is 3,773 high-impact news events strictly aligned with T0, T+5m, T+15m, and T+1h price action.

**The brutal takeaway across 3,700+ events:** The initial conclusion holds up even stronger across a 2-year timeline. If you are manually trading breaking news, you are simply exit liquidity for algorithms. The actionable price impact (the alpha) is absorbed almost entirely within the first 10-15 minutes, regardless of whether it's a bull or bear market. By the time a human reads the headline and opens an exchange, it's over.

I’ve updated my Kaggle notebook with the new 2-year charts so you can visually see how the volatility decays over a statistically significant timeframe.

You can check out the updated analysis and the code here: [**https://www.kaggle.com/datasets/yevheniipylypchuk/bitcoin-news-vs-1m-btc-price-action-2025-26**](https://www.kaggle.com/datasets/yevheniipylypchuk/bitcoin-news-vs-1m-btc-price-action-2025-26)

For the quants here: do you still try to trade news sentiment on lower timeframes, or have you completely moved to macro trends? Would love to hear your thoughts on the expanded dataset.