Crypto_Painter

Crypto_Painter

88,805 Twitter followers

Follow

Posts

The relationship between crypto analysts and contract traders should be roughly equal to: Is the weather forecast the same as the weather betting on PolyMarket?

After a night of repeated evolution, the system has begun generating strategies that surpass Beta returns! The training used BTC-USDT data from 2018 to 2021. Data after 2021 is all out-of-sample data. The equity curve shows numerous flattening and jagged inflections, indicating that the strategy is gradually approaching the behavior of real traders. The previously stable and smooth curve now looks rather artificial... However, a new pitfall has been encountered... This evolutionary mechanism leads to elite factors continuously dominating the dominant species, severely reducing species diversity. The next goal is to ensure the diversity of factor evolution by limiting the over-proliferation of a single type of strategy. Ideally, the GP system will evolve two highly effective species, one focusing on trend and the other on oscillation. After subsequent training to differentiate market regimes, the two multi-factor strategies can be integrated into one! The previous approach of prioritizing Regime fitting seems too hasty. Without good alpha, Regime fitting only introduces fitting noise, which is a classic "chicken or egg" problem... Interesting~ twitter.com/CryptoPainter/stat...

Previously, a bunch of accounts in the English-language forums were chanting "withdrawal movement," but now they're silent... After all this time of withdrawals, how did they even increase Binance's market share? So we have to admit, in this industry, you can never look at what others are saying, but rather what they are doing.

CoinGlass 2026 Q1 Cryptocurrency Market Share Research Report One key takeaway from the Report: Trading stayed active, but liquidity and capital became even more concentrated at the top. Binance remained the clear leader in volume, OI, depth, and reserves, while Hyperliquid

Ever since having a child and discovering that sudden death is a common occurrence, I can't help but wonder what to do with crypto assets. Upon seeing the #OKX feature, I immediately added an emergency contact through the security settings in my account settings. This is undeniably a very user-friendly design! twitter.com/CryptoPainter/stat...

Another new pitfall... The factor expressions output by genetic algorithms lack logical auditing, so often even after extensive backtesting, the output still shows severely overfitted and contradictory factors, such as the following: where_signal((delay(1h_TD_Seq_4) <= (1d_Volatility_Compression_10_Z50 * 70)), where_signal((sma_20(1h_Donchian_Position_20) <= ema(1h_BB_Width_20, ema_fast(1h_Donchian_Position_20))), 1, 0), where_signal(cross_up(1d_Std_Compression_10_Z50, 15m_OBV_Relative_20_Z200), 0, -1)) How can this be solved? I integrated the Gemini 3.1 API, analyzing the high-scoring factor expressions every 20 generations. If overfitting or logical contradictions were found, I intervened in the evolution process to forcibly remove the factor. However, Claude believes I shouldn't use Gemini for this. He suggests regulating factor output by optimizing the output logic, rather than adding a black box to the end of an already heavily randomized pipeline. Okay~ I'll give it a try~ twitter.com/CryptoPainter/stat...

Loading..