Visnu Gonzalez

[Visnu Gonzalez @VisnuGD]

@No_thing_007 I'm in the deepseek with 3k

[Visnu Gonzalez @VisnuGD]

@No_thing_007 You crazy? No way. I bought myself Solana to see his bluff I'm up 0.00678 something, so basically nothing. It's 🗑️

[Visnu Gonzalez @VisnuGD]

@GuoShenCN7 🤣🤣🤣🤣🤣🤣 SCAAAAAM

[Visnu Gonzalez @VisnuGD]

@GuoShenCN7 No such ticker as $SICAT

[Visnu Gonzalez @VisnuGD]

@Vivek6935 @GuoShenCN7 Yeah I will spam his posts though 🤷

[Visnu Gonzalez @VisnuGD]

@GuoShenCN7 Darn bot

[Visnu Gonzalez @VisnuGD]

@GuoShenCN7 U dumb bot

[Visnu Gonzalez @VisnuGD]

@GuoShenCN7 @GuoShenCN7 this guy won't even comment here. I think it's a bot. About to report it

[Visnu Gonzalez @VisnuGD]

@GuoShenCN7 Go ahead

[Visnu Gonzalez @VisnuGD]

@grassi77951 @aifinancelabs @joinautopilot You from st Andrews?

[Visnu Gonzalez @VisnuGD]

@GuoShenCN7 🏆

[Visnu Gonzalez @VisnuGD]

@RichBoogeyMan Do you announce your swing entries? If so where? Let me tag along to test your way on a small 5k I set aside. Thanks

[Visnu Gonzalez @VisnuGD]

@QuettaBrowser Same stuff. I have same extensions on 3 different chromium browsers, they ALL work fine. It's obviously not my phone. 🤯

[Visnu Gonzalez @VisnuGD]

Now I'm getting really frustrated, I think I spoke to soon and I might go back to Misses browser.not only are extensions not working but now sound isn't working in sites like https://t.co/Ilvau2GUFW can't watch a video because sounds won't work no matter what I do
@QuettaBrowser

[Visnu Gonzalez @VisnuGD]

@QuettaBrowser No response yet?

[Visnu Gonzalez @VisnuGD]

This is incredible, only 2 greens, is @deepseek_ai going "deep" under the ocean? 🤷
@aifinancelabs

Independent live and academic evaluations of LLM trading are far less flattering. In multi-season simulated live trading (crypto and U.S. stocks, same prompt and rules across models), only about 46% of model-seasons finished profitable and rank in one period did not predict rank in the next. A 24-day real-time fund experiment found most flagship models, including DeepSeek-V3, posted net losses; only one Grok variant was slightly positive. Longer backtests of LLM timing or news-based strategies often fail to beat simple buy-and-hold once transaction costs, broader stock universes, and different market regimes are included—LLMs tend to be too conservative in strong bull markets and too aggressive in downturns. Directional hit rates on short-horizon price moves or chart interpretation usually sit near 50-60%. Crypto “arena” contests produced occasional strong short bursts for DeepSeek (trend-following style with favorable risk-reward), but results were volatile and not consistently first-place after corrections.

Key limitations that reduce real-world accuracy include look-ahead or training-data leakage in backtests, inability to model true market impact or regime shifts, and the fact that markets quickly incorporate public information that LLMs also see. Specialized fine-tuning or multi-agent setups can improve sentiment or ranking tasks modestly, but they have not produced reliably repeatable alpha as standalone traders. The Autopilot experiment is a useful public data point because holdings and reasoning are disclosed, yet it remains one implementation during a period when certain thematic bets (AI power demand, energy, defense) paid off. Treat any single “AI beat the market” claim as one sample, not proof of systematic edge.

This is concerning enough to bail out sooner than later 🤐

[Visnu Gonzalez @VisnuGD]

@0DonaldTrumpJr1 93