L'IA ne prédit pas l'avenir | El Mostafa Belkhayate | TEDxENCGA

The talk is a personal success story and motivational speech rather than a rigorous scientific presentation. The speaker, El Mostafa Belkhayate, claims to be a three-time World Trading Champion and to have achieved extraordinary returns (1000% in one year) and to have been ranked top gold manager by Bloomberg. However, these claims are not substantiated with verifiable data or references. The audience is encouraged to trust his authority based on his achievements, but no evidence is provided. The core concept of ’time-of-day profiling’ is interesting and plausible: financial markets exhibit intraday patterns, and certain times may be more predictable due to liquidity, volatility, and market participant behavior. However, the speaker does not provide any empirical evidence or references to support this idea. He simply asserts that 90% of time windows are not predictable and that focusing on one 30-minute window is key. This is an oversimplification and lacks scientific rigor. The speaker also makes a strong claim that AI cannot predict the future, which is a valid point in the sense that financial markets are complex and unpredictable. However, he does not discuss the extensive research on machine learning in finance, which often focuses on probabilistic forecasting rather than deterministic prediction. He introduces the term ’nowcasting’ but does not explain it in depth, and his advice to ’erase the past and future’ is vague and not actionable. The talk includes a live demonstration where he asks the audience to use AI to generate code for a predictability coefficient, but this is not shown in detail. The overall quality of information is low: it is mostly anecdotal, with no references to academic literature or data. The argumentation is based on personal authority and emotional appeal rather than evidence. The title is somewhat misleading: while the speaker does say AI cannot predict the future, the talk is more about his success and motivational advice. The adequacy between title and content is moderate. The talk is aimed at a general audience, but the analysis should not penalize that. The lack of sources and verifiable claims significantly reduces its reliability. The speaker’s advice to focus on a single time window may be useful as a heuristic, but it is presented as a universal truth without supporting evidence. Overall, the talk is more inspirational than educational, and its scientific value is limited.

The title accurately reflects the central message that AI cannot predict the future, but the talk focuses more on personal success stories and trading advice.

4/10

The talk is primarily anecdotal and motivational, with limited technical depth and no verifiable data or sources. The speaker's claims of world championships and performance are unsubstantiated. The core advice on time-of-day profiling is plausible but presented without rigorous evidence.

Markers derived by PSI from the transcript: the creator did not define chapters.

• 00:00Introduction and personal story of winning the first online trading championship in 1999.
• 02:00Working for Michael Jackson and managing a gold fund in Canada; ranked top gold manager by Bloomberg.
• 05:00Building a trading algorithm and app with 1000% returns; obtaining US SEC license.
• 08:00Winning the Chicago Exchange competition and being nominated for the World Championship in Abu Dhabi.
• 10:00Live trading challenge at Abu Dhabi and winning the championship.
• 12:00Introduction to time-of-day profiling: analyzing 30-minute windows for predictability.
• 15:00Advice to use AI to generate code for a predictability coefficient and focus on one time window.
• 18:00Emphasis on nowcasting over forecasting and the unpredictability of the future.
• 20:00Encouragement to be patient and trade only once per day per market.

• TEDx Program — Mentioned in the description as the organizing body for TEDx events.

• Intraday seasonality — Supports the idea that market behavior varies by time of day.

• Efficient-market hypothesis — Challenges the notion that predictable patterns can be exploited consistently.