TL;DR - Across the 53 published interviews on this site, 35 traders bring up AI unprompted or when asked. The pattern is consistent: the quant competitors use large language models to generate and stress-test hypotheses, and then make the data decide; the discretionary traders use AI to automate research and say it will not replace the judgement at the trade. Nobody with a verified placing describes handing execution to a model. The 2025 IQC runner-up built LLM-driven alpha pipelines; his teammate's warning is that once everyone can ask an LLM for ten alphas, asking is no longer an edge.
The question "do winning traders use AI" has a boring answer and an interesting one. The boring answer is yes, most of them, in some form. The interesting answer is what for, and the interviews on this site give it in the traders' own words. Thirty-five of the fifty-three published interviews touch on AI, machine learning or algorithms. The quotes below are chosen because the speaker has a verified competition placing and because the statement is specific enough to act on.
The quants: AI generates, data decides
The clearest picture comes from the WorldQuant International Quant Championship, where the work is building "alphas," predictive signals tested across thousands of stocks.
Sumit Kumar finished second at the 2025 IQC Global Finals in Singapore, the only Indian in the top three, with no finance background at all:
"I come from an engineering background and had no prior knowledge of finance. My approach was more mathematics driven. For building Alphas I used techniques like template generation, combinatorial approaches and LLM-driven pipelines. I also used various statistical methods and custom key performance indicators to test the quantitative robustness of my Alphas."
The model generates candidates. The statistics decide which survive. That order is the whole method.
Siddhesh Kamble, alpha researcher on Team Smoooth Operator, the 2026 IQC India National Winners, describes the same split and then names its limit:
"We used LLMs extensively for research, understanding concepts, finding limitations in our hypotheses, and cross-verifying our ideas."
"Once everyone has access to AI, simply asking an LLM to generate ten alphas isn't a meaningful advantage."
"Everything had to be validated through data and backtesting."
His teammates describe the validation culture that makes the generation useful. Tisha Hirani: "you have a hypothesis, you turn it into something testable, you run it against data, and the data gets the final say." Prathamesh Dharmaraja: the most underrated skill is "the discipline to kill your own idea when the data tells you to." An LLM can produce a hundred ideas an hour. It cannot kill one.
Chia-Chun Chung, a senior AI researcher from Hsinchu who took bronze at the 2025 IQC, is the one interviewee whose day job is building the models, and he wants more from them than automation:
"Rather than simply using AI to automate existing workflows, I'm interested in building AI systems that can help researchers think more creatively, generate new hypotheses, and accelerate scientific discovery."
The discretionary traders: AI does the chores
The futures, forex and equity traders in the series use AI, when they use it, for the work around the trade.
Florian Philippi, a data engineer who placed third in the 2024 U.S. Investing Championship Enhanced Growth division at +311%, built his own AI-powered stock research tool to automate the screening and preparation that used to take his evenings. The decision to buy stays with him.
Christian Flanders, second in the 2024 USIC at +433.5%, was asked directly whether AI is changing momentum trading and whether the patterns he trades will still work in thirty years:
"I don't think it's AI per say, I think as a bull market goes farther and farther in time, you have more and more traders participating and they anticipate breakouts/patterns forming which makes everything happen "faster". Bear markets "reset" this dynamic because most traders get wiped out in the decline."
"Yes, I believe VCPs and Code 33 will still work and exist in 30 years. VCP's are a visual representation of supply and demand."
That is the discretionary view in one paragraph: the thing being modelled is crowd behaviour, and crowd behaviour is older than the tools used to watch it.
Vincent Mischke, a classical chart pattern trader and TradingView Editors' Pick, makes the same argument from the other end of the century: "the classical patterns, which worked 100 years ago, are still intact today and generate strong trading signals." His analysis is deliberately indicator-free. Whatever automation sits behind it, the published chart is a human reading price.
The two questions traders actually ask
Should I let AI trade for me? Nobody in the series with a verified placing says they do. The closest is the quant camp, where a model proposes signals that are then combined, neutralised and traded as a portfolio, under a research process that discards most of them. Even there, the trader owns the kill decision.
Will AI make my method obsolete? The discretionary traders answer no and give a reason: the patterns they trade are expressions of supply, demand and crowd psychology, which more participants make faster but do not remove. The quant traders answer differently: the method is safe, but the edge from any individual tool evaporates as soon as the tool is common. Kamble's line is the one to remember. Access is not an advantage. The research loop around the access is.
What this means for a trader entering a tournament
Three practical conclusions follow from the interviews.
If you compete in a quant format, build the validation pipeline before the generation pipeline. The 2025 and 2026 IQC placings in this series all came from teams that describe testing, stress-testing and discarding in more detail than they describe generating.
If you compete in a discretionary format, use AI where it saves time without touching the decision: screening, research, journal analysis. The traders with audited returns in this series keep the entry and the sizing manual.
In either case, treat "I used AI" as a statement about workflow, not about edge. The leaderboard does not record which tool found the trade. The Hall of Fame records the return, and so far every entry in it was made by a person.
Frequently asked questions
Do professional traders use AI?
Most of the traders interviewed on this site do, in one of two ways. Quant competitors use large language models to generate and stress-test trading hypotheses, then let backtests decide. Discretionary traders use AI for research, screening and automation around the trade while keeping entry and sizing decisions manual.
Is AI trading better than human trading?
The interview record does not show AI beating humans at the decision itself. It shows AI making research faster. The 2025 IQC runner-up built LLM-driven alpha pipelines; his teammate notes that once everyone has the same access, the generation step is no longer an edge, and the validation discipline is.
Will AI replace chart patterns and discretionary trading?
The discretionary traders in the series say no, because the patterns describe supply, demand and crowd behaviour rather than a technique that can be copied. More participants make moves happen faster; bear markets reset the dynamic. The patterns persist because the people do.
Where do these quotes come from?
From the interview section of each trader's profile on trading-tournaments.com, conducted by Vitalii Kaminskyi. Each quote is verbatim and linked to the profile where the full interview appears.
Related
- Gut or data? Tournament champions give opposite answers the decision-making half of the same series.
- The G.O.A.T. canon what a verified placing is and why it outlasts the tool.
- How to win a trading tournament the practical playbook from past champions.
- WCTC vs USIC how the championships quoted here are audited.
Background on the competition the quant interviewees won: the WorldQuant International Quant Championship and the U.S. Investing Championship.
Last reviewed 2026-10-13 by the trading-tournaments.com editorial team. Quotes are verbatim from published interviews; corrections via the Suggest a change form.