Discussion Topic: 2/2 Guest speaker comment.2/2 Guest speaker comment.
What did you learn from today's speaker, John Niccolai? What was the
best question from the audience? Did you ask it?

Reply from Felix Zou

I learned a lot about Citadel and products, though admittedly I did
not understand everything given my lack of financial literacy.

I thought the best question (given my CS background) was the one about
neural nets. John Niccolai's answer that the "fanciest arrows in the
quiver" don't always works the best aligns with my own understanding
of this as well in general. Of course, I am less familiar with trade
horizons and HRT vs. Citadel, but that element of the answer was
interesting to hear as well. I did not ask that question.

Reply from Zikang Chen

Some market overviews regarding FX, rates, mortgages and some
reflections on those. The best question is how math could be applied
to those market analysis. I am not the person who asked it.

Reply from Thomas Chung

I think it was really cool to have someone like John talk about topics
that I've heard of but never really understood. I didn't know what
things like Fixed Income and the Mortgage Market actually were until
today, but he explained them really well. I think the best question I
heard was whether we should buy Silver in lieu of the sharp drop the
other day, and very intelligently John avoided the question.

Reply from Gavin Onghai

One thing I learned was how inflationary markets work and what are
some current methods to estimate the market (namely monte carlo
similations). The best question, I thought, was his views on the 50
year mortgage plan. I thought it was interesting to hear an investor's
take on it.

Reply from Alex Lu

I learned that real markets have fat tails, i.e. they are more
volatile than gaussian which make them harder to simulate. The best
question was what one should do as a trader with the belief that AI is
looking like a bubble like the dot-com bubble. While Niccolai didn't
give personal trading advice his answer was still insightful. I did
not ask this question.

Reply from Omar Abdellall

I learned more about a broad overview of finance and many concepts and
terms that I wasn’t familiar with before, which he explained
clearly. I thought the best question was about whether the current
excitement around AI being a 'bubble', since his response gave helpful
perspective of what a trader thinks. I didn’t ask the question.

Reply from Arya Bhushan

One thing I learned today was how complicated products like fixed rate
mortgages actually are, and how mortgages in general are different in
the US vs. many other countries (in other countries tend to be
floating rate)

The best question was about the impact of the Federal Reserve (such as
who is chosen to lead it) and how that affects decisions made by the
macro desk at Citadel. Prof. Slade asked this question

Reply from Joanne Zhao

I learned about how the popularity of 30 year fixed mortgages in the
US enables the Fed to hike interest rates during economic
downturns. Most other countries have shorter term adjustable rate
mortgages, so when interest rates are raised, people may be unable to
pay their mortgages and be left without a home.

The best question from the audience was the one about the AI bubble,
and I thought Niccolai's response that the AI bubble is different from
past bubbles because the companies that are heavily investing in it
have massive profits to be gained was very interesting and not one
that I've heard of before.

Reply from Tony Chang

One thing I learned a lot about was the inner workings of Citadel and
integrating software into trading. The best question from the
audience, I believe, was the question about John's perspective on the
AI bubble, as that has become an increasingly relevant topic lately,
given the newly released profit margins of ChatGPT. I did not ask this
question.

Reply from Pranay Kapoor

I was excited to hear about Citadel's offerings and how they are
trying to bring computer science in the trading space. The best
question was about the car that John Niccolai drove. No, I didn't ask
the question

Reply from Weixing Zhang

I learned that every time you trade, there is a unique
product/instrument being used; however, there is no transparency with
data and prior information. In France, 65-year-old retirees now have
higher incomes than working-age adults.  An interesting question was
"whether to double down or not " for gold/silver metal trading.

Reply from Jiashu Huang

I learned that real market is not following standard Gaussian
distribution and we need to account for that when developing
models. The best question was about whether current AI is a bubble. I
did not ask it.

Reply from Olivia Ye

I learned that the US has a mortgage market unlike other countries,
especially Europe. I found it super interesting that there is a
correlation between hikes in rates and how fast people vote out
incumbents in politics there, compared to the US.

A question asked I found interesting was if they would use neural
networks for evaluating risk instead of Monte Carlo Simulations. He
talked about time horizons and how the plays into it, which is a good
example of choosing the proper decision system doesn’t always mean the
most complex one available.

Reply from Xianhang Lin

John’s lecture topic and content really appealed to me. I‘ve always
been interested in applying CS domain knowledge—such as ML, CNNs, and
Transformers—to financial trading time-series data, especially since I
also minored in finance during my undergraduate studies.

Reply from Cindy Chen

I learned a lot more about financial topics and how quantitative
finance firms like Citadel apply computer science and math to real
markets. I think the best question from the audience was the one
relating to AI. I didn't ask this question.

Reply from Lakxshanna Raveendran

I learned a lot about mortgages, which I had very little prior
knowledge of. I wasn't entirely familiar with some of the vocabulary,
though, so it took me a bit more effort to follow. I thought the
question on the effect of AI was interesting, but I did not ask it.

Reply from Emmett Seto

One thing I learned from John Niccolai was that Citadel does not have
a Singapore, Hong Kong or Tokyo office because they could trade out of
their London office or other offices.  Additionally, I learned that
the 30-year fixed mortgage is uniquely American and does not apply to
other countries. One question asked from the audience was whether or
not we should buy metals such as silver and gold. I did not ask this
question.

Reply from Will Yang

John gave an overview on the financial landscape, with an emphasis on
the fixed income and foreign exchange part, and how Citadel works in
those various markets. I also learned about the interesting case study
of the Swiss franc flash crash in 2015. The best question asked is
about whether the recent development in AI might change the neutral
rate of interest, r*, following a discussion on the similarities
between AI and dotcom bubble, and the way and timeframe those
technical developments change people's life. I didn't ask the
question.

Reply from Sean Lee

I gained a more thorough understanding of various concepts in finance
such as mortgage markets. I also learned about the difficulties of
predicting and simulating the market. I thought that the best question
from the audience was the one about the AI bubble, since it is a topic
that is very relevant today. I didn't ask that question.

Reply from Xinyuan Zhu

I learned that the fanciest arrows in the ML quiver aren't always the
best tools for every trade. The highlight was the question about the
AI bubble; John's take on how current profit margins differentiate
this era from the dot-com crash was a real eye-opener. Even with my
limited finance background, seeing how CS is used to navigate
'fat-tailed' market volatility was insightful. I didn't ask questions.

Reply from Diana Shyshkova

I learned a lot about Citadel's structure and the range of products
they trade. What impressed me most was how the speaker handled the
question he couldn't answer directly

Reply from Dane Keahi

One thing I learned from today's speaker was about Mortgages. This was
an important topic because it affects the general population. I
learned that the US is a very interesting place for mortgages. The
benchmark is a 30-year fixed mortgage, which is a very long time. This
was interesting to me because people in the US will know for sure that
they can pay off the house (as long as they don't lose their job or
anything), but this is not the same for people in Europe.

There weren't a lot of questions, but the best question from the
audience was whether Citadel is thinking about using more neural nets
to analyze market trends instead of a Monte Carlo simulation. This was
especially interesting because it hinted on the integration of
artificial neural networks to solve problems. I did not ask it.

Reply from Youxuan Ma

I learned that the government and policies can affect the market a
lot, and that although the people trading Macros like volatilities and
fluctuations to some extent, they fear regime change. I also learned
that the government and big tech companies are now spending huge
amounts in building data centers, which could very likely drive up
electricity prices for consumers and households, which are what we
don't wanna see.

I think the best question was: "wouldn’t the uncertainties around
tariffs and inflations create some novel arbitrage opportunities?",
which I did not ask.

Reply from Rena Wang

Before the lecture, I had very limited financial knowledge, but I
found the talk really engaging and easy to follow. I learned how risk
in fixed income markets is always “live,” especially due to different
trading hours and liquidity across products. The discussion on stale
bond prices and infrequent ticks showed the challenges of real-time
risk management. The best question was whether the current AI craze is
a bubble. It was interesting to hear John’s perspective on this from
the viewpoint of the financial industry.

Reply from Daniel Xu

I learned how quantitative models and computer science are used in
real financial markets, particularly in fixed income. A key takeaway
was that simpler, well-understood tools can often be more effective
than overly complex models. The best question was about whether the
current AI boom resembles a bubble, since it highlighted how traders
think about long-term technological shifts and risk; I did not ask the
question.

Reply from Hubert Wang

John discussed a wide range topics including global markets, trading
strategy, and Citadel's role in it. One of the most interesting
questions was about using neural nets in the prediction strategies. It
was particularly enlightening to hear that the most sophisticated
tools aren't always the most effective. Often, the most robust
strategies rely on clarity rather than complexity. I didn't ask the
question.

Reply from Cixuan Zhang

I learned how the structure of the U.S. mortgage market affects how
interest rate changes impact households and the broader economy. The
lecture also highlighted why realistic market models must account for
fat tails and extreme risk rather than assuming Gaussian behavior. One
question I found interesting was whether neural networks could replace
Monte Carlo simulations for risk evaluation. The discussion on time
horizons showed that choosing the right decision framework depends on
context, and that more complex models are not always the most
appropriate choice.

Reply from Vanesa Aguay Guerra

I learned about how exciting it is to be in the space applying time
series models to financial decision making. The fact that markets do
not reflect Gaussian models creates ample space for the exciting work
of working with live an messy data to iteratively build better
insights while focusing on thinking critically about prior
assumptions.

Reply from Kemi Omoniyi

I learned how Citadel’s Fixed Income and Macro business connects
rates, FX, and mortgages, and why real markets are harder to model
than simple textbook assumptions. John Niccolai emphasized that the
most complex tools, like neural nets, aren’t always the best choice
because the right approach depends on time horizon and context. The
most interesting audience question for me was whether AI looks like a
bubble, and his response offered a measured trader’s perspective
without giving personal advice.

Reply from Gabriel Brown

Something I found interesting from Dr. Niccolai's speech was the
conversation about mortgages. I did not realize that fixed rate
mortgages were a US thing and that in other countries variable rate
mortgages are more common. This is interesting, as Niccolai said,
because it means that raising the interest rates has different impacts
in the US vs other countries.

The most interesting question was whether Citadel was using neural
nets to make decisions in regard to their trading because this is
something that makes sense and I hadn't thought about before. It was
interesting to learn that they didn't consider this a viable
investment strategy for their position. I did not ask it

Reply from Alba Quintas Núñez

I learned a lot about Citadel's different areas of focus and which
profiles they are interested in for industry internships. I am not
very knowledgeable about the world of finance, but the most engaging
part of the talk for me were the cases Niccolai walked us
through. This helped me understand the application of the concepts he
mentioned. Finally, I think the best question was about neural
networks for predicting price movements. I did not ask it, but
Niccolai's answer that the prediction horizon is very different was
enlightening. It made me reflect on the risk and portfolio standpoint,
where with neural nets, we would achieve higher production and a
better Sharpe ratio, but capacity would be low.

Reply from Elizabeth Schaefer

I had an excused absence, but I am very sorry for missing! I bet it
was a very interesting talk.

Reply from Helen Mao

I learned that mortgages in the US are very unique because they last
30 years. This allows the government to do more to the interest rates
without enraging the public. The best question from the audience was
whether his math background helped his job at all. I did not ask it.

Reply from Sasha Spiegel

I enjoyed learning about the different ways in which one can make
money off the stock market.  Even something like inflation rates can
be traded.

The best question was the one about neural nets and HRT, which is
interesting because I've done internships at HRT and enjoy learning
about how the company works.

Reply from Andy Ma

I was surprised to learn that finance is very broad and most people in
finance only work in one small section of the finance world (equities,
real estate, fixed income, etc.).

I think the best question was about the use of neural nets and other
SOTA ML techniques at Citadel. I was pleasantly surprised to hear that
these SOTA techniques are not the final word, and that older
techniques are still applicable.

Reply from Bende Doernyei

It was so interesting to see how a COO is able to think so casually
and honestly about their company and was great to hear how certain
decisions, like the Asia office expansion were deliberated within
leadership. After the talk, finance seemed less secretive and more
approachable -- I especially liked Dr. Niccolai 's frank discussion
about how investment decisions are made, how they try to appeal to the
investors (they sometimes need to portray somewhat different things
than what they actually do in the background, but that still serves
the interst of the customer very much), and I liked how he discussed
how they make the investment packages more appealing (e.g. you "cannot
sell and S&P500 investment with a markup, you need to make it look
more complicated") :D

Reply from Joseph Yu

I gained a more thorough understanding of just how complicated the
market is, particularly regarding mortgage markets. It was fascinating
to learn how factors from individual mortgage rates can influence the
national economy, and understanding more of this connection would
really interest me. The standout moment for me was the audience
question regarding the AI bubble—a highly relevant topic, though I
wasn't the one who asked it.

Reply from Miranda Selin

I thought the discussion about the 'winner's curse' & how many
counterparties to request a quote from was particularly interesting. I
knew previously the sell-side perspective: that you should worry if
you give the most aggressive quote. However, I had never thought
through the implications on the buy-side perspective--that you would
want to minimize the counterparties' worries over 'winner's curse' via
not asking too many companies for quotes. It's a fun game-theory-style
puzzle.

I thought the question on how often one sees a regime shift was very
interesting, though I did not ask it.

Reply from Meghana Chamarty

I learned that markets have messy dynamics, so even in a place like
Citadel the goal is choosing robust models and tools that fit the time
horizon, not just the fanciest technique. The best audience questions
were about AI and the math behind markets, and I didn’t ask those
questions.

Reply from Samuel Lee

John was really good at explaining the state of the economy currently
and the big headlines in each market. I found the part about mortgages
interesting, how European mortgage rates differ from traditional US
fixed rates. Especially the impact this difference had in a recent
event such as the COVID-19 pandemic, significantly increasing an
European borrower's monthly rent payment while American mortgage
payments didn't increase. I didn't ask this but the question related
to the impact of regime changes was pretty revealing as to how
seemingly unrelated global events can affect trader decisions.

Reply from William Wang

I liked his perspective of artificial intelligence of coding and his
perspective on PhD. I think the best question was about his car. I
didn't ask the question

Reply from William Wang

I learned about his perspective about artificial intelligence with
coding and mortgage. The best question was about his perspective about
the value of PhD. No, I didn't ask it

Reply from Sam Meddin

Learned about what types of simulation citadel uses to maximize their
performance and really manager for risk. The question I asked which
was my favorite was about to what extent citadel uses neural nets and
John let us know they really aren’t using them to that much of an
extent.

Reply from Yuwang Ma

Tried my best to understand, but it was intereting. I learned about
the great company Citadel, and especially about "How does fixed income
impact you“. How macro rates markets show up in real life and how a
professional macro platform thinks about trading, risk, and
execution. The questions about AI bubbles are interesting. I didn't
ask because I can barely keep up what is going on.

Reply from Gio Martinez

I learned a lot about Citadel and basically a lot about the finance
world. I think the best question was about how the AI bubble looks
like the dot com bubble and what we as traders should do. I did not
ask it.

Reply from Emma Slagle

John Niccolai's discussion of recency bias was memorable for me. His
mention of the common mentality that the "housing market never goes
down" as an example of recency bias forgets that investors had to sell
at large losses during the recession, and illustrates how cognitive
biases become embedded in decision-making. It gave an example to our
recent lectures, which point out that the goal of algorithmic
decision-making in machines isn’t to be perfect, but to be better than
humans. It reminds us that automation doesn't eliminate human error,
but additionally can actually scale and accelerate flawed assumptions
if we're not thoughtful about the behavior we encode into our systems.

Reply from Ethan Smith

I had a harder time connecting with John Niccolai's presentation than
I did with Dan Russell's presentation. I'm not too well versed in the
finance world, and the question I was most interested in asking (but
didn't ask because I felt uncomfortable) was whether
quant/finance/trader people ever face a crisis of whether their work
produces value or just moves it around. I'd imagine the answer would
be that it's valuable to move investments around because it smooths
out the economy and puts money towards entities that will use it well
in periods of growth.

As for the actual content of the presentation, I didn't know that the
US had been missing the ideal target for inflation before COVID. Also
it freaks me out a little every time we talk about national debt. But
then again, I don't think I'd heard anyone talk about the debt-to-GDP
ratio before, and how the GDP grew much faster than the debt after
WWII, making it seem like debt was going down. I guess the investments
worked then.

I appreciated Prof. Slade's questions because they seemed really
insightful, but they were a little too far over my head to remember
the questions or the answers well.

Reply from Yide Jin

I learned that quantitative trading is much more complex than applying
machine learning or deep learning models in isolation. The speaker
emphasized that models like LLMs cannot realistically simulate true
market conditions, especially issues like liquidity, execution, and
market impact.

The best question from the audience asked whether modern AI models
could fully replace traditional quantitative strategies. I did not ask
the question, but I found the discussion very helpful.

Reply from Ephraim Akai-Nettey

I got a very quick and insightful view into what citadel does. I also
learned about the models people come up with evaluate risk (monte
carlo simulations etc) and how the models got it wrong during the 2008
financial crisis Best question was whether we should double down or
metals or not considering the huge dip in recent days.

Reply from Vivian Kaleta

I learned that markets aren't just noisy they are really just
structurally messy. Niccolai really stressed that fat tails are a
feature of our markets and extreme events happen more often than
Gaussian models predict. Crises in the market also aren’t “once in a
lifetime” for traders, they are recurring. So events like COVID or
2008 weren't just noise or anomalies but rather they are structural
properties of the system. This matters because it emphasizes that
stress testing matters more than making point forecasts. A question I
liked was "Is AI a bubble?"