"Cut the risk" to what level? "Until it's a short." That's where risk management becomes discretionary trading. Rob Carver, independent systematic trader and author, recounting an unnamed fund’s risk override in an April 2021 interview. Rob Carver explains: "They had a position on, and someone comes over: ‘I think you should cut the risk on that position.’" "‘That's reasonable. To what level?’ And the guy said, ‘Until it's a short.’" "Now to me, that's crossing the line." "It's one thing to say the model doesn't know about a particular risk, so this position is too big." "The moment you're changing the sign of the position, you've moved from risk management to discretionary trading through the back door." "There’s a clear line for me, but before that there’s probably a big gray area."

Menon’s framework at Prometheus: carry, trend and reversion—with balanced exposure and macro-driven tilts. Aahan Menon of Prometheus Macro on his organizing framework for systematic macro (ReSolve Riffs, Dec 2025). Aahan Menon explains: "There are three big forces, and those are the only things that matter, at least to me." "Every T+1 exposure you have is a function of either carry, trend, or reversion — whether you're trading vol or the S&P 500." "We want dynamic but balanced exposure to those factors. You don't wanna go all in on one and have a lost decade." "What determines whether you tilt from carry to reversion to trend is some macroeconomic circumstance. The tilting is the secret sauce." "You can't take one set of rules and apply it to every market — that's not how it works." "Bonds trend very differently from commodities. Commodities have a preponderance of trend; equities are very mean-reverting." "We believe carry, trend, and reversion define all forward returns."

The cost of building and distributing a company has collapsed. Ken Griffin’s point: when capital, compute, and distribution are widely accessible, the harder moat becomes the people you surround yourself with. Ken Griffin, founder of Citadel, explains: "Access to capital has never been as straightforward as it is today." "The barriers to competition have come down dramatically — whether it’s generative AI, cloud compute, or access to the internet." "If you have a product, you can now touch hundreds of millions or billions of people in the blink of an eye." "This is the best of times." "Looking back on the 35-year history of Citadel, always try to surround yourself with the best and brightest people." "Don’t try to clone yourself." "Find people who are really bright but really different than you are." "The tension between you in debate and dialogue will make you much more effective as a leader."

Tomorrow’s newspaper still wasn’t enough: oversized bets hurt players even when they could read the news a day early. Victor Haghani, Elm Wealth founder and The Missing Billionaires co-author, describing the experiment on Rational Reminder in October 2026. Victor Haghani explains: "We created a game where we give people the front page of the Wall Street Journal a day ahead, historically, and let them trade on that in stocks and bonds and see how they do." "We ran the experiment with real money, mostly with MBA and finance graduate students to begin with and some practitioners. Then we just opened it up, and we’ve had 100,000 people or more play it since then, for fun." "On average, people didn’t do very well despite having this newspaper a full day ahead of time." "People oversized their bets. They would be 40 times leveraged, and the market would go down 2%. They lost 80% on that day, and they couldn’t come back from that." "The sizing just didn’t correspond to sensible sizing or proportional to how clear the signals were." "The economy is stronger than we thought—that could be good for stocks, or that could be bad for stocks because interest rates might be going up. It’s just bad for bonds generally." "When we brought in some really senior macro type traders to play, we got six or seven of those guys to play it, and they did well. They only had a hit ratio of like 60%, but still they did well." "They doubled their money on average over 15 sets of bets. It was possible to do well, but quite a lot of experience was needed."

If “increasing” earns +1, what happens when embezzlement is increasing? Asness on the limits of word counts. AQR’s Cliff Asness at iConnections Global Alts New York in June 2025. Cliff Asness explains: "For years we've tried to use fundamental momentum — are things getting better from the ground up, not just price momentum?" "Quants have tried to process earnings calls for good news or bad news. We did it with word counts — see 'increasing,' plus one." "You build a giant lexicon over time, more complicated phrases, different weights — but that's basically what you're doing." "If the actual sentence was 'unfortunately, massive embezzlement has been increasing,' plus one is probably not good fundamental momentum." "We have a massive amount of data, which is often necessary for machine learning." "Language is inherently nonlinear — this sentence might modify one 10 sentences later." "It's just way better than our old methods."

"Optimization is an art, not a science. Get a coffee, and when you come back, all will be well." Rob Carver, in an April 2021 interview, recalling colleague Simon’s approach to portfolio optimization. Rob Carver explains: "Everyone said they fitted portfolios using some fancy optimization process. I remember the first time I watched somebody actually do this." "Simon said, 'It's okay, Rob. Optimization is an art, not a science. Go get a coffee, and when you come back, all will be well.'" "I came back, and there was this beautiful set of weights." "He'd added constraints to the portfolio until the weights looked like he thought they should look." "No disrespect to Simon — I know loads of people who do exactly the same thing. I've done it myself." "If we're gonna just put in weights we think look right, let's just do that — but in a robust way."

The computing power that once gave the biggest firms an enormous advantage is becoming accessible to almost anyone. Ken Griffin argues that AI is steadily eroding the moats that protect established businesses. Ken Griffin, founder of Citadel, explains: "We now use data centers full of billions of dollars of hardware." "But a startup can lease the same hardware footprint if it wanted to." "These moats to competition continually get eroded." "It's a really incredible time to be an entrepreneur." "A friend of mine had a pet insurance business, and he turned the keys over to his 25-year-old son." "The son said, 'I can use social media. I can find when somebody posts a photo of a puppy on Facebook.'" "I can determine the breed of dog through AI and deliver a custom message to that consumer." "'Congratulations on your new golden retriever. You'll love your dog for the rest of its life. Buy Spot Pet Insurance.'" "They sold the business a few weeks ago for a billion dollars."

Menon’s test: even a perfect year-ahead macro forecast did not make daily allocation changes outperform the underlying beta. Aahan Menon of Prometheus Macro, describing his own allocation experiment on ReSolve Riffs in December 2025. Aahan Menon explains: "There’s a very common saying in discretionary macro: it’s very hard to predict the next couple of days, next couple of weeks, next couple of months, but it’s much easier to forecast the next six to 18 months." "When I started Prometheus, I went out and I tested this, and it’s not even close to true." "If you look at changing your asset allocation on a daily basis based on a one-year forward, 100% accurate growth and inflation forecast, you will not outperform your beta." "We can’t say that they can predict the one-day forward return. But everything from the day after tomorrow until the next year, you know with perfect precision." "You know whether GDP is gonna be up or down. You know whether the S&P 500 is gonna be up or down. You know whether inflation is gonna be up or down." "The most important thing is the trading horizon that you’re trading right in front of you." "If you’re not getting that right, you’re giving yourself a little bit of comfort that there’s more time until your forecast pans out."

Simpler models can be more resilient. Seibert’s trade-off: the return path still isn’t smooth. Moritz Seibert of Takahē, in April 2025, recalling his earlier obsession with model complexity. Moritz Seibert explains: "At some point I was obsessed with complexity. This is probably 15 years or so ago—PCA and covariance matrices and all that type of stuff." "It’s very easy as a quantitative trader to think that with the toolbox that we have, there’s always ways to improve something. Your Python code is there, your Excel spreadsheet is there, you feed it the data." "You think it must be able to do better, and you get very easily into data fitting and overfitting." "We stripped ourselves from that requirement or that necessity to search for complexity or find the solution in complexity." "Oftentimes the simplest things work best. Least degrees of freedom, most resilient." "They're inconvenient though — this type of trading produces erraticness. It's not smooth."

At most firms, a bad decision turns into a search for someone to blame. At Citadel, Ken Griffin says the real advantage is knowing when you're wrong and moving on. Ken Griffin, founder of Citadel, explains: "Every business leader is making decisions under uncertainty." "You have to make your best guesses, be flexible, and be willing to pivot as information unfolds." "We're going as fast as we can in the direction we think we need to travel." "And when we realize we're going in the wrong direction, we stop and change direction." "Many large businesses find themselves unable to commit at all and end up going nowhere." "Great businesses do not suffer from sunk cost fallacy." "One of my partners told me the biggest difference between Citadel and the global bank where he worked was how mistakes were handled." "Where I worked, everyone would try to figure out who to blame and how to bury the mistake in a pile of paperwork." "Here, it's just: 'We screwed that up. Next.'"

In Haghani’s model, return chasing keeps increasing the bet while momentum flips between overweight and underweight. The outcome depends on who has more capital. Victor Haghani, Elm Wealth founder and The Missing Billionaires co-author, discussing his research framework in October 2026. Victor Haghani explains: "Momentum is binary. You're either overweight or underweight. As soon as momentum flips, you do that." "Return chasing is gradual. The better historical returns are, the more you want to own of equities." "Market's going up, returns getting better, you're more and more overweight. Then it starts going down and you're unloading." "That behavior is what results in the return chasers doing poorly and the momentum guys doing well." "It really depends on the relative size of these guys. If return chasers are just a few, you don't get that anymore — everybody's having impact." "What we think is the fact is that the return chasing behavior is operating on a lot more capital than time series momentum is in general."

The 'miss the 10 best days' stat is garbage — miss the 10 worst days and it's equal and opposite. They never show you that half. AQR’s Cliff Asness at iConnections Global Alts New York in June 2025. Cliff Asness explains: "The famous stat: what if you miss the 10 best days? I've hated that stat for at least 26 years." "If you miss the 10 worst days, it had an equal and opposite positive effect — which they never tell you." "You could put the exact same study out, not show them the other one, show them yours, what if you miss the worst days, and show, “See, you gotta time the market. Look at how much you can make.”" "It’s a ridiculously extreme strategy. If 10 times over a 30-year investment period you went from 100% stocks, which you are every other day, to 0% stocks, and those were the 10 worst possible times to do that, then that would be bad." "You probably shouldn't time the market because, like me, you probably suck at it. That's a pretty good reason not to do something." "If you think you’re good at it, you should do it proportional to how much skill you think you have at it, like you should do any other symmetric investment strategy."

The real innovation behind IEX wasn’t just slowing high-frequency traders down. It was designing an exchange that didn’t make more money by letting customers get picked off. Michael Lewis, author of Flash Boys, explains: "At every other exchange, high-frequency traders put their machines right next to the exchange machine." "They get the fastest information out of the exchange and respond before everyone else." "So IEX built what they called the magic shoebox." "They coiled 60 kilometers of fiber inside it." "When a high-frequency trader tried to trade, the signal had to go around the coil first." "It slowed them down." "Then IEX built a fast route to all the other exchanges, so customer orders could get there before high-frequency traders raced ahead." "They identified the conflict of interest: the other exchanges were owned by brokers and high-frequency traders." "So this exchange would be owned by investors — people who just wanted a fair place to buy and sell stocks." The clever part wasn’t the delay. It was removing the business model that made speed worth selling.

Before U.S. markets fully went electronic, Thomas Peterffy was already treating exchanges like software endpoints. If the exchange wouldn’t send trades through the wire, he built a camera to read the screen. Thomas Peterffy, founder of Interactive Brokers, explains: "In 1991, the Deutsche Terminbörse became electronic and started an options exchange." "I told them we had a completely automated system showing bids and offers to our traders on the floor." "Now that they had an electronic exchange, we wanted to pipe our bids and offers directly into it." "That way, they would have liquidity." "Nobody was trading because it took forever to type in bids and offers as the stocks kept moving." "So we hooked our system to their system and automatically piped in our bids and offers." "At first, they wouldn’t let us receive trades on the wire." "They insisted on showing the trade on the screen." "So we built a camera looking at the screen." "Then we built a program to translate the pixels and interpret the numbers." "By 1993, we had a big international network where we made automated markets." That was the edge: not just knowing the right price, but building the plumbing to quote it everywhere before the market caught up.

The most interesting part of Flash Boys isn’t that HFT found a speed edge. It’s that the entire incentive stack made the edge hard to fight. Michael Lewis, author of Flash Boys, explains: "Brad Katsuyama went to the SEC and described how he was being front-run." "The room instantly divided." "The older SEC staffers were upset and disturbed by what they were hearing." "The younger SEC staffers were more sympathetic to the high-frequency traders." "RBC later found that more than a couple hundred SEC staffers had left to work for HFT firms or lobbying firms tied to HFT." "Brad realized he wasn’t going to get anything done there." "Whenever a new rule gets made, clever people find a way around it." "So the question became: is there a market solution?" "Can you create incentives in the market for a fair market to arise?" "He decided to quit Royal Bank of Canada and create a stock exchange." The technical fix was building an exchange. The harder trade was convincing real money to stop feeding the old one.

A millisecond doesn’t feel like time to a human. In market structure, it can be the whole trade. Michael Lewis, author of Flash Boys, explains: "It wasn’t seconds. It was milliseconds." "The difference between the market ordinary investors see and the market high-frequency traders see is just a couple milliseconds." "From a computer’s point of view, that’s plenty of time." "People have a hard time getting their minds around how a computer perceives time." "To the naked eye, when you hit buy, it looks like you got it right away." "But that’s not what happened." "Your order got sold." "The right to execute your order was sold by your broker to a high-frequency trader." "The exchanges sell special access and speed to high-frequency traders." "The incentives are all screwed up in the system." "The stock market is supposed to allocate capital." "This is a market run by intermediaries for intermediaries."

When Nasdaq told Thomas Peterffy his automated bids had to be “keyed in through the keyboard,” he took the rule literally. So he built a machine that typed faster than any human could. Thomas Peterffy, founder of Interactive Brokers, explains: "We had a Nasdaq terminal, and we started to trade quite a lot." "One day, the vice chairman of Nasdaq came in." "He expected to see twenty people on terminals constantly making markets." "Instead, there were two or three people sitting around, mostly writing computer code." "He asked, ‘Who is doing all this trading?’" "I pointed to the computer." "He said, ‘This is not going to work. You clearly have an advantage over people typing bids and offers by hand.’" "I said, ‘That’s life. Technology grows, and people who use the technology have an advantage until everyone else catches up.’" "Then he found a Nasdaq rule saying bids and offers had to be keyed in through the keyboard." "So we built pistons over the keyboard." "The computer drove the pistons." "When we had something to type, it didn’t go bam, bam, bam." "It went brr." The rule said the orders had to touch the keyboard. Peterffy made sure they did.

By 2009, the edge wasn’t knowing Microsoft better. It was seeing the order before the rest of the market could react to it. Michael Lewis, author of Flash Boys, explains: "When Brad hits buy 50,000 shares of Microsoft, the signal goes up the West Side Highway and out the Lincoln Tunnel." "The first exchange it reaches is BATS." "Physical distance matters." "It’s light traveling through fiber." "High-frequency traders are sitting there waiting to find out what he wants to do." "They have built faster routes from that exchange to all the other exchanges." "They race ahead of him." "They buy the Microsoft he wants, then sell it back to him at a higher price." The trade wasn’t being beaten by a better thesis. It was being beaten by the pipes.

Before options pricing became textbook finance, Thomas Peterffy was brute-forcing fair value at home on an Olivetti computer. The edge was simple: simulate thousands of trades, find the break-even prices, then compare them to the market. Thomas Peterffy, founder of Interactive Brokers, explains: "People were buying and selling puts and calls, but nobody really knew what they were worth." "If you could figure out how to price them, maybe we could make some money." "I had no immediate answer." "But I kept thinking about the problem." "It would not leave me for months." "I created a program to generate random numbers and used them to create a price series." "On that price series, I bought and sold puts and calls across different durations and strikes." "I summarized the outcomes to determine the prices where I would break even." "Obviously, this would have been impossible without the computer." Before the model was obvious, the edge was doing the work nobody else had automated yet.

Roulette is unbeatable — unless you build a computer into your shoe. Thorp + Shannon: 44% edge. Edward O. Thorp, invented card counting, ran a quant Princeton Newport Partners, explains: "Claude Shannon was probably the king of gadgeteers. Robots that ran mazes, machines that played chess." "We decided to make an all-out effort to build a machine that would predict the outcome of our roulette game." "We built a small computer with about 11 transistors in it. It's now at the MIT Museum." "Nine or 10 months, working in Shannon's basement almost full time, we built this wearable computer — the first wearable computer according to the MIT Media Lab." "One person entered push-button info on the position and velocity of the ball and wheel. The computer would instantly tell you where to bet." "The bettor heard musical tones. When the last tone stopped, it told him what section of the wheel to bet." "We divided the wheel into eight sections and had a massive edge of 44%. The piles of dimes became huge piles very quickly."