In May I published a case study on running alpha research through a team of LLM agents: four roles, instrumented handoffs, and roughly a 2 lift in hypotheses tested per week. The post was careful about what it measured. It was silent about a risk I had not thought to price. Here is the risk. If your research agent and your competitors research agent are the same base model, then whatever that

Take two stocks that both end the month flat. The first bled lower for two weeks, then clawed all the way back. The second ran up for two weeks, then gave it all back. Same start, same finish, same zero return. Sort every stock in the market by past return and these two land in the identical bucket, because return only sees the endpoints. Gulen and Woeppel show that the shape between the endpoints

An 8-week dual-momentum switch between GLD and IBIT returns 79.91% a year at a Sharpe of 1.64. Vojtko and Dujava report that number on a Quantpedia note covering 31 December 2018 through April 2026, weekly rebalance at Wednesday's close, cash when both lookbacks print negative. Hold the other column: maximum drawdown is still -43.94%. They tested ten lookbacks and published the peak. The

In September 2017, Alpha Architect published VIX and Trend-Following, the Killer Combo?, an empirical examination of whether volatility information could improve a traditional trend-following allocation model. The central idea was intuitive: market volatility may contain useful information about how quickly an investor should measure momentum. A shorter momentum window may react more effectively

Factor timing is the seductive idea that you can do better than holding a basket of factors: lean into value when value looks cheap, load momentum when momentum is running, cut a factor when its volatility spikes. Every one of those moves has a tidy formula and a plausible story. And when you run them out of sample on a clean universe, almost all of them lose to the dumbest thing you could have

I recently came across Lpez de Prado et al.'s (2026) treatment of the Sharpe ratio estimator: Normal, with the true as its mean, and a variance (see the Normal section of the image below) conditioned on sample size, the return series' autocorrelation, its skewness, its kurtosis, and the level of itself. It's an elegant correction but stubbornly a bell curve underneath.

Markets have been behaving unusually lately. In May, equity indices rose while volatility and skew also increased, a relatively rare occurrence historically. Since last week, the same phenomenon has emerged again, with the spot/volatility correlation turning positive. Is this still a rare occurrence? We dont know. But one thing is clear: regime detection is becoming increasingly important in

Three papers crossed my desk this month. One detects market manipulation with graph neural networks and reports 98.7% accuracy at 8.3 milliseconds per decision. A second, same author cluster, detects anomalies in high-frequency trading and reports a 15% accuracy improvement with an F1 of 0.915. A third uses XGBoost on Bitcoin technical indicators and reports 92.40% accuracy with a ROC AUC of

You lock in gains, you stop giving winners back, and, the part everyone repeats, your win rate goes up. I tested it on 3,966 of my own trades. Two of those three claims are false, and the third one cost me a fortune. The setup I run an automated futures system with a fixed exit: each strategy takes its full target or its full stop. No partials, no breakeven moves; the trade rides to one of two

Where this came from. A reader replied to Autopsy 001 with a Morningstar page showing NANC ahead of the index and a fair question: what am I missing? They were not missing anything. The fund really has outperformed. This is the answer to what that outperformance is made of. Unlike our other case files this is not a pre-registered test of our own hypothesis. It is a descriptive analysis of

Over the past few years, the liquid alternatives space has moved decisively into a new ETF era. What was once a narrow category now stretches from CTA to hedge-fund replication strategies, sometimes pairing those return streams directly with core stocks and bonds exposures. And thanks to their low-cost, daily liquidity and ease of access, these vehicles are attracting progressively more inflows.

The pitch One screenshot: $6k into weekly calls, $1.2M out. The logic sounds like physics options have capped downside and uncapped upside, so keep buying cheap lottery tickets and one asymmetric win pays for everything. The post-2021 refinement adds a mechanism: find the gamma squeeze, ride the dealers' forced hedging, be the flywheel instead of the fuel. The test Twenty years of S&P

A well-known finance paper showed that an AI model could read a news headline about a company and say, better than chance, which way the stock would move. The same paper predicted the effect would fade as more traders started using the same tools. Thanks for reading! Subscribe for free to receive new posts and support my work. We tested that prediction with 2026 data, a different AI model, and

Run pairwise Granger tests over a basket of thirty instruments and you get a causal network that looks like a plate of spaghetti. AUD/USD drives copper, copper drives the Aussie back, oil leads the loonie, the loonie leads oil, and somewhere in the mess a small-cap ETF appears to Granger-cause the ten-year yield. Most of those arrows are fake. They come from three things markets are drowning in:

The dividend discount model (DDM) says that a stocks current price ought to reflect the discounted value of its expected future dividend stream (1) begin{equation*}mathrm{Price} = sum_{t=1}^{infty} frac{mathbb{E}[mathrm{Div}_{t}]}{(1{+}r)^t}end{equation*} mathbb{E}[mathrm{Div}_t] is the companys expected dividend in t years, and r > 0% is the firms discount rate. The Gordon

Every investor who holds only their home market has made the same decision. Over the last century and a half that decision paid an Australian seven percent a year in real terms and a Portuguese investor less than one. Nobody chose which of those they were born into. The number everyone quotes Equities return about seven percent a year after inflation. That figure comes from one country, measured

You backtest a strategy, subtract "a spread" for costs, and move on. Where did that number come from? If you typed in a guess, or pulled a single quoted spread from a vendor, you are almost certainly wrong, and wrong in the one direction that flatters your worst assets. The effective spread you actually pay is unobservable without trade-and-quote data, and quote data for anything outside

This article presents methods for adjusting the duration of major countries in global bond indices using point-in-time measures of local economic conditions. The macro factors include inflation, credit conditions, real estate price growth, yield-curve valuations, and economic surprises. Each has clear theoretical implications for interest-rate markets. An equally weighted composite score of these

How a single-value performance metric can distort the ranking of trading strategies and how to avoid falling for it and gain deeper insight into strategy comparison. Diagram comparing probability distributions of trading strategies A and B, showing why a single higher performance value (V_B) doesn't mean strategy B is truly better than A Picture two trading strategies, A and B. You run

Behavioral finance research has established that investors dislike negative skewness because it exposes them to rare but severe losses, while they embrace positive skewness because it offers the chance of occasional outsized gains the lottery-like appeal that persists even when expected payoffs are modest. In behavioral models, this preference for positively skewed assets bids up their prices,

This is an independent test of a novel strategy from David Varadi: Inflation Compass. It builds on his earlier Growth and Inflation strategy by adding a direct market-based measure of expected inflation. Were testing two versions of his new strategy: Original and Enhanced (more on this later). Backtested results from 1990 follow. Results are net of transaction costs see backtest

A time series is mean-reverting if it tends to return to a stable long-run level after being displaced from it. In pure form, that means the process has a well-defined unconditional mean and a variance that does not grow without bound; shocks decay rather than accumulate. This is the opposite of a random walk, where each innovation is permanently absorbed into the level and variance grows linearly

You already ran the models. The old article "From One Tree to Forests to Boosting" walked you from a single decision tree to XGBoost, "How a Decision Tree Engineers a New Alpha" showed a tree carving conditional edges out of order-book features, and "Ridge Above 1h, XGBoost Below 5min" handed you a timeframe rule for which model to point at which horizon. None of

What Is the Ornstein-Uhlenbeck Process? The OrnsteinUhlenbeck (OU) process is the simplest continuous-time model of a mean-reverting random process. It was introduced in 1930 by Leonard Ornstein and George Uhlenbeck as a physical model of the velocity of a Brownian particle experiencing friction (Uhlenbeck & Ornstein, 1930), and it has since become one of the most widely used stochastic

It cleared the control test that kills almost everything I try and that I still wont trade. Thats because theres a gap between those two things and it is the most useful thing Ive learned building this system, so let me show you what actually happened. The pattern For this particular example, I am trading the market on the daily gaps. I noticed what looked like a pull toward round

Open any retail futures forum and you will find the same fourteen setups traded as gospel: the opening range breakout, the gap fade, the gap continuation, the volume spike, the liquidity grab reversal, the Asia session expansion. Each comes with a chart, a win rate, and a story about why it works. Mathias Mesfin took the whole list, wired it to a bar-close signal and next-bar-open fill on Micro

The last few posts have been a lot to chew through. So here is a simple ETF rotational strategy that you can do to capture the upside of equities while eliminating half the risk. The inspiration for this strategy comes from the paper Looking for Synergy with Momentum in Main Asset Classes which you can read for yourself here. Not a subscriber? Become one today for more research delivered to your

The only sustainable competitive advantage is to learn faster than your competition. Arie de Geus Implementing research papers can sometimes work, though a perfect replication often fails. Its never wasted effort, though: the ideas in the paper end up feeding new ideas and good conversations with other researchers. Heres the 4th edition of the Quantitativo weekly, featuring papers that

If you learned to trade from YouTube, you probably learned the Supertrend. It topped our folklore hunt being the single most-taught beginner tool on the internet, complete with the green-line/red-line screenshots and the promise that the trend will tell you when to start riding the new trend. So we did the honest thing: we tested it exactly the way it's taught. Figure 1 is the tool itself, on

Both June and July saw SPX close lower. Seasonality for those two months it is generally favorable. But since 1950, two of the weakest months are August (3rd worst) and September (worst). So in this weekends letter I decided to see how August and September fared when we already had June and July slipping. The table below gives us an answer. After down closes in both June and July. August has

In one of our previous articles, we built a neural-network-based volatility forecaster that beats baseline models in all volatility regimes: Volatility Forecasting using Neural Networks Vertox Jun 22 Volatility Forecasting using Neural Networks Today, we are gonna look at something weve never done in any article before: Neural Networks. Read full story Now no need to be sceptical about the

The Black-Scholes-Merton model is one of the cornerstones of modern quantitative finance. Despite its elegance and widespread use, its simplifying assumptions limit its ability to capture many features of real financial markets. As a result, researchers continue to extend the model to make it more realistic and applicable in practice. In this post, we discuss two such extensions. The first

Below I have shared my story of an idea that passed four checks Id set in advance and died on the fifth, and why the fifth one is now the first thing I run. The setup I like to day-trade MNQ, and Id built a framework around specific price levels that I have created. The core claim was simple: price reacts at certain levels within each hundred-point block. Falls into one, bounces. Thanks for

Every mean-reversion trade starts with a lie you tell yourself about where price "should" be. You call it fair value, you subtract it from the spot price, and you bet the gap closes. Most traders reach for a moving average and stop thinking. Xu, Firoozye, Koukorinis, Treleaven, and Zhu at UCL take the question seriously and reframe fair value as the output of a tunable low-pass filter,

Move like the wind, be still as the mountain, the old strategists taught. The moving-average ribbon promises to tell you which moment you're in. The catch: it's built from the past, so the wind it reads has already blown. Stack a fan of moving averages on a chart and assign colours to them. Wait for the bands to fan out and align, and price is "in a trend." When the lines

Split one Bitcoin day into two pieces and only one of them predicts anything. The piece that runs while US stock exchanges are closed, from yesterday's 4pm close to today's 9:30am open, carries a signal for tomorrow's VIX. The piece that runs while those exchanges are open carries nothing. Gu, Lin, and Liu ran five-minute Bitcoin data from 2018 to 2023 through that split and found

We spend our trading lives looking at charts, plugging in different indicators, not to realize that we are painting on top of the same old picture. Instead of trying a different color of paint, we need to change the canvas. And in doing just that, you may have won yourself a spot in the lead. Heres the thing: Everyone is looking at the same charts. Everyone has access to the same information

Every indicator you use on a rolling window, momentum, RSI, Bollinger bands, discards the order in which events occurred. Two 2-hour windows with identical total return and identical range expansion score identically, even if one saw price rally first and volatility respond, while the other saw volatility spike first and price catch up later. These are different market events (a breakout versus a

Trend following was one of the most persistent anomalies in finance for nearly two centuries, yet its performance deteriorated sharply after the 2008 financial crisis. An analysis of approximately 100 liquid futures contracts from 1995 to 2025 shows that this decline is highly selective. The decisive factor is not asset class, liquidity, market electronification, or strategy crowding, but

Most quantitative backtests start with prices: a clean historical database, ideally free of survivorship bias and the usual data traps. We have written a lot about that already, and shared practical ways for independent researchers to build more reliable datasets. At some point, though, research usually asks for more than prices. How does a strategy behave around major macro announcements? Does

Key Takeaways An OHLC bar can show that take-profit and stop-loss prices were both touched. It cannot show which came first. Run the same search under both TP-first and SL-first before promoting a candidate. We evaluated 48,825,000 identical configurations twice, forming 97,650,000 backtest runs. Total return changed in 37,168,124 pairs, or 76.13%. Among the changed pairs, the median SL-first

Twenty years of retail research points the arrow one way. You read the VIX, and the VIX tells you where the S&P 500 is going. Oversold VIX means complacency, sell stocks; spiked VIX means panic, buy the dip. Connors built a cottage industry on it, and every trading forum still runs some version of the "VIX says buy" screenshot. Rob Hanna ran the tests both directions and found the

Modern portfolio theory is built on a remarkably intuitive idea: investors should earn higher expected returns for bearing greater risk. This principle lies at the heart of the Capital Asset Pricing Model (CAPM), one of the most influential models in financial economics. According to the theory, stocks with higher systematic risk should compensate investors with higher long-term returns. Yet among

In this episode of House of Quants, listeners will discover: My personal perspective: The episode traces the journey from engineering, statistics, data science, and algorithmic trading toward quantitative research, explaining how each field contributed to a deeper understanding of uncertainty, validation, execution, and risk. Why long-term market forecasting often is useless: It examines the

When inflation shifts from low to high, a traditional 60/40 equity-and-bond allocation breaks down because both fall together. You need real assets to act as the ballast. Ray Dalio Inflation is one of the most powerful forces in asset allocationand one of the hardest to measure in real time.Everyone understands why it matters. The harder question is how to capture it before markets

Alpha rarely arrives as one heroic discovery. It is more like an ant colony carrying a leaf many times its own size: dozens of small contributions, each unimpressive on its own, somehow producing an impressive result. Unfortunately, research also resembles an ant colony in another respect. Much of the work involves running in circles. The first version

In a previous blog post, I described the NAAIM Exposure Index, which represents the average exposure to U.S. equity markets as reported by members of the National Association of Active Investment Managers (NAAIM) in a weekly survey. In this second post of this series on sentiment indicators, I will show how that survey of professional money managers can be turned into a relative sentiment

I wrote my newest paper mainly to challenge the conventional way of judging an investment or trading strategy through a single observational point of its performance metric an approach that discards precious information about a strategy's effectiveness and can lead to false conclusions. Rather than routinely collapsing a strategy's performance metric to one observational scalar, I