QRAFT AI-Enhanced U.S. Large Cap Momentum ETF (AMOM)

Machine learning applied to stock selection is a natural bet, but a momentum factor built on a computer’s eye for patterns lives and dies by the stability of those patterns — which tend to break when they matter most.

AMOM is an exchange-traded fund launched in 2021 by QRAFT Technologies, a fintech firm specializing in machine-learning-driven index construction. The fund selects from the universe of large-cap U.S. stocks using an artificial intelligence model trained to identify momentum characteristics — stocks that are trending higher and are likely to continue trending — and it reconstitutes the portfolio monthly. QRAFT’s model ingests thousands of data points: price patterns, technical indicators, earnings estimate revisions, options positioning, analyst sentiment, insider trading activity, and others. From this, it scores each eligible stock on its momentum profile and builds a diversified portfolio of the highest-scoring names.

The origin of QRAFT and the momentum thesis

QRAFT Technologies was founded in 2015 by engineers and data scientists with backgrounds in machine learning and quantitative finance. The firm’s thesis is that human analysts, even professionals, struggle to process the volume and complexity of data points that matter for stock selection, and that machine learning can spot patterns faster and more reliably. The team’s earliest models were applied to stock and sector rotation. Over time, they moved into launching publicly traded products, starting with AMOM.

The momentum factor itself is well-established in academic finance: over decades, stocks that have recently outperformed tend to continue outperforming for a period, and stocks that have lagged tend to lag further. This pattern is not universal — momentum crashes exist, and they are often severe — but the edge has been durable enough that momentum has become a mainstream factor in institutional asset management. QRAFT’s innovation is not the discovery of momentum but the application of machine learning to identify the stocks whose momentum is most robust and likely to persist.

How the algorithm works and what it holds

The fund’s model scores large-cap U.S. stocks (those in the Russell 1000) on multiple momentum dimensions: price strength over various lookback windows, earnings surprises and estimate revisions, technical support and resistance levels, and what QRAFT calls “cross-asset momentum” — signals from options markets, futures, and related instruments. The model weights these signals by their historical predictive power (a technique called backtesting). It then selects 40 to 60 stocks for the portfolio, emphasizing those with high momentum scores while maintaining diversification across sectors.

The portfolio is reconstituted monthly — the scores are recalculated and stocks are added or removed — which allows the fund to respond quickly to changes in momentum. A stock that was a strong momentum candidate last month but has started to fade will exit the portfolio, and fresh momentum leaders will be added. This high turnover is a cost (trading expenses, tax drag in taxable accounts) but also a feature: the fund is explicitly designed to capture trend-following returns and shed names that no longer fit.

The strengths and the fault line

The machine-learning approach has real advantages. The model can process far more information than a human portfolio manager and can backtest its logic across decades of market history, adjusting weights to maximize historical returns. The fund avoids human biases: it does not fall in love with a company’s story, does not overweight high-profile names, and does not let recent headlines push it into panic-driven trades.

The fault line is the one that affects all momentum investing: momentum is a relative-strength phenomenon, and it breaks hardest when the market reverses. When a bear market arrives, yesterday’s winners become today’s victims, and momentum-following strategies sell into falling prices — precisely wrong timing. A machine learning model trained on 20 years of data has never experienced the specific patterns of the next bear market, because the next bear market is always unique. AMOM would not have predicted the 2020 pandemic crash or the 2022 Fed hiking cycle. When the market regime shifts, an algorithm’s predictions become less reliable, even one trained on vast amounts of data.

The other risk is overfitting: the model’s rules, refined and back-tested over years, may have captured quirks and data artifacts of historical markets rather than true relationships. That is why momentum factor performance varies so widely from year to year — sometimes it crushes the market, and sometimes it underperforms badly — and why live fund performance can diverge from backtested results.

Costs and reconstitution drag

AMOM’s expense ratio is moderate for a smart-beta or factor-tilted ETF. The real cost is hidden in the monthly reconstitution: the portfolio turns over substantially (turnover ratios for momentum funds typically run 50 to 100 percent annually), which creates trading costs, bid-ask spreads, and in taxable accounts, capital-gains distributions. Over time, that drag matters.

The fund is liquid — it trades actively and has reasonable assets under management — so execution costs should be manageable for retail investors entering or exiting at market prices.

Who AMOM is for and how to research it

AMOM is for investors who believe in the momentum factor and in the value that machine learning can add to factor implementation, and who are comfortable with the volatility and drawdowns that come with any single-factor strategy. It is not a core holding for a long-term diversified portfolio; it is a tactical allocation, often paired with other factors or broad-market exposure. Investors should understand that momentum’s returns are cyclical — it excels in bull markets and strong uptrends but suffers in reversals.

The fund’s prospectus and QRAFT’s educational materials explain the methodology and the historical back-tested performance. The live performance (available on the fund’s website, ETFdb, Morningstar, and Yahoo Finance) should be compared against both a broad large-cap benchmark and other momentum-focused or factor ETFs. Anyone considering AMOM should also read the academic literature on momentum crashes and be prepared for the reality that the fund’s returns in any given year are likely to be erratic and highly correlated with broader market sentiment.