New Strategies:
#1278 – 52-Week High and the Conditional Betting-Against-Beta
Period of rebalancing: Monthly
Markets traded: equities
Instruments used for trading: stocks
Complexity: Complex
Backtest period: 1963-2022
Indicative performance: 15.66%
Estimated volatility: 24.86%
Source paper:
Sun, Licheng: Beyond Volatility: The 52-Week High and the Conditional Betting-Against-Beta Puzzle
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6516598
Abstract: We show that the conditional betting-against-beta (BAB) puzzle goes beyond volatility. Using price-to-52-week-high (PTH), we uncover a two-dimensional structure in BAB: at the stock level, BAB is strongest among stocks trading far below their own 52-week highs, while at the aggregate level, the common BAB factor is strongest when the market as a whole is trading closer to prior highs. To organize these findings, we develop a stylized latent-state model in which stock-level PTH captures drawdown-related mispricing in the cross section, whereas aggregate PTH helps reveal whether marketwide price weakness is more likely to reflect a persistent adverse state. In direct horse-race tests, PTH dominates the realized-volatility conditioning emphasized by Barroso, Detzel, and Maio (2025), while TED, the funding proxy emphasized by Frazzini and Pedersen (2014), explains only part of the evidence and does not subsume the role of aggregate PTH. Overall, PTH emerges as a central organizing variable for conditional BAB.
#1279 – Relative-Value Reversal in Precious Metals and Mining ETFs
Period of rebalancing: Daily
Markets traded: equities, commodities
Instruments used for trading: ETFs
Complexity: Simple
Backtest period: 2015-2026
Indicative performance: 25.6%
Estimated volatility: 47.32%
Source paper:
Shrestha; Sangam: Relative-Value Reversal in Precious Metals and Mining ETFs
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6476099
Abstract: This paper studies reversal and relative-value strategies in precious metals and mining equities, with a focus on gold, silver, and junior versus senior mining ETFs. I first test a baseline cross-sectional reversal strategy across a universe of precious-metals-related mining equities. I then test benchmark-adjusted residual reversal and several ETF spread reversal strategies, including junior-versus-senior miner spreads and miner-versus-metal spreads. Strategy performance is evaluated across multiple lookback horizons using annualized return, volatility, Sharpe ratio, turnover, and implied holding period, with transaction costs explicitly incorporated. The results show that raw cross-sectional miner reversal and residual miner reversal are weak after costs, while ETF-based relative-value reversal performs materially better. The strongest strategy is the reversal of junior silver miners relative to silver itself, represented by the SILJ-SLV spread, with the best net Sharpe ratio of 0.541 at a 10-day horizon. Overall, the evidence suggests that relative-value reversal is more robust than raw standalone reversal, and that silver-based dislocations are stronger than comparable gold-based dislocations.
#1280 – Gold-Bitcoin Dual Momentum Strategy
Period of rebalancing: Weekly
Markets traded: commodities, cryptos
Instruments used for trading: ETFs
Complexity: Simple
Backtest period: 2018-2026
Indicative performance: 12.01%
Estimated volatility: 8.77%
Source paper:
Vojtko, Radovan and Dujava, Cyril: Dual Momentum Allocation Between Physical Gold and Bitcoin (Digital Gold)
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6729918
Abstract: From the trading desk to the portfolio committee, investors face a familiar question: how should alternative stores of value fit into a diversified portfolio? This research explores that question through a systematic dual-momentum framework comparing Bitcoin and physical gold in a rules-based tactical allocation model. Rather than debating ideology, we focus on practical portfolio construction and risk-adjusted returns. The goal is to examine whether “digital gold” can complement its physical counterpart within a disciplined investment process, and whether the distinct behavior of these assets can be used to build a more effective systematic strategy.
#1281 – Orthogonal Residual Commodity Momentum
Period of rebalancing: Weekly
Markets traded: commodities
Instruments used for trading: futures
Complexity: Very Complex
Backtest period: 2018-2026
Indicative performance: 16.25%
Estimated volatility: 14.84%
Source paper:
Ding, Yiyi and Kang, Wenjin and Yu, Jianfeng and and Zhao, Shen: Momentum and Reversal on the Short-Term Horizon: Evidence from Commodity Markets
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6425598
Abstract: This paper documents the coexistence of momentum and reversal on the short-term horizon. Utilizing unique investors’ position data from commodity futures markets, we decompose returns into a flow component tied to speculators’ net trading (Q) and an orthogonal residual component (R nonQ). We find that R nonQ can significantly and positively predict next-week return (short-term return-based momentum), while Q forecasts next-week return in a negative manner (short-term trading-based reversal). The short-term momentum effect applies to the entire cross-section of sample commodities and also to different subsamples of commodities with different characteristics. Further analysis suggests that this short-term momentum primarily reflects speculators’ trend-chasing trading behavior, which strengthens when volatility/uncertainty is low and when the expected profitability of momentum strategy is high. We also show that short-term momentum signals (R nonQ) can be aggregated to enhance the traditional intermediate-term momentum strategy and substantially improve its performance. By combining our empirical findings with a parsimonious model, we synthesize the microstructure-based liquidity-provision and behavioral trend-following mechanisms. Moreover, with the help of more granular investors’ position data, we find that the liquidity-based short-term reversal should be characterized as a reversal strategy based on past speculators’ trading rather than past commodity returns. Overall, our study demonstrates that momentum and reversal coexist on the same short-term horizon in commodity markets, overturns the previous canonical “reversal-at-short-horizon, momentum-at-intermediate-horizon” segmented perception, and therefore deepens our understanding about the overall return predictability in financial markets.
#1282 – Large Language Models and Generative Factor Discovery in Crypto Markets
Period of rebalancing: Daily
Markets traded: cryptos
Instruments used for trading: cryptos
Complexity: Very Complex
Backtest period: 2024-2025
Indicative performance: 15.43%
Estimated volatility: 26.86%
Source paper:
Sun, Xinyu and Wang, Zhiyi and Zhang, Huaxi: Large Language Models and Generative Factor Discovery in Crypto Markets
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6461691
Abstract: This study presents an end-to-end framework for automated factor discovery in cryptocurrency markets using GPT-5.2. Using USD-quoted perpetual contracts on BitMEX, we generate alpha factors via zero-shot generation and few-shot iterative enhancement. The top long-short factor delivers 658.01% annualized excess return with a Sharpe ratio of 4.13, and we find that leading signals remain robust in a strictly forward out-of-sample window defined by the model’s training cutoff, even during a sharp downturn. Cross-model comparisons (GPT, Gemini, Grok) suggest that frontier LLMs can internalize and operationalize crypto-specific market microstructure logic—especially liquidity–leverage traps—rather than merely fit historical correlations. We also document important boundary conditions: few-shot refinement can increase complexity and overfitting, and distinct-looking factors may be logically redundant, motivating explicit redundancy control in generative factor mining.
New research papers related to existing strategies:
#11 – Stock Return Reversal within Industries
#946 – Industry-adjusted Reversal
Stosik, Jan and Zaremba, Adam: Short-Term Reversal Persists Globally-If Properly Measured
https://ssrn.com/abstract=6630998
Abstract: Short-term reversal appears to have weakened in international equity markets because standard measures mix firm-specific and industry-wide returns. When evaluated relative to industry peers, the reversal persists. Using data from 64 countries, the conventional strategy delivers negligible returns, whereas an industry-adjusted strategy earns 0.53% per month with a Sharpe ratio of 0.74. The effect reflects the reversal of firm-specific shocks masked by persistent industry components. A regret-based signal captures the same industry-relative variation and adds no independent return. Reversal did not disappear; it was mismeasured.
#814 – Post Earnings Announcement Drift in China
Li, Chong: Stock Market Reactions to Financial Statement Announcement: An Empirical Investigation based on Chinese A-Shares
https://ssrn.com/abstract=5235276
Abstract: Since the discovery of post earnings announcement drift, there has been a lot of research dedicated to this stock market anomaly. With Chinese A-share stock performance data from 2017 to 2024, we found a market-adjusted abnormal return of about 4.5% in the 120-trading day period following financial statement announcement, ranging from -10.0% to 11.8% depending on years. Using a panel model with fixed effects for both individual firm and fiscal year to identify the driving factors of this post-announcement drift, we found that the security financing facility has mitigated the positive announcement drift by reducing the average abnormal daily returns by 0.044, 0.059 and 0.063 percentage points for the 30, 60, and 90-day period, respectively. Our final model can explain 19.0%, 23.4% and 25.3% of the variation in post announcement average abnormal returns for the 30, 60, and 90-day period, respectively. Relevant market factors such as investor divergence of opinion and liquidity, information quality corresponding to the financial statement, investor inattention, and corporate fundamentals also play a role in explaining the post announcement abnormal returns.
#1176 – OHLCT Multi-Scale CNN Strategy
Guan, Ketian and Li, Sida and Si, Jia: Understanding the Convolutional Neural Network’s Stock Return Predictability
https://ssrn.com/abstract=6389479
Abstract: Jiang, Kelly, and Xiu (2023) demonstrate that a convolutional neural network (CNN) trained on stock price chart images achieves striking out-of-sample return predictability. Building on their framework, this paper aims to understand the origin of the CNN’s predictive power. We investigate which visual elements drive the CNN’s predictions, whether richer image content can improve them, and where the model focuses its attention within the image. Using controlled ablation experiments, we establish a clear hierarchy among image elements: OHLC (open-high-low-close) price geometry is the dominant source of predictive power, the moving average line is redundant, and volume bars contribute marginally. We further show that augmenting images with 52-week high-low price anchors, grayscale price-level channels, or market index overlays uniformly degrades performance — a “less is more” result indicating that the CNN’s optimal input is the parsimonious OHLC representation. Gradient-based saliency maps reveal that the CNN concentrates attention on closing prices near daily high-low extremes in the most recent trading days, with volume spikes as secondary cues. Comparing saliency patterns across image designs uncovers a novel attention-performance paradox: more focused attention is associated with worse out-of-sample performance, while broader, more diffuse scanning of the OHLC geometry yields superior predictions. Our baseline long-short strategy achieves an annualized Sharpe ratio of 3.744 over 2020-2024, confirming that CNN-based predictability remains robust after the original sample period of the paper.
#699 – Stock and Bond Returns Predict Currency Returns
Yamani, Ehab Abdel-Tawab: The role of stock and bond returns in the cross-section of currency returns
https://ssrn.com/abstract=6459665
Abstract: This paper offers a new risk-based explanation for the cross-section of currency excess returns using local financial market drivers emanating from returns in stock and bond markets across 10 developed countries for the period of 1996-2024. The results show that local stock and bond returns are indeed empirically powerful risk factors in explaining the cross section of their corresponding currency excess returns, even after controlling for popular sources of cross-sectional pricing. This pricing power translates into economic profits for investors when using stock and bond returns as joint predictors for currency trading. This new evidence on the importance of stock and bond information in the currency market provides important insights to researchers and practitioners alike.
#1193 – Grid Trading Strategy for BTC
Zhai, GuangSen: The Prediction Paradox: Why Machine Learning Models Fail to Predict Cryptocurrency Prices in Live Trading
https://ssrn.com/abstract=6566940
Abstract: A growing body of literature claims high accuracy in cryptocurrency price prediction using machine learning models, with reported backtest accuracies ranging from 85% to 99%. However, these claims are almost exclusively validated through historical backtesting, with live trading results conspicuously absent. This paper presents a rare empirical study of three sequentially-developed ML-based trading systems deployed on Binance ETHUSDC perpetual futures over a three-month period (January-April 2026). Each system represented a distinct architectural evolution-from bar-based to tick-level, from 142 features to 45 streaming features, from 10x to 50x leverage-yet all converged on the same outcome: live trading performance statistically indistinguishable from random. System V1 achieved AUC=0.85 in walkforward cross-validation but generated only $0.12/day in live PnL. System V3, with 75x larger model complexity, produced cumulative-22.2 basis points across 5 live trades. We identify five structural failure modes-asymmetric TP/SL illusion, execution-prediction mismatch, regime non-stationarity, overfitting through complexity escalation, and survivorship bias in literature-and argue that the cryptocurrency prediction paradigm suffers from a fundamental methodological flaw. As a counterpoint, we present a non-predictive grid market-making strategy that achieved consistent profitability on the same exchange under identical market conditions, suggesting that edge in crypto markets derives from market microstructure exploitation rather than price forecasting.
And several interesting free blog posts that have been published during the last month:
An Index of Commodity Futures Returns Since 1871
Commodity markets are back in investors’ focus. After years in which equities and growth assets dominated portfolios, the recent rise in geopolitical tensions, inflation uncertainty, supply-chain fragmentation, and renewed resource nationalism has reminded allocators that commodities remain a critical macro asset class. That is why a newly released research paper, An Index of Commodity Futures Returns Since 1871, is particularly timely. Using a hand-collected database covering more than 150 years of U.S. commodity futures history, the authors provide one of the most comprehensive long-term perspectives yet on commodity investing — showing not only that diversified commodity futures historically delivered equity-like risk premia, but also that their return drivers were meaningfully different from stocks, offering valuable diversification across economic regimes.
Who Profits from Prediction Markets?
In the high-stakes arena of prediction markets, a counterintuitive pattern emerges: retail traders who correctly pick winners more than half the time still lose money, while automated traders with coin-flip accuracy pocket nine-figure profits. Using 222 million prediction market tradeswith directly observable terminal payoffs, the paper “Who Profits from Prediction? Execution, Not Information” presents a clean answer to why it is so. The authors decompose trader returns into a directional component and an execution component, revealing that the execution component, not the directional component, determines which trader types earn positive returns.
A Century Without Data: Reconstructing Emerging Markets Equity History
For U.S. equities, fixed income, and commodities, reconstructing long-term historical datasets is relatively straightforward, and we have already explored these challenges in several previous studies, including 100 Years of Multi-Asset Trend Following, Extending Historical Daily Bond Data to 100 Years, and Extending Historical Daily Commodities Data to 100 Years. Moreover, the broader methodology of reconstructing missing market histories shares many similarities with the techniques discussed in How to Replicate Any Portfolio. Emerging markets, however, represent a particularly interesting opportunity for historical reconstruction, as reliable long-term data is often unavailable for much of the 20th century despite the growing importance of these markets in modern portfolio construction and asset allocation. In this article, we present the framework we developed to extend emerging market histories in a consistent and economically meaningful way, enabling more robust long-term quantitative research and modelling.
Active Dual Momentum GTAA Strategy
Our study explores a weekly-rebalanced dual-momentum-based Global Tactical Asset Allocation (GTAA) strategy applied to a diversified set of ETFs. The strategy selects assets based on relative momentum and applies an absolute momentum filter to avoid declining investments. Ultimately, a single combined strategy was created by merging two sub-strategies, incorporating both shorter- and longer-term momentum signals. Backtesting over an extended period demonstrates that this approach delivers attractive risk-adjusted returns, achieving attractive Sharpe and Calmar ratios, while maintaining lower drawdowns compared to a simple equally weighted benchmark.
Quantpedia Awards 2026 – Winners Announcement
Welcome to the Quantpedia Awards 2026 winners announcement. For the third time, we are proud to celebrate excellence in quantitative research and recognize the researchers behind innovative studies in quantitative trading. We are also pleased to see that the Quantpedia Awards have become an established and recognized brand within the quant community. This is the moment we have all been waiting for: who made it into the top five, and what will the authors of the winning papers receive?
Plus, the following trading strategies have been backtested in QuantConnect in the previous two weeks:
1271 – High Volume Stocks Outperform Overnight
1272 – Similarly Priced Stocks Momentum
1273 – Anomaly-Driven Demand Factor
1274 – Liquidity Improvements Factor
