New Strategies:
#1303 – Dynamic Conversion Arbitrage on the Magnificent 7
Period of rebalancing: Daily
Markets traded: equities
Instruments used for trading: stocks, options
Complexity: Very Complex
Backtest period: 2023-2026
Indicative performance: 7.5%
Estimated volatility: 2.7%
Source paper:
Deng, Zhaoren: Dynamic Conversion Arbitrage on the Magnificent 7
https://www.quantconnect.com/research/20937/dynamic-conversion-arbitrage-on-the-magnificent-7/p1
Abstract: Conversion arbitrage is a classic Options trade that produces a fixed payoff by combining a long stock position with a long put and a short call at the same strike and expiry. This strategy hunts for conversions on the Magnificent 7 stocks whose annualized return exceeds a minimum threshold, opens the most attractive one, and dynamically swaps into a better conversion if one surfaces. It trades the Equities and their Options and parks idle capital in the SHV Treasury ETF. From January 2023 to June 2026 the strategy earned a 7.5% compounding annual return with a maximum drawdown of 1.1% and a portfolio beta of 0.017, a nearly market-neutral profile.
#1304 – Filing Language Stability as a Selection Signal
Period of rebalancing: Monthly
Markets traded: equities
Instruments used for trading: stocks
Complexity: Very Complex
Backtest period: 2020-2026
Indicative performance: 21.09%
Estimated volatility: 35.15%
Source paper:
Sun, Emily Xinyu: Filing Language Stability as a Selection Signal
https://www.quantconnect.com/research/20966/filing-language-stability-as-a-selection-signal/p1
Abstract: Every public company files a 10-K with the SEC each year and a 10-Q after each of its first three quarters. These are the audited annual report and the lighter quarterly updates that disclose a firm’s financial results and business risks. They draw investor attention to the reported numbers, yet the language of the filing carries information of its own. Cohen, Malloy, and Nguyen (2020) document that firms which barely change their filing language from one year to the next earn higher subsequent returns than firms that substantially rewrite their disclosures, an effect they call Lazy Prices. We turn this finding into a long-only stock-selection strategy on the 100 most liquid US Equities. Each month we rank them on the language similarity of their latest 10-K and 10-Q filings, drawn from the Brain Language Metrics on Company Filings dataset, and hold a long-only portfolio of the most stable filers, weighted to maximize the portfolio Sharpe ratio. From January 2020 to June 2026, the strategy produced greater risk-adjusted returns than the benchmark SPY ETF.
#1305 – Polymarket Mean-Reversion Strategy
Period of rebalancing: Daily
Markets traded: crypto
Instruments used for trading: crypto
Complexity: Simple
Backtest period: 2024-2026
Indicative performance: 22.09%
Estimated volatility: 17.97%
Source paper:
Vojtko, Radovan and Dujava, Cyril: Exploiting Mean-Reversion in Decentralized Prediction Markets: Evidence from Polymarket Binary Contracts
https://dx.doi.org/10.2139/ssrn.6726362
Abstract: This study examines the profitability of mean-reversion trading strategies applied to binary outcome contracts on Polymarket, the world’s largest decentralized prediction market platform. We analyze three distinct contracts representing varying risk profiles: a quasi-risk-free instrument (No to “Will Jesus Christ return in 2025?”) and two high-yield speculative contracts (No to “Will China invade Taiwan in 2025?” and “Will the US confirm that aliens exist in 2025?”). Using high-frequency price data sampled at 10-minute intervals over approximately one year, we implement a parameterized mean-reversion framework across twelve strategy variants, testing robustness under varying liquidity constraints and transaction cost assumptions. Our findings reveal that while mean-reversion signals generate substantial alpha under passive limit-order execution (zero-spread scenario), strategy performance degrades significantly when more aggressive market orders are accounted for.
#1306 – QQQ-Bitcoin Breakout Rotation
Period of rebalancing: Daily
Markets traded: equities, crypto
Instruments used for trading: ETFs, crypto, futures, CFDs
Complexity: Simple
Backtest period: 2019-2025
Indicative performance: 43.69%
Estimated volatility: 25.85%
Source paper:
Vojtko, Radovan and Dujava, Cyril: Silicon vs. Satoshi: Tactical Asset Rotation Between NASDAQ-100 and Bitcoin
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=7055018
Abstract: We investigate a Donchian breakout rotation strategy between QQQ (NASDAQ-100) and Bitcoin (BTC), with a cash fallback during consolidation, and test it across eight lookback horizons (5–50 trading days) and two priority variants over a seven-year sample spanning 2019–2026. The strategy consistently outperforms passive benchmarks on a risk-adjusted basis, achieving Sharpe ratios up to 1.69, Calmar ratios exceeding 2.2, and maximum drawdowns compressed by 50–79% relative to buy-and-hold allocations. Results are robust across the 5–30 day lookback range, demonstrating that retail attention dynamics between technology stocks and cryptocurrency represent exploitable structure amenable to systematic harvesting. In the modern retail attention economy, Bitcoin and the NASDAQ-100 are not merely separate assets; they are competing narratives. Both appeal to the same pool of speculative capital, the same appetite for asymmetric upside, and the same behavioral forces of FOMO, herding, and recency bias. When technology stocks dominate the imagination, capital clusters around QQQ and the artificial intelligence trade. When Bitcoin breaks out, the crowd’s attention pivots toward crypto’s promise of explosive upside. This paper tests whether that rotation in attention leaves a systematic footprint. Using Donchian breakout signals across QQQ and Bitcoin, with cash as a fallback during periods of consolidation, we examine whether investors can harvest momentum without remaining permanently exposed to either asset’s full drawdown profile. The results suggest that the answer is yes: retail attention does not move randomly. It rotates, it concentrates, and—when measured through price breakouts—it can be systematically exploited.
#1307 – Degree of Extrapolation Effect in Commodities
Period of rebalancing: Weekly
Markets traded: commodities
Instruments used for trading: futures
Complexity: Complex
Backtest period: 1987-2024
Indicative performance: 13.11%
Estimated volatility: 18.67%
Source paper:
Wang, Zhe and Xu, Qi and Ye, Yang: Extrapolative Commodity Returns
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6745679
Abstract: We investigate the role of extrapolation in commodity markets. Non-commercial traders exhibit extrapolative beliefs while commercial traders do not. We construct a novel holding-based commodity measure of the degree of extrapolation (DOX). DOX positively predicts the cross-section of commodity returns. The predictive power cannot be explained by existing risk factors or characteristics. The predictability is consistent with the liquidity provision due to gradual unwinding of extrapolators and limits to arbitrage. Investors even extrapolate the term structure. We further offer an extrapolation-based interpretation for commodity momentum and value. An extrapolation-augmented factor model explains commodity anomalies and individual commodity returns well.
#1308 – China-U.S. Dual-Anchor Commodity Strategy
Period of rebalancing: Weekly
Markets traded: commodities
Instruments used for trading: futures, CFDs, ETFs
Complexity: Complex
Backtest period: 2016-2026
Indicative performance: 36.5%
Estimated volatility: 21.99%
Source paper:
Tang, Shuiping: Dual Anchors: Dual Anchors: China-U.S. Factors Jointly Determining Global Gold and Crude Oil Pricing — Systematic Evidence Based on Exhaustive Testing with VIF Constraints
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6740898
Abstract: Traditional commodity pricing literature has long anchored gold and crude oil pricing to U.S.-centric variables, leaving China-linked factors systematically excluded from quantifiable pricing frameworks. This study employs a rolling regression framework with a 756-trading-day window and variance inflation factor (VIF) constraints, exhaustively testing 462 factor combinations for gold and 141 for crude oil. The optimal gold model (China-U.S. Dual-Anchor Three-Factor Gold Valuation Model) consists of the TIPS real yield, the China-U.S. 10-year sovereign yield spread, and the gold-oil ratio, achieving a rolling full-sample R² of 0.908 and a static full-sample R² of 0.814 (2016–2026), with all VIFs below 5. The optimal crude oil model (China-U.S. Dual-Anchor Four-Factor Crude Oil Valuation Model) consists of Shanghai crude oil futures (SC), the VIX, the synthetic time spread, and the China-U.S. yield spread, achieving a full-sample R² of 0.888 and a fixed-parameter out-of-sample R² of 0.611. Under Newey-West autocorrelation-consistent corrections and LASSO regularization, all China-linked factors retain statistical significance and model survival. The China-U.S. yield spread is the only factor entering both optimal models—its independent explanatory power in gold traces back to at least 2006 (three-factor static R² = 0.836), while in crude oil its pricing function strictly begins with the launch of Shanghai crude oil futures in March 2018, as confirmed by an institutional instrumental variable test identifying the causal activation effect (first-stage F = 2,425.75). The SC factor maintains a 100% positive sign across all six rolling sub-periods with zero sign reversals. This paper further develops the theoretical concept of the “institutional interface”—whether a monetary policy signal can enter the pricing equation of a commodity class depends on whether that class possesses a tradable benchmark contract denominated in the domestic currency—offering a new analytical perspective for research on global commodity pricing power. The global commodity pricing regime has evolved from a single-anchor to a China-U.S. dual-anchor structure.
#1309 – Intraday Donchian Trend-Following Framework
Period of rebalancing: Intraday
Markets traded: equities
Instruments used for trading: CFDs, ETFs, futures
Complexity: Complex
Backtest period: 2023-2026
Indicative performance: 28.49%
Estimated volatility: –
Source paper:
Wong, Zhen Hao: Experimental Evaluation of an Algorithmic Trading Strategy Against the S&P 500 Benchmark
https://ssrn.com/abstract=6722841
Abstract: This study presents an empirical evaluation of a rule-based algorithmic trading strategy designed to outperform the S&P 500 benchmark under controlled risk conditions. The proposed model is based on a Turtle Donchian trend-following framework combined with volatility-adjusted risk management using the Average True Range (ATR). The strategy is tested on the US100 index over the period January 2023 to April 2026 using multiple risk configurations, including fixed-risk and dynamic risk allocation models. Performance is evaluated using key financial metrics such as total return, maximum drawdown, Sharpe ratio, and recovery factor. Results indicate that the fixed-risk model is capable of outperforming the S&P 500 benchmark while maintaining comparable drawdown levels. Higher-risk configurations demonstrate significantly increased returns at the cost of elevated drawdown, illustrating the trade-off between return maximisation and risk control. Monte Carlo simulation and stress testing are also applied to evaluate the robustness of the fixed-risk and dynamic-risk models under randomised trade sequences and adverse execution assumptions.
#1310 – ENSO Signals in Soft Commodity Futures
Period of rebalancing: Weekly
Markets traded: commodities
Instruments used for trading: futures
Complexity: Very Complex
Backtest period: 2004-2026
Indicative performance: 1.18%
Estimated volatility: 1.12%
Source paper:
Apte, Mohit: ENSO Signals and Out-of-Sample Predictability in Soft Commodity Futures
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6789516
Abstract: This paper examines whether publicly available El Niño–Southern Oscillation (ENSO) information improves out-of-sample return prediction for ICE soft commodity futures. Using NOAA Niño 3.4 sea surface temperature (SST) anomalies and WRDS Datastream daily settlement prices for coffee, cocoa, sugar, cotton, and orange juice over January 2000 through January 2026, I combine three climate-derived features with three price-based features in a strict expanding-window walk-forward framework with explicit lag treatment to prevent look-ahead bias. Comparing Ridge regression, XGBoost, and LightGBM, the transaction-cost-adjusted Ridge specification achieves an out-of-sample Sharpe ratio of 1.05. Feature ablation reveals that climate variables are complementary rather than independently predictive: adding ENSO features to a price-only model raises the Sharpe ratio by approximately 0.10, whereas climate-only specifications produce negligible forecasting power. The predictive content is concentrated during strong El Niño regimes but remains positive across neutral and moderate La Niña conditions. Results are robust to alternative information-lag assumptions, forecast horizons, transaction-cost levels, and a frozen 2023–2026 holdout period. The findings suggest that ENSO information contains modest but economically meaningful incremental predictive value for soft commodity futures when combined with market-based signals, with relevance for commodity hedging, procurement timing, and risk monitoring.
#1311 – Seven Factor Model for Crypto
Period of rebalancing: Weekly
Markets traded: crypto
Instruments used for trading: crypto
Complexity: Very Complex
Backtest period: 2020-2026
Indicative performance: 80.3%
Estimated volatility: 61.33%
Source paper:
Babayeva, Farid and Aliyev, Nijat: Crypto Has Fundamentals: A Seven-Factor Model for Digital Asset Returns
https://ssrn.com/abstract=6818558
Abstract: Cryptocurrency markets are widely perceived as speculative and disconnected from fundamentals. This perception keeps many traditional investors on the sidelines: without understanding what drives digital asset returns, they cannot decompose portfolio performance, hedge unwanted exposures, or distinguish skill from systematic risk. This paper addresses that gap. We introduce the QuantNest-7 (Q-7) model, a seven-factor framework that extends the three established factors of Liu et al. (2022) (market, size, and momentum) with four new dimensions drawn from blockchain data and derivatives markets: on-chain quality, fundamental value, residual volatility, and perpetual futures funding sentiment. Using 90 tokens over six years (January 2020 through May 2026), we find that five of six style factors carry statistically significant risk premia. On-chain quality, value, volatility, funding sentiment, and short-term reversal are all significant at the 1% level, while size becomes insignificant after controlling for volatility, suggesting the apparent size premium is a volatility premium in disguise. Notably, we document short-term return reversal rather than the momentum effect reported by Liu et al. (2022), indicating that past losers outperform past winners in our broader, more recent sample. The Q-7 model exhibits essentially zero overfitting, explains 63% of per-token return variation and 85% of diversified portfolio returns out of sample. Conditioning on market sentiment reveals regime-dependent premia: reversal earns its strongest returns during fearful markets, value during optimistic ones, while quality delivers positive returns across all sentiment environments. For practitioners, the Q-7 model provides the first comprehensive factor-based toolkit for digital assets: portfolio managers can decompose crypto returns into systematic exposures, construct factor-neutral portfolios, run performance attribution, and begin treating digital assets with the same analytical rigor applied to equities and fixed income.
#1312 – Cross-Sectional Return Prediction in Cryptocurrency Markets Using Variational Latent Factor Models
Period of rebalancing: Daily
Markets traded: crypto
Instruments used for trading: crypto
Complexity: Very Complex
Backtest period: 2024-2024
Indicative performance: 47.6%
Estimated volatility: 41.03%
Source paper:
Boyer, Sacha: Cross-Sectional Return Prediction in Cryptocurrency Markets Using Variational Latent Factor Models
https://ssrn.com/abstract=6860920
Abstract: This paper investigates the application of variational autoencoder-based latent factor models to cryptocurrency markets. We adapt the FactorVAE architecture, originally designed for equity markets, to a universe of 50 major cryptocurrencies using technical indicators derived from price and volume data. Our empirical analysis covers the period 2019-2024, with out-of-sample testing in 2024. We implement and compare seven distinct trading strategies exploiting the model’s cross-sectional predictions, including directional, market-neutral, volatility-adjusted, and momentum-filtered approaches. Results demonstrate that the model generates economically significant signals, with six out of seven strategies achieving positive Sharpe ratio s exceeding 1.0. The Momentum Filter strategy emerges as the most effective approach, achieving a 269.8% return with a Sharpe ratio of 2.12. We further examine latent factor stabilization techniques and find that temporal regularization produces more interpretable factor dynamics without sacrificing predictive accuracy. These findings suggest that deep generative models offer promising tools for understanding and exploiting cross-sectional structure in cryptocurrency markets.
New research papers related to existing strategies:
#458 – End-of-Month Treasury Returns
Tian, Yuan and Sun, Jialu: Private Money, Public Debt: Quarter-End Price Pressure in Treasury Bills
https://ssrn.com/abstract=6913784
Abstract: This paper shows that quarter-end bank window dressing creates a substitution channel from private to public money-like assets. Around regulatory reporting dates, foreign banks reduce borrowing from U.S. money market funds, forcing these funds— which are constrained to hold only short-term, high-quality assets—to reallocate dis- placed cash into Treasury bills. As a result, short-maturity U.S. Treasury yields fall by 3–9 basis points at quarter-ends. Exploiting predetermined fund-bank relationships and cross-sectional variation in Treasury bill holdings across funds, we causally es- timate that money market fund demand inflows equal to 1 percent of Treasury bill market value lower yield spreads by 4.9 basis points. We use this estimate to quantify the fiscal consequences of bill-heavy debt issuance under alternative Treasury financing scenarios
#887 – Pairs Trading in Cryptocurrencies
#1301 – GARCH-LSTM Pair Trading Strategies on ETH-BTC pair
Lebiedź, Damian and Ślepaczuk, Robert: Dynamic Multi-Pair Trading Strategy in Cryptocurrency Markets with Deep Reinforcement Learning
https://ssrn.com/abstract=6825378
Abstract: This study aims to determine whether the application of Deep Reinforcement Learning (DRL) as a specialized execution overlay can enhance pair trading in highly volatile cryptocurrency markets. Although classical implementations of the strategy have proven successful in traditional equities, they frequently exhibit rigidity and suffer from severe divergence risks when applied to high-variance environments. To address this need, this research introduces novel concepts. To construct a robust system, we developed a hierarchical “Filter-then-Rank” pair selection methodology and a proprietary “Fixed Risk, Adaptive Mean” execution model. The system employs a Proximal Policy Optimization (PPO) agent with a Long Short-Term Memory (LSTM) layer to govern execution decisions within strict deterministic risk management boundaries. Evaluated on 1-hour interval data from the Binance USD-M Futures market, the optimized RL policy achieved an Out-Of-Sample performance that substantially outperformed the heuristic baseline. A stationary circular block bootstrap robustness check confirms that the Agent’s risk-adjusted outperformance is statistically significant at the 10% level. Although falling marginally short of the stricter 5% threshold, this result highlights the extreme idiosyncratic variance characteristic of digital assets. Ultimately, this thesis contributes to the quantitative finance literature by introducing a hybrid architecture that combines statistical arbitrage with DRL execution policies. Furthermore, it delivers a novel framework for safe reinforcement learning via deterministic shielding, proving that anchoring a neural policy to statistically robust boundaries successfully mitigates severe divergence risks.
And several interesting free blog posts that have been published during the last month:
Can AI Do Financial Research?
Large language models are already capable of summarizing financial research, but are they ready to conduct it? In their latest paper, researchers from Google, Boston College, and Columbia introduce a framework where a large language model doesn’t just fetch data—it acts as an autonomous AI research agent capable of navigating the “hypothesis discovery loop.” By placing an LLM within a human-designed laboratory—complete with a symbolic language of 66 accounting primitives and a standardized backtesting pipeline—the authors tested whether AI can move beyond black-box predictions to generate economically legible and statistically robust signals. This isn’t just about throwing a transformer at a price series; it is a systematic attempt to automate the “propose–test–reflect” cycle that defines empirical finance.
Commodity Crisis Analysis – How Portfolios React to Commodity Shocks
Financial markets are often viewed primarily through lens of equity index movements, as they attract most of the attention. However, commodities represent an important component of the global economy, and shocks in commodity markets can have a significant impact on broader financial assets.
From time to time, market stress originates outside equities. A recent example are the repeated US attacks on Iran, which increased uncertainty in energy markets and raised the risk of an oil supply shocks. A similar dynamic was observed in 2022 during the Russian invasion of Ukraine, when commodity prices moved sharply higher or during the US invasion of Iraq in 2002, when uncertainty in oil markets led to increased price volatility.
These events highlight the importance of analyzing portfolio behavior not only during equity bull and bear markets, but also during commodity-driven shocks.
Getting the Target Right in Return Prediction
Recent interesting research from Cakici and Zaremba, highlights an often-overlooked aspect of machine learning for equity return prediction: the choice of prediction target. Rather than focusing on increasingly sophisticated model architectures or feature engineering, the authors show that how returns are represented during training has a much larger impact on predictive performance. In particular, models trained to predict stock ranks instead of raw return levels generate substantially stronger portfolio performance—roughly doubling both returns and Sharpe ratios in large-cap universes.
From Backtest to Benchmark: Validating New Strategies with Quantpedia API
A profitable backtest is rarely the end of a research process. In professional quantitative research, the more important question often comes after the first positive result: is the strategy genuinely new, or is it simply another version of an already known factor, timing rule, or anomaly?
This is especially relevant when a researcher develops a new systematic strategy with a clean historical equity curve. The strategy may have acceptable risk-adjusted performance, stable drawdowns, and a logical trading rule, but those statistics alone do not prove that the idea is unique. A silver strategy, for example, may look different on the surface while still behaving like a known commodity timing model, a trend-following strategy, a volatility filter, or a broader macro factor exposure.
This article shows how Quantpedia API can be used as a benchmark dataset for validating new research. Instead of evaluating a new backtest in isolation, the strategy is compared against the Quantpedia universe of documented quantitative strategies. The workflow identifies nearest neighbours, assigns the strategy to a factor cluster, calculates a uniqueness score, and produces a research robustness report that can be used for deeper validation.
The goal is not to replace human research judgment. The goal is to create a structured robustness checker that helps researchers understand whether a new strategy is truly differentiated, redundant with known effects, or simply a variation of an existing Quantpedia strategy profile.
Is Trend Still Your Friend?: A Microstructural Account of the Demise of Short-Term Trend-Following
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 volatility-normalized tick size. After 2008, trend-following profits collapsed almost entirely on small-tick contracts across all signal horizons, while remaining largely intact on large-tick contracts. This finding suggests that modern trend-following portfolios are fundamentally split into two distinct regimes governed by market microstructure rather than traditional asset classifications.
Plus, the following trading strategies have been backtested in QuantConnect in the previous two weeks:
Plus, the following trading strategies have been backtested in QuantConnect in the previous two weeks:
1260 – Hedged Industry Momentum
1278 – 52 Week High and the Conditional Betting Against Beta
1285 – Cross-sectional FOMC Announcement Premium in Stocks
1289 – Volatility-Targeted QQQ–TLT Rotation with a Multi-Scale CNN-LSTM Forecast
1293 – Trend-Following Pullback Reversal Strategy
1294 – Value-Growth Spread Timing Strategy
1298 – Large Cap-Mid Cap Spread Strategy
