Study results and what follows

A new, academically led analysis shows that modern predictive systems can anticipate a very large share of what active managers execute, with the headline figure centered on 71%. The research isolates short-horizon trades and trains models against observable signals; the models then recover a substantial portion of trade direction and timing. That alone does not prove every manager is obsolete, but it does puncture the assumption that skilled human judgment remains uniquely opaque.

Market mechanics respond quickly. When algorithmic predictors can map order flow, price adjustments occur sooner, compressing the windows where traders convert information into profit. Expect execution patterns, not just portfolio construction, to migrate toward faster, model-informed approaches. Some active shops will accelerate investment in data and execution technology. Others will double down on edge sources that are harder to model.

Regulatory and governance threads will surface. If predictability stems from common signals or inadvertent leakage, exchanges and compliance teams face new pressure to audit order routing and information flows. Market structure fixes — from batch auctions to disclosure rules — will re-enter conversations because predictability changes who captures value from trades.

For allocators, the study forces a re-evaluation of expected alpha. Fee structures tied to historical outperformance may look unattractive if predictable trades mean quicker arbitrage and narrower margins. Some investors will shift toward managers who demonstrate nonpublic informational advantages or specialize in longer-horizon, illiquid niches.

The academic finding sits alongside an active market narrative: in recent quarters corporate earnings calls and investor transcripts show a sharp rise in explicit AI-related discussion, and market participants are already acting on those narratives. Traders and algorithmic strategies have pared positions in companies perceived as most exposed to AI-driven disruption, and sell-side earnings forecasts have not uniformly caught up to the price action. That divergence — price discovery leading fundamentals — magnifies volatility and demonstrates how a predictive edge in trade flow can interact with thematic repricing.

Wider market effects are visible beyond equities. Credit desks report wider spreads and weaker secondary prices for smaller software vendors and single-product businesses perceived as vulnerable to automation or higher capex needs. The combined picture: predictability at the trade level can accelerate thematic market moves that reprice risk before analysts adjust models, with real effects on funding costs and liquidity for smaller firms.

Technology and supply-chain dynamics are relevant enablers. Demand for compute, cloud capacity and alternative data has stepped up, benefiting hyperscalers and large platform providers able to self-fund R&D and procurement. At the same time, onshore foundry incentives and construction/talent limits create timing risk that concentrates advantages with larger, cash-rich vendors.

Not every implication is immediate or uniform. Replication across markets, asset classes, and timeframes will determine how broadly active management economics change. The study is a directional accelerant; its full impact plays out over the coming 6–12 months as firms test and respond. In particular, the headline predictability is model- and horizon-specific: operationalizing similar performance requires comparable datasets, feature engineering, and robust out-of-sample validation.

In sum, the paper reframes a familiar debate: whether AI supplements human judgment or replaces parts of it. Here, the evidence leans toward substitution in the trade-execution layer, prompting strategic moves in data acquisition, compliance, and product positioning. At the same time, market-level AI narratives — separate but complementary — mean price discovery may preempt fundamentals, amplifying the immediate impact on managers and smaller technology firms.