In 1976, ecologist Eric Charnov published a short, elegant paper answering a question that seems almost too obvious to need an equation: when should a foraging animal give up on a patch of food that’s getting picked over, and move on to a fresh one? His answer, the marginal value theorem, has since become one of the most cited ideas in behavioral ecology. Half a century later and in a completely different discipline, quantitative trading firms have spent considerable money and mathematical effort solving what turns out to be an almost identically structured problem: when should a trading strategy abandon a signal whose profitability is fading, and redeploy capital into something fresher? Neither field, as far as the published literature shows, has ever cited the other.
Scientific Foundation
Charnov’s marginal value theorem describes an animal foraging in an environment where food is clustered into discrete patches, separated by travel time during which no food is available at all. Within any given patch, the rate of intake starts high and declines as the easiest food gets consumed first — a diminishing-returns curve. The theorem’s central insight is that an optimally foraging animal shouldn’t wait until a patch is fully depleted to leave. It should leave the moment the patch’s current, marginal rate of return drops to match the average rate of return available across the whole environment, accounting for the time it costs to travel to the next patch. Leave too early, and you’re abandoning food you could still profitably harvest; leave too late, and you’re wasting time on a patch that’s no longer competitive with what’s available elsewhere. The theorem has been validated across an enormous range of species and contexts, from bumblebees to macaques foraging for social information, and recent work has extended it to handle the messier, uncertain environments real animals actually face, where the “average rate elsewhere” isn’t known with certainty and must be estimated using Bayesian updating as new information arrives.
Cross-Domain Connection
Quantitative trading has an almost mechanically parallel problem, dressed in entirely different vocabulary: alpha decay. A trading signal — some statistical edge that predicts future price movement — doesn’t stay profitable forever. As more traders identify and exploit the same inefficiency, or as market conditions shift, the signal’s predictive power erodes, typically following a documented decay curve; industry analysis puts annual alpha decay costs at roughly 5.6 percent in U.S. equity markets and 9.9 percent in European markets, with those rates reportedly worsening over time as trading technology accelerates the competitive pressure. A 2025 academic paper frames the resulting decision problem in strikingly familiar terms: a trader holding a position whose signal is decaying faces a trade-off between continuing to hold and realizing diminishing returns, versus paying a transaction cost to exit and redeploy capital elsewhere, formalized as an infinite-horizon Markov decision process solved to maximize long-run average expected reward per period. A separate, earlier academic treatment working with real commodity futures data found that signals which mean-revert, and therefore decay, faster command a larger required return to compensate for the transaction costs of trading them more frequently — a direct, quantified trade-off between marginal signal strength and the cost of switching, worked out independently using an entirely different mathematical toolkit than Charnov’s.
Line the two problems up and the structural overlap is hard to miss: both involve an opportunity whose rate of return declines the longer you stay in it; both explicitly weigh that declining marginal rate against the cost of moving to the next opportunity; and both are solved, in their respective fields, by explicitly maximizing a long-run average rate rather than a one-time payoff — Charnov’s “average capture rate for the habitat” and quant finance’s “maximum average expected reward” are, mathematically, doing the same job with different names.
What Remains Undemonstrated
This is convergent mathematics, not a borrowed idea, and it’s worth being precise about that distinction. No published finance research cites the marginal value theorem or behavioral ecology literature in developing alpha decay models, and no ecology paper references quantitative trading. The two fields appear to have independently arrived at structurally similar optimization frameworks because they were both, without realizing it, solving the same general class of problem: an optimal-stopping decision under diminishing marginal returns and switching costs, maximized over the long run rather than a single instance.
There’s also a real mechanistic difference worth naming rather than glossing over. A foraging patch depletes because the resource itself is physically being consumed — a fixed, finite quantity of berries or insects genuinely running out. A trading signal decays for a different reason: not because the underlying market inefficiency is a fixed pool being consumed, but because other traders notice it and compete it away, a reflexive, strategic process with no direct ecological equivalent. A berry bush doesn’t get less rewarding because other foragers noticed you were there; a trading signal often does. That’s a meaningful difference in mechanism sitting underneath the mathematical similarity in structure.
There’s a genuinely underexploited opportunity buried in this gap, though. Behavioral ecology has already built out a substantial body of work extending the marginal value theorem to handle exactly the kind of uncertainty quantitative trading grapples with — foragers who don’t know the true average rate of the environment and must estimate it adaptively using Bayesian updating as they go, adjusting their patch-leaving threshold as new information arrives. That’s a close mathematical cousin of the problem a trader faces estimating an uncertain, decaying signal’s true remaining value in real time. Nothing in the published finance literature suggests anyone has gone looking at the ecological uncertainty-extension literature for tools that might transfer.
Why It Matters
The value here isn’t a plug-and-play formula finance could import wholesale — market dynamics, competitive crowding, and transaction cost structures are different enough from foraging ecology that a direct transplant would oversimplify. It’s the more interesting observation that two fields, working on superficially unrelated problems with no contact between them, converged on nearly the same abstract optimization logic because the underlying problem — when to abandon a depleting opportunity for a fresh one, weighed against the cost of switching — turns out to be a genuinely general one, showing up wherever an agent, biological or financial, has to decide how long is too long to stay somewhere.
Human Dimension
There’s something almost funny about picturing a bumblebee, working out by evolutionary trial and error exactly when a flower patch has stopped being worth the visit, and a quant on a trading desk, running a Markov decision process to work out exactly when a signal has stopped being worth the position, arriving at recognizably the same shape of answer. Neither one read the other’s work. Neither one needed to. The math that governs “when do I stop harvesting this and go find something better” turns out not to care much whether the thing being harvested is nectar or basis points.
Sources:
1. Wikipedia — “Marginal value theorem” — https://en.wikipedia.org/wiki/Marginal_value_theorem
2. PMC (National Institutes of Health) — “Foraging Under Uncertainty Follows the Marginal Value Theorem with Bayesian Updating of Environment Representations” — https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10996644/
3. bioRxiv — “Taking fear back into the Marginal Value Theorem: the risk-MVT and optimal boldness” — https://www.biorxiv.org/content/10.1101/2023.10.31.564970.full.pdf
4. PMC (National Institutes of Health) — “Foraging animals use dynamic Bayesian updating to model meta-uncertainty in environment representations” — https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12068741/
5. arXiv — “On the Effect of Alpha Decay and Transaction Costs on the Multi-period Optimal Trading Strategy” — https://arxiv.org/abs/2502.04284
6. Microalphas — “Signal Decay Analysis: Understanding Alpha Lifecycles” — https://microalphas.com/signal-decay-patterns/
7. Foxholm Financial — “Signal Decay and Alpha Erosion: Why Trading Signals Lose Their Edge” — https://foxholm.com/q/concepts/signal-decay/
8. National Bureau of Economic Research — “Dynamic Trading with Predictable Returns and Transaction Costs” — https://www.nber.org/system/files/working_papers/w15205/revisions/w15205.rev0.pdf
9. ScienceDirect — “Alpha decay and Sharpe ratio: Two measures of investor performance” — https://www.sciencedirect.com/science/article/abs/pii/S0264999321001474
Idea originated at artificialideas.org. Article researched and written by Claude Sonnet 5. Published at artificialideas.org.
Artificial Ideas 