agentsclimarketplace

Behavioral economics

Skill alzadjaliaafra-hash/murshidi-knowledge-layer/models/behavioral-economics

Modular, fine-tuning-ready knowledge architecture for financial-domain LLMs — 9 domain models, each an activatable skill with knowledge corpus, instruction dataset, and held-out evals.

Install
npx -y skills add alzadjaliaafra-hash/murshidi-knowledge-layer --skill behavioral-economics

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • 27 days oldThe repository was created 27 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Behavioral economics framework — the twelve cognitive biases that drive real financial decisions (loss aversion, anchoring, availability, confirmation, sunk cost, overconfidence, herding, framing, recency, endowment, hyperbolic discounting, illusion of control), with application patterns for advisory communication, incentive design, and platform UX. Activate with /behav, "behavioral", "bias", "client psychology", or "decision design".

SKILL.md

3.8 KB, 666 tokens by cl100k_base, as published. Nobody here has run it

Behavioral Economics in Finance

A working model of how humans actually make financial decisions — not how they should. Used to predict counterparty behaviour, structure incentives, frame advisory communication, and design platforms that align with human nature rather than fight it.

The Twelve Biases

#BiasCore principleFinancial manifestation
1Loss aversionLosses hurt ~2× as much as equivalent gainsHolding losers too long; delaying profit-taking
2AnchoringThe first number becomes the reference pointOpening offers frame entire negotiations
3Availability heuristicRecent/memorable events are overweightedPost-crash over-caution; trend-chasing
4Confirmation biasPeople seek confirming informationWarning signs on existing holdings ignored
5Sunk cost fallacyPast investment drives continued commitmentUnderwater positions held; failed projects continued
6OverconfidenceKnowledge and predictive skill overestimatedOvertrading; concentrated bets
7HerdingSafety is sought in crowd behaviourMomentum bubbles; capitulation selling
8Framing effectPresentation changes the decisionIdentical economics accepted or rejected by wording
9Recency biasLatest data dominates the long seriesShort-window extrapolation of returns
10Endowment effectOwned assets are overvaluedRefusal to sell at fair market prices
11Hyperbolic discountingNear rewards dominate far larger onesUnder-saving; short-term deal preference
12Illusion of controlInfluence over random outcomes is overestimatedExcessive active management; timing conviction

Application Patterns

Advisory communication. Lead with loss framing where action is needed ("protect what you have earned" outperforms "capture more upside"). Control the anchor — always present the first number. Introduce counter-evidence only after acknowledging the client's existing thesis (confirmation-bias-aware sequencing). Replace point estimates with probability ranges to calibrate overconfidence. Separate past from future explicitly to defuse sunk cost ("the only question is forward return from today").

Incentive & platform design. Make forward-looking metrics the visual default (against sunk cost and recency). Use "typical range" displays to anchor expectations honestly. Design exits that feel like securing gains rather than admitting losses. Counter herding with base-rate displays; counter availability with long-window context charts. Structure commitment devices against hyperbolic discounting (auto-escalation, default enrolment).

Ethical boundary. These mechanics are applied to align decisions with the decision-maker's own stated objectives — de-biasing, not manipulation. Any application that exploits a bias against the client's interest is out of scope.

Output Protocol

Every behavioral analysis names the operative bias, the observable evidence for it, the reframe that neutralises it, and the design or communication change that operationalises the reframe.

Resources

  • knowledge/methodology.md — full bias catalogue with application detail
  • dataset/train.jsonl — instruction-tuning pairs
  • eval/eval.jsonl — held-out evaluation questions

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.