{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,28]],"date-time":"2026-09-28T08:57:50Z","timestamp":1790585870594,"version":"4.1.0"},"posted":{"date-parts":[[2026]]},"group-title":"SSRN","reference-count":0,"publisher":"Elsevier BV","license":[{"start":{"date-parts":[[2026,1,1]],"date-time":"2026-01-01T00:00:00Z","timestamp":1767225600000},"content-version":"unspecified","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"abstract":"<jats:p>Investors on the Malawi Stock Exchange face a frustrating problem: the standard tools for building a portfolio assume that the past is a reliable indicator, yet in a small market shaken by currency devaluations and policy shifts, yesterday's averages can mislead. This paper asks a simple question. What happens if, instead of feeding the classical Markowitz model long-run historical averages, we give it a forecast that pays attention to what the market has been doing lately? We use a neural network suited to time-series data (Long Short-Term Memory, or LSTM) to produce these forecasts, show mathematically that the combined model is sound and stable, and test it on eight years of monthly prices for all sixteen listed Malawian companies, keeping the final eighteen months completely untouched for an honest evaluation. The market showed substantial variation across stocks, with average monthly returns from 0.5% to 5.8% over the estimation period and several banking stocks repricing by more than 50% within a few months of the November 2023 kwacha devaluation. The results are encouraging: the prediction-based portfolio earned more than the classical one, a cumulative return of 275 % compared with 195 % over the hold-out period, with a better return for the risk taken and, notably, slightly less volatility. Even simpler forecasting rules improved on the classical model, which points to the real lesson. In markets like Malawi's, how you estimate expected returns matters more than the optimisation model itself. Everything, from data to results, can be replicated from our published code.<\/jats:p>","DOI":"10.2139\/ssrn.7520998","type":"posted-content","created":{"date-parts":[[2026,9,25]],"date-time":"2026-09-25T12:23:36Z","timestamp":1790339016000},"source":"Crossref","is-referenced-by-count":0,"title":["Beyond Historical Mean Returns: A Long Short-Term Memory Approach to Portfolio Optimisation on the Malawi Stock Exchange"],"prefix":"10.2139","author":[{"ORCID":"https:\/\/orcid.org\/0009-0003-1528-6822","authenticated-orcid":true,"given":"Amos","family":"Mngoli","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-7074-0608","authenticated-orcid":true,"given":"Nelson","family":"Dzupire","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-3958-1415","authenticated-orcid":true,"given":"Patrick","family":"Chidzalo","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"78","container-title":[],"original-title":[],"deposited":{"date-parts":[[2026,9,28]],"date-time":"2026-09-28T08:25:01Z","timestamp":1790583901000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.ssrn.com\/abstract=7520998"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026]]},"references-count":0,"URL":"https:\/\/doi.org\/10.2139\/ssrn.7520998","relation":{},"subject":[],"published":{"date-parts":[[2026]]},"subtype":"preprint"}}