Written for the 2025 LSESU Essay Competition by Neerav Soni

Armed conflict and peace are difficult to predict. They occur at a variety of scales and contexts.. Authoritarian states and democratic states alike are involved in armed conflict and peace. Conflicts erupt ranging from drug wars in Mexico to border conflicts in Kashmir and full scale war in Ukraine. There is a unique constellation of factors surrounding each event (or non-event) meaning there can be no one-size-fits-all metric for prediction. Therefore, this essay argues that the best economic predictor is one that adapts to the situations and factors presented by aggregating the various viewpoints and importance of metrics – the prediction market. 

I considered several other metrics including GDP/Capita, demographics (particularly youth bulges), inequality, and higher education levels before coming to prediction markets. However, the common issue with all of the metrics considered was that they could not be used alone to make a prediction and instead had to be considered alongside other metrics. For instance, Urdal’s (2006) study establishes that youth bulges have an effect on the level of political violence, however it is influenced by many factors. The most important of which include development levels (measured using GDP/Capita) and regime type. Societies with higher development levels and purer democratic and autocratic governing styles are less conflict prone with youth bulges. However, economic growth and tertiary education levels are also influencers of terrorism displaying how armed conflict and peace are caused by a variety of factors. Urdal (2006) also establishes that there is no threshold past which youth bulges will begin to cause armed conflict. The absence of a clear causal threshold makes prediction inherently uncertain. The main implication from these findings is that the degree of subjectivity used in predicting peace and armed conflict will increase as different perspectives may weigh metrics differently. This makes agreeing on predictions difficult unless some aggregation method, such as a prediction market, is used.

The reliability and availability of metrics are also a cause for concern, especially in the most conflict-prone areas which may be the regions where prediction is most vital. Statistics may be falsified for a variety of reasons by authorities or there may be unreliable data collection due to absence of funding. For example, during the Second Congo War, Zimbabwe would under-report it’s military spending to the IMF (Pearce, 2000). Increased military spending can have a direct effect on both development and economic growth and these affect the level of armed conflict.

A prediction market has 2 common forms: binary and index contracts. Binary contracts have 2 possible outcomes, and generally the contract payout is $1. The price (p) of the contract multiplied by 100 is the % probability of the event occurring, where 0 < p < 1 (at the boundary values the market will not form so the inequality is strict). For instance, if the price of a contract is $0.57 for Gavin Newsom to become president in 2028 then the market believes Newsom has a 57% chance of winning the 2028 Presidential Election. Index contracts are where there are a range of outcomes and the payout is proportional to the outcome. Using the same example, if the contract in a prediction market for the vote share % for Newsom in 2028 has a price of $0.36, this implies the market expects him to get 36% of the popular vote. This is because the contract will pay out $0.01 per % of vote share. This represents the market’s mean vote share expectation. Traders can buy or sell contracts. 

These contracts can be for predicting armed conflict and peace by assessing geopolitical situations. An example is a binary contract on Polymarket showing the probability of a ceasefire, 20% at the time of writing, between Hamas and Israel before August 31st (Polymarket, 2025a). The probability increases when looking at October 31st, currently at 48% ( Polymarket, 2025b). This is useful in determining when the market indicates a ceasefire to be more likely and shows how prediction markets can track expectations over time.

I believe that prediction markets solve many of the issues discussed previously, due to the Efficient Market Hypothesis (EMH). EMH states that in an efficient market, prices will reflect all available information (Fama, 1970). In prediction markets, EMH generally holds in its semi-strong form (which means all publicly available information is reflected). Due to EMH, prediction markets can aggregate different viewpoints which weight metrics differently and also compensate for doubts about reliability and accuracy of statistics. Traders in the market can buy or sell contracts at the market price depending on their beliefs of the outcome. Prices will shift depending on how many traders buy or sell contracts at the current price until the equilibrium price is reached with all current publicly known information and the viewpoints of all traders are aggregated. The reliability and accuracy of statistics will also be priced in due to traders using Bayesian reasoning (either formally or intuitively) to calculate probabilities of statistics being reliable/accurate. They will then use further Bayesian reasoning to update their beliefs on the outcome of the event and therefore buy or sell contracts, shifting market prices to reflect the new probabilities/mean expectations of the outcome. Traders apply Bayesian reasoning to all new noisy information, not just for statistical reliability and accuracy. EMH and Bayesian reasoning allow prediction markets to integrate various metrics as opposed to other predictors which are more limited.

In order for traders in a prediction market to make accurate predictions,  four requirements must be met. Traders should: have private information (diversity of opinion), even if it is just their own beliefs or intuition; not be influenced by the opinions of others (independence of opinion); use a mechanism to aggregate all private judgments (aggregation); and draw on specialised, local knowledge (decentralisation) (Surowiecki, 2004). Without these four requirements, the collective wisdom of the participants and therefore the accuracy of the prediction markets can be weakened.

Diversity of opinion can be limited by the specific platform and regulations of different countries. Kalshi, one of the largest prediction markets, is restricted to US residents (Osipovich, 2024). Its largest competitor, Polymarket is banned in regions including the US, UK and Singapore (CFTC, 2022; Sale, 2025; Arbusers.com, 2025). Regulatory blockades affect accuracy as increased international presence will increase traders’ diversity of opinion. This will be especially important if the market concerns an event happening in a restricted country, as this will severely restrict the accuracy of the market due to the absence of relatively more informed traders.

The aggregation requirement is carried out by public platforms such as Polymarket and Kalshi. However, while predicting armed conflict and peace for some situations with national security implications there is a scope for an closed government-run prediction market. An example of this is the short-lived FutureMAP by DARPA. This would ensure participants in the market have more accurate and reliable information as participation would be limited to those working in the defence departments (and therefore have clearance for information and context about the subject). This would also encourage staffers to offer their own true opinions as they will not have the fear of disagreeing with superiors publicly.These factors will offer more accurate predictions as the independence and diversity of opinion will be greater. Furthermore, with more accurate information, prices will be closer to the true probabilities/market mean expectations.

The US army released an article detailing the usefulness of prediction markets on public platforms for threat prevention and prediction (Ferris and Ferris, 2025). However, in order for public prediction markets to provide well calibrated probability forecasts, there must be a sufficient number of traders in the market providing liquidity. This can also be measured in the trading volume of a market. If the trading volume/market liquidity is low then prices may fail to reflect true probabilities as there are not enough traders to back the other side of a contract. This means prices adjust according to the market forces and are instead stuck in place. Therefore, public prediction markets should only be used for armed conflict prediction when the issue is extremely topical and there is a guarantee of sufficient market participation e.g Israel-Hamas Conflicts. For less well known debates, closed prediction markets may provide better probability forecasts despite having less traders in general.  Some markets should also be kept closed as if they were public, this might lead to a self-fulfilling prophecy of individuals making the outcome does occur. For example, an assassination market, where having contracts on when an individual will die may lead to their demise by parties who hold those contracts. There is also the ethical consideration of profiting on armed conflict which is why FutureMAP was short lived.

Prediction markets compensate for statistical unreliability and subjective interpretation without being dependent on one viewpoint. They combine the strengths of the predictors currently used while limiting their weaknesses. Their main advantage is their adaptability to any situation, which is why prediction markets are the best method for predicting armed conflict and peace in this multi-polar world order.