Brand Stories Quantifying Skill in Opinion Trading Platforms: Key Findings from IIT Delhi Research Quantifying Skill in Opinion Trading Platforms: Key Findings from IIT Delhi Research
IIT Delhi's research confirms that opinion trading is a skill-driven market. By analyzing player performance and establishing criteria like skill predominance and consistency, the study advocates for its classification as a legitimate skill game, comparable to fantasy sports and poker.
Opinion trading is a form of gamified information market where participants trade on the outcomes of future events, from sports matches to elections. In these platforms, users buy and sell virtual shares or contracts that pay out based on an event's outcome. For example, a contract might correspond to a prediction that a certain team will win a match; if the prediction is correct, those holding the contract profit, otherwise they lose their stake.
Under law, the “predominance” test often determines a game's status: if skill predominates over chance in influencing the outcome, the game is considered skill-based.
Researchers at IIT Delhi have addressed this question in a March 2025 study titled “Quantifying Skill on Opinion Trading Platforms.” The study examines opinion trading through the lens of four criteria commonly used to establish a game as skill-based: (1) Predominance of skill over chance, (2) Consistency of player performance, (3) Learning effect over time, and (4) Existence of a skill gradient among players. Below, we summarize how the researchers theoretically and empirically validated each of these criteria using mathematical models and real-world data, and what their findings imply for the legal and societal understanding of opinion trading.
Predominance of Skill Over Chance
One fundamental test for a skill-based game is whether skilled players reliably outperform what pure chance would yield. In the context of opinion trading, this criterion asks: do informed, strategic traders earn better returns than if they were simply guessing outcomes at random?
Theoretical Model: The IIT Delhi team developed a rigorous mathematical model of an opinion trading market to compare a random strategy against a skilled strategy. In their model, a random player makes arbitrary trades (buying outcomes at random prices) with no knowledge of the event, whereas a skilled player possesses superior knowledge or insight about the true outcome
The analysis showed that a purely random trader is very likely to lose money over time. In fact, when all players trade randomly, the expected return of a typical player ends up lower than the amount invested once platform fees are accounted for. This means that without skill, a trader's ROI (return on investment) tends to fall below 1 (a net loss). By contrast, a trader with even a modest edge—modeled as knowledge of the eventual outcome—can exploit that information to earn positive returns. The researchers prove that a player who knows the outcome and acts optimally can achieve an expected return significantly above break-even, even when facing many random-playing opponents. In short, skill (in the form of better predictions) was shown to confer a clear theoretical advantage, whereas a lack of skill leads to eventual losses. This mathematical result establishes, under controlled assumptions, that skill can indeed predominate over chance in opinion trading.
Empirical “Skill Dilution” Test: To complement the theory, the study introduced a novel skill dilution experiment using real market data. The idea is simple: if a game is purely chance, randomly altering some outcomes post hoc shouldn't systematically hurt performance. If the game is skill-based, however, injecting randomness should degrade skilled players' success. Researchers took historical event data from an opinion trading platform and “diluted” the outcomes with varying probabilities
In practice, for a given dilution level α (0 ≤ α ≤ 1), they would flip a fraction α of the event results to the opposite outcome (as if some outcomes were decided by a coin toss). They then recalculated all players' win rates and returns under this counterfactual scenario. The results were striking: even a small amount of outcome randomization caused a significant drop in overall win rates, and higher dilution levels led to progressively worse performance. In other words, when the “skill signal” in the outcomes was partially replaced with noise, players (on average) could not maintain their usual success rates. The entire distribution of win rates shifted downward under randomized outcomes, indicating that what was originally driving those win rates was skill, not luck. Statistical tests confirmed the robustness of this effect: the study reports that for each tested dilution level, the null hypothesis that performance metrics remain unchanged (i.e. unaffected by random outcome flips) was decisively rejected (with p-values on the order of 10^−100). The distribution comparison is provided here for reference.
This skill dilution test provides strong empirical evidence that skill is the dominant factor in real-world opinion trading outcomes. If success were mostly due to chance, random outcome perturbations would have little impact; instead, performance plummeted with increasing randomness, which underscores that skill makes the crucial difference in this market.
Consistency of Performance
Another hallmark of skill-based games is consistent performance by the same players over time. Simply put, a skilled player tends to perform well repeatedly, whereas luck-based outcomes would cause performance to fluctuate randomly from one period to the next. The researchers examined whether opinion traders who did well in the past continue to do well going forward, as expected if skill persists.
Using a large dataset of users' trading records across the entire year 2024, the study tracked two key success metrics for each player on a month-to-month basis:
- Win Rate (WR): the fraction of trades or event contracts that yielded a positive return for the player.
- Return on Investment (ROI): the ratio of money earned to money invested by the player.
If opinion trading is skill-based, one would expect a player's January performance to correlate positively with their February performance, and so on throughout the year. Indeed, the analysis found strong positive correlations between consecutive months' performance metrics for individual players
For every adjacent month pair (Jan–Feb, Feb–Mar, … Nov–Dec), the Spearman correlation of users' win rates was significantly above zero (typically in the 0.5–0.6 range), and similarly for ROI. All these correlations were highly statistically significant (the probability of seeing such consistency by chance was effectively zero).
To visualize consistency over longer periods, the researchers plotted a heatmap of correlation coefficients for every pair of months in 2024, not just consecutive ones. The resulting map revealed a clear pattern: players' win rates remained positively correlated even across many months of separation. For example, the correlation between win rates in January 2024 and December 2024 was about 0.59, indicating that many of the top performers in January were still outperforming by year-end. Overall, the correlations between any two distinct months ranged from roughly 0.52 to 0.65, a remarkably high level of persistence for a large user population. Such figures suggest that whatever factors made certain users successful in one month (e.g. superior insight, strategy, or knowledge) tended to remain with them in subsequent months, as one would expect if those factors are rooted in skill.
Heatmap of win-rate correlation among months (Jan–Dec 2024). Each cell shows the correlation between the win rates of all users in the two respective months. Warmer colors (toward orange) indicate higher correlation. The consistently moderate-high values (mostly 0.52–0.65 off the main diagonal) demonstrate that user performance in opinion trading is persistent over time. A slight dip in correlations during Mar–May is observed, coinciding with a major cricket tournament that attracted many new, inexperienced users (introducing more variability).
It is worth noting the March–April–May dip visible in the heatmap and analysis. During those months, a large influx of new users entered the platform, drawn by a major event (the Indian Premier League cricket season). These newcomers, being beginners, injected more random variance into the skill pool – many likely performed unpredictably – which temporarily lowered the correlation between early-season and late-season performance. Even so, the correlations remained strongly positive, and once the sample stabilized, consistency rose again. The overall takeaway is that opinion trading shows persistent performance patterns: skilled traders tend to remain ahead, month after month. Such consistency is a signature of skill-based activities (much as a top chess or poker player will usually stay on top over many games, barring only short-term upsets). It stands in contrast to pure games of chance, where last month's winners have no better chance of winning again this month.
Learning Effect: Do Players Get Better Over Time?
In games of skill, we typically observe that players improve with practice – newcomers start at lower proficiency and, through experience, increase their skill and success rates. The presence of a learning curve in the aggregate player base would further indicate that opinion trading rewards knowledge and practice, not random luck. The IIT Delhi study looked for this learning effect by tracking how individual players' performance changed as they played more and more event contracts.
For this analysis, researchers assembled a cohort of 37,000 new users and followed each user's first 720 trades on the platform
By looking at performance as a function of the number of trades completed (the “event rank”), they could assess whether more experienced traders achieved better outcomes. Two metrics were computed for each user after i trades (for i = 1 to 720):
- Cumulative ROI after i events, and
- Win rate (proportion of winning trades) after i events.
Aggregating these across all users, the study found a clear upward trend: both the average and median win rates of users increased as they played more events
In fact, the correlation between “number of events played” and performance was significantly positive, especially during the initial phase of usage. Statistical tests confirmed that the improvements were not due to chance – there was a strong positive correlation between experience and success with p ≈ 10^−100, essentially proving a genuine learning effect. The learning curve for opinion traders was visualized by plotting the median win rate against the event count (from 1 to 720) for the cohort. The resulting curve showed a steep rise early on, which gradually tapered off as users became more seasoned.
Interestingly, the shape of the curve follows a power-law form, which is commonly seen in skill acquisition for various games In practical terms, this means newcomers tend to make rapid gains in skill initially (as they learn basic strategies and avoid rookie mistakes), but the improvements slow down later, approaching an asymptotic performance level once advanced strategies are mastered. The researchers noted that this “plateauing” of learning is a well-known phenomenon in games like chess or video games, where novices climb quickly at first and then improvements become incremental. Another fascinating insight is that top performers learned faster. When users were grouped by their ultimate performance level (top 1%, top 10%, top 25%, etc.), the best players not only reached higher peak win rates but also had a steeper initial learning curve. In other words, those who eventually became experts showed stronger improvement early on than those who remained mid-tier – a sign that innate ability or early adoption of winning strategies accelerated their growth. However, all groups eventually leveled off, indicating that even the best traders hit a skill ceiling where additional experience yields diminishing returns. This pattern again mirrors other competitive skill-based domains, where high performers usually exploit learning opportunities more efficiently than others.
Overall, the presence of a learning effect in opinion trading – evidenced by rising win rates with experience and a characteristic learning curve – reinforces that success on these platforms is not random. Participants can actively improve their outcomes through practice, education, and strategy refinement. Such improvement would be impossible if outcomes were governed by luck alone. Instead, like fantasy sports or poker, opinion trading appears to reward those who study and gain expertise in the activity.
Skill Gradient and the OpTraS Score System
The final criterion the study examined is the existence of a skill gradient in the player population. A skill-based game usually has a wide distribution of ability: some players are very skilled, some average, some unskilled – and their results should differ accordingly. For instance, in chess or fantasy sports, one can rank players by skill level and observe that higher-ranked players win more often. The researchers sought to demonstrate a similar gradient in opinion trading by constructing an objective skill ranking system for traders and seeing if it correlates with success.
To do this, the study introduced OpTraS – the Opinion Trading Score, a composite rating formula designed to capture each trader's skill level. OpTraS was inspired by rating systems in games like chess (Elo ratings) and incorporates multiple facets of performance. Specifically, the OpTraS score rewards a player for:
- Success: How much net profit the player has made (total earnings) and their return on investment (efficiency of profit)
ROI ensures that simply spending more is not rewarded unless it's profitable. - Engagement: The level of activity, measured by the number of event contracts the player has participated in
This reflects experience and commitment to the platform. - Adroitness: Strategic acumen, quantified by the number of exit trades the player executes
Exit trades refer to selling off one's position before an event outcome is decided – much like cutting losses or taking profits early. A higher number of timely exits can indicate tactical skill in managing risk and reading market signals.
All users who had made at least a minimum number of trades (to have a performance history) were assigned an OpTraS score by the researchers. The score was computed with a formula that balances the above components, with recent performance weighted more (using a decay factor so that current form matters more than older results)
The details of the formula are in the paper, but conceptually OpTraS acts as an analog of a player rating – higher OpTraS should correspond to a more skilled (and successful) trader.
To validate the skill gradient, the research team divided players into groups based on their OpTraS ratings and compared their performance. If the score truly measures skill, then higher-rated groups should consistently outperform lower-rated ones. This is exactly what the study found. For example, one analysis compared “consistent” vs “inconsistent” players: those who engaged with opinion trading regularly (e.g. 20+ trades every week) versus those who participated sporadically.
The consistent-engagement group tended to accumulate higher OpTraS scores over time (not surprisingly, since more engagement and likely more learning), and when tracking their outcomes over nine months, they outperformed the inconsistent group in every period Figure 6 of the paper shows the skill score trends of these groups: the active, presumably more skilled players maintained a substantial lead in success metrics month after month. This persistent outperformance of one group over another is clear evidence of distinct skill levels – a high-skill cohort and a low-skill cohort with a measurable gap between them. In short, there is a skill gradient: some players consistently achieve higher returns and win rates, and a ranking system like OpTraS can identify them.
Crucially, the authors have made the OpTraS scoring system open and available for other platforms or researchers to use
By demonstrating a working skill metric for opinion trading, they provide regulators and industry stakeholders a tool to quantify player skill on an ongoing basis. Just as Elo ratings revolutionized how we rank chess players, OpTraS (or similar indices) could be used to maintain leaderboards, run “pro vs amateur” analyses, or even as a basis for compliance (e.g. ensuring games remain fair by detecting abnormally skilled bots, etc.). For the purpose of this study, OpTraS enabled a robust conclusion: opinion trading isn't a one-level playing field of pure luck – it has novices and experts, with experts consistently doing better, characteristic of a skill-driven domain.
Regulatory and Societal Implications
Establishing that opinion trading satisfies all four skill criteria – predominance of skill, consistency, learning by doing, and skill differentials – has important implications for how these platforms are viewed by law and society. Opinion trading is indeed a game of skill, it could justifiably be placed in the same category as fantasy sports, poker, rummy, and other skill-based games.
Legal Perspective: The findings of this research lend credence to the argument that opinion trading should not be lumped under non-skilled statutes, but rather treated as a skill-centric activity akin to fantasy sports. In the fantasy sports arena, both U.S. and Indian courts have noted that success requires substantial knowledge of players and statistics, strategic lineup selection, and informed decision-making.
The parallels to opinion trading are clear: instead of sports rosters, opinion traders analyze real-world events, weigh probabilities, and make strategic trades. The demonstrated predominance of skill means regulators could apply the same “dominant factor” test outcome as with fantasy cricket or football – concluding that while chance plays a role (as it does in any marketplace), skill is the overriding factor determining success. Moreover, the existence of a measurable skill gradient (via something like OpTraS) provides a quantitative basis to differentiate skilled play from random play, which could be a useful evidence in legal settings.
In jurisdictions like India, where the Public Gambling Act (1867) prohibits games of chance but exempts games of skill, this study's evidence can guide policymakers. We've seen state courts in India uphold the legality of online rummy and fantasy sports precisely because of their skill componentsf. With opinion trading now shown to share those same components, a strong case can be made that it too deserves the status of a legitimate skill-based venture. This could pave the way for clearer regulations that treat opinion trading platforms similarly to fantasy sports platforms.
Parallels with Other Games: The evolution of public policy around poker, rummy, and fantasy sports offers a roadmap for opinion trading. Poker, once seen purely as gambling, has increasingly been recognized for the skill of its top players (with even some U.S. case law supporting that view)
Rummy was explicitly declared a skill game by India's highest court decades ago, which allowed a whole industry of rummy and later other card game platforms to flourish legally. Fantasy sports faced legal challenges in both the U.S. and India; however, judicial opinions (such as White v. Cuomo in New York, 2022, and Varun Gumber v. Chandigarh in India, 2017) highlighted the skillful nature of managing a fantasy team, leading to their acceptance. The IIT Delhi study on opinion trading provides that compelling evidence for this new genre. By rigorously analyzing data and drawing analogies to established skill games, it effectively argues that opinion trading is no more a game of chance than fantasy sports or poker.
Societal Perspective: Classifying opinion trading as a skill game also invites a broader discussion on its potential benefits and risks to society. The researchers of the study suggest a positive view: opinion trading, by engaging users in analysis and critical thinking about real-world events, could serve an educational or cognitive function. The research finds that opinion traders become more informed about current events, data trends, and probabilistic reasoning. The paper even speculates that in the long run, such platforms might contribute to a more informed and engaged citizenry, as players have a tangible incentive to educate themselves on topics ranging from politics to economics While this is a hopeful outlook, it aligns with arguments made by economists (like Arrow et al.) that prediction markets harness the “wisdom of crowds” and improve information dissemination.
In conclusion, the March 2025 IIT Delhi study makes a compelling, data-driven case that opinion trading is predominantly a game of skill. It joins a growing body of research that seeks to quantify skill in interactive platforms, echoing the journey of fantasy sports and poker from dubious status to mainstream acceptance. By meeting the key tests of skill – outperforming random chance, showing consistent and improvable performance, and exhibiting clear skill stratification – opinion trading platforms seek a rightful place as legitimate market-based games. For policymakers, legal experts, and investors, these findings provide valuable clarity: they can approach opinion trading as an emerging skill-based industry. As with any innovation at the intersection of technology, economics, and law, ongoing dialogue is essential. This research is an invitation to that dialogue, offering evidence that could inform balanced regulations and encourage further study.
Disclaimer: This article is sponsored content curated by HT Syndication. The inputs and details accounted for in the article do not necessarily reflect those of HT, and HT does not endorse or assume any responsibility for the information provided.
Want to get your story featured as above? click here!
Catch all the Latest Tech News, Mobile News, Laptop News, Gaming news, Wearables News , How To News, also keep up with us on Whatsapp channel,Twitter, Facebook, Google News, and Instagram. For our latest videos, subscribe to our YouTube channel.
Follow our Page
Follow our Page
Follow our Channel
