Polymarket
Algorithmic vs HumanFeesUser LTV AnalysisTop TradersDiversification AnalysisMost Traded QuestionsTaker Speedbump Analysis
Methodology: Algorithmic vs Human Classification
Hypothesis: Humans need to sleep, so their trades cluster during waking hours. Algorithms trade 24/7, so they have a flat hourly distribution.
Method: We analyze the coefficient of variation (CV) of hourly trade counts for each user:
- Low CV (< 0.5) = flat distribution = algorithmic
- High CV (> 1.5) = concentrated distribution = human
- We also consider the number of active hours (algorithms trade in more hours) and total trade count
Classification: Users are classified as algorithmic, likely_algorithmic,likely_human, or human based on their hourly trading patterns.
Key Finding:
While algorithmic and likely algorithmic traders represent only 29.1% of all users, they account for 84.6% of total trading volume, demonstrating their outsized impact on platform activity.
Trading Volume (7-Day Average)
Capitola ResearchDaily Active Users (7-Day Average)
Capitola ResearchVolume per User
Capitola ResearchTrader Type Statistics
| Trader Type | Number of Users | Avg Lifetime Volume | Median Lifetime Volume | Avg Lifetime Trades | Median Lifetime Trades | % Active | % Active (7d) | Avg Trading Duration (days) |
|---|---|---|---|---|---|---|---|---|
| Algorithmic | 29,688 | $4,922,960 | $127,832 | 82,767 | 15,298 | 79.7% | 51.6% | 140 days |
| Likely Algorithmic | 524,097 | $200,048 | $32,668 | 2,029 | 1,021 | 72.3% | 35.6% | 162 days |
| Likely Human | 211,515 | $83,585 | $22,900 | 662 | 597 | 61.5% | 29.9% | 194 days |
| Human | 1,135,984 | $24,735 | $9,170 | 145 | 321 | 50.7% | 22.8% | 59 days |