Algorithmic share
bot_share_history.csv
Share of weekly Polymarket counterparty events (maker + taker) attributed to algorithmic wallets.
DV-PMI · Data downloads
Data downloads
Updated every Sunday 22:00 Pacific · 185 weeks · CC BY 4.0
Full weekly history for every index, plus the master wide-format panel. Each file is published as CSV with the complete history since November 2022. The master table merges all indices into a single date-indexed panel, analogous to the Baker-Wurgler sentiment file or Ken French’s factor library.
671Munderlying trades
185weeks of history
12downloadable files
Master wide-format panel
polymarket_indices_weekly.csv
All indices merged into a single date-indexed panel. Canonical citation file. Use this for time-series research, panel regressions, or as a Baker-Wurgler-style supplement.
Per-index histories
Each card shows the smoothed (13-week MA) sparkline. Click "Index page" to see the full deep-dive.
Probability weighting
pwi_history.csv
Weekly Prelec α from the calibration curve. α = 1 is rational pricing; α < 1 is the classical inverse-S weighting.
Execution edge
execution_history.csv
Bot-minus-retail Prelec α gap. Positive means bots weight tails more heavily — the behavioral signature of execution-quality divergence.
Market efficiency
efficiency_history.csv
How cleanly the Prelec model explains each week's calibration gap. Higher = stable behavioral signature; lower = idiosyncratic shocks.
Longshot pricing gap
price_gap_history.csv
Realized win rate of trades at prices < 10% minus the band midpoint. Positive = longshots underpriced.
Cumulative P&L
cumulative_pnl_history.csv
Cumulative resolved P&L per wallet type since 2022. Bots end at +$136M; active retail at −$82M.
Snapshots
Cumulative aggregations across the entire 671M-trade panel. No time-series component.
Calibration curve
snapshot
calibration_snapshot.csv
20 price bins × realized win rate × CI bounds × n_trades. Prelec α = 0.664, R² = 0.987.
Insider screen (PII)
snapshot
pii_snapshot.csv
Per-wallet flagged status, MNPI category breakdowns, multiple-testing survivors. The Paper-2 forensic dataset.
Top markets (live)
snapshot
top_markets_latest.json
Top 30 markets by USD volume with bot share, flagged-wallet count, execution gap, and live odds via the briefings.
Market microstructure
snapshot
market_microstructure_latest.json
Per-(market, token, wallet_type) volume, average price, and execution-gap calculations for the briefings cards.
Read the data
Copy-paste these snippets to load the master panel.
Python
import pandas as pd
url = "https://jdellavedova.com/data/polymarket_indices_weekly.csv"
dvpmi = pd.read_csv(url, parse_dates=["date"]).set_index("date")
dvpmi[["pwi_alpha", "bot_share", "alpha_gap"]].plot() R
library(tidyverse) url <- "https://jdellavedova.com/data/polymarket_indices_weekly.csv" dvpmi <- read_csv(url) dvpmi |> select(date, pwi_alpha, bot_share, alpha_gap) |> pivot_longer(-date) |> ggplot(aes(date, value, color = name)) + geom_line()
Stata
import delimited "https://jdellavedova.com/data/polymarket_indices_weekly.csv", /// varnames(1) clear gen date_d = date(date, "YMD") format date_d %td tsset date_d tsline pwi_alpha bot_share alpha_gap
Cite this dataset
@techreport{dellavedova2026dvpmi,
title = {Della Vedova Prediction Market Indices},
author = {Della Vedova, Joshua},
year = {2026},
version = {0.1.0},
institution = {University of San Diego, Knauss School of Business},
url = {https://jdellavedova.com},
note = {Weekly-updated dataset, 671M Polymarket trades since November 2022}
} Data released under Creative Commons Attribution 4.0 (CC BY 4.0). Please cite. Column schemas follow semantic versioning; breaking changes trigger a major-version bump with a one-release deprecation window.