Polymarket Market Metrics: Depth, Spread, Volume, and Open Interest Per Frame

• Per-frame: volume (cumulative, at 250ms), open interest, top-of-book bid/ask depth, mid, and the spread derived from the ladder.
• The same row shape covers every market type — Up/Down, sports, event-linked — so one metric query serves the whole record.
• Open interest exists at the frame level because the archive snapshots it; a 4h-sampled OI is an approximation of the same series.
• Spread is derivable (ask minus bid) but the archive carries the ladder levels that make the spread's composition visible, not just its width.

Volume is the dominator of the ladder: the September sample's median trade of 42,254 shares (p90 65,081) is the volume unit that moves these books, and depth at the top is often smaller than one median trade.

Open interest is the slow variable — it builds through a market's life and collapses at settlement — while volume is the fast variable that spikes in a reprice window.

Liquidity is where these two meet: a 90%+ one-sided frame count means OI is real, resting depth is not, and any liquidity metric must state which side it measures.

• Screen the archive: across hundreds of series, pick the markets where low liquidity (thin depth) meets high pending volume — the repricing candidates.
• Benchmark market quality: 98.6% of sampled frames price at 0.99/0.995, so a metric of "tick-at-the-round" per market is a direct book-quality measure.
• Detect regime: volume at rest vs at repricing is a two-mode series; the archive's timestamps let you classify frames by the reference move instead of eyeballing charts.

• Pull the metrics series for five markets of different types and print one day per market: volume, open interest, top-of-book depth, derived spread, at 250ms.
• Compute the one-sided ratio per market: share of frames where exactly one side carries size; expect the high-90s percentile most days from the measured behavior.
• Rank the five by pending volume per unit of depth — the screening metric that flags repricing candidates before the reference moves.
• Save the daily aggregate to a small table; your dashboard and your sanity share the same numbers.

The metrics series is the museum of the book: volume is what traded, open interest is what is still at stake, depth is what could trade now, and spread is the price of immediacy.

On Polymarket these four disagree constantly — a market with huge pending volume and a thin single-side book is common (90%+ one-sided frames), and the disagreemen is exactly where both signal and risk live.

The archive's design point is that all four share one timestamp; any cross-metric study is a single frame join, not a reconciliation project.

In a dashboard, the metrics series is the feed that keeps every panel honest: volume drift and OI collapse before settlement are both visible as regime changes in the same columns.

In research, the metrics series is the screening layer before the books layer; pick candidates from metrics, then study them in full depth.

Metrics are the screening layer of the whole stack: volume, open interest, top-of-book depth, and spread at one timestamp describe a market's state faster than any other page, and every repricing story this archive records started as a metrics anomaly.

Use the four-number signature (volume, OI, both depth sides, spread) to rank markets before opening the books — the pending-volume-versus-depth screen producing the repricing candidates is the practical payoff.

The one-sided ratio is your liquidity conscience: a market that is one-sided 90%+ of the time is a candidate, not a pool; the metric series is what says so in a number instead of a vibe.

Store the daily metrics aggregate as your standing "market scanner" table; it is the cheapest way to watch every series at once and the honest sibling to the deep individual studies.

• volume explained
• open interest explained
• depth explained