Trades tell you what happened. Depth tells you what can happen next. In one real Kalshi book we checked, the quote was 55¢ / 56¢. But a $5,000 buy would have paid 56.89¢ on average, 1.39¢ above mid. If you want to estimate slippage, size an order, or check if liquidity is real, you need the orderbook. Trades and hourly bars cannot answer those questions alone.

What did Dune launch?

On May 21, 2026, Dune launched unified Kalshi and Polymarket data. The dataset puts both venues in one schema. It includes per-fill trades, hourly OHLCV candles, hourly open interest, market metadata, and resolutions. Kalshi history goes back to June 2021. Polymarket history goes back to November 2022. The data refreshes about every hour.

This is useful work. A single schema across two venues saves a lot of cleanup. Hourly bars and trades are good for backtests, calibration studies, and volume reports.

We checked the public table list for Dune's prediction-markets catalog. We did not find an orderbook or depth table there. That is not a flaw. It is a different data type, with a different job.

What do trades and hourly bars tell you?

A trade is one fill. It has a price, a size, a side, and a time. An hourly bar sums those fills into open, high, low, close, VWAP, and volume.

This data answers questions about the past:

• What price did the market trade at?

• How much volume traded in an hour?

• How did the probability move before an event?

But a trade only shows liquidity that was used. It does not show liquidity that was waiting. It also does not show orders that were placed and then pulled.

What does orderbook depth tell you?

Depth is the list of resting bids and asks at each price level. It shows how much size sits at each price, at a given time.

This data answers questions about cost and risk:

• What is the spread right now?

• How much can I buy before the price moves 2 cents?

• What average price would a 5,000-contract order get?

• Did a large wall of orders vanish before it was hit?

A market can show a 60¢ last trade and still have almost nothing behind it. The last price looks stable. A real order would move it.

How do the two data types compare?

The short version: trades and bars show the outcome. Depth shows the conditions.

Text version: trades and bars answer what traded and how much. Depth answers spread, slippage, size per level, and pulled quotes. Both show probability moves.

What does a real book look like?

Here is one real snapshot. It is one market at one moment. It is not a typical figure.

• Market: Kalshi, "BTC price up in next 15 mins?" (ticker KXBTC15M-26OCT250915-15).

• Time: 2026-10-25 13:01:16 UTC (9:01 AM ET). This was one minute into a 15-minute window.

• Source: Kalshi's public orderbook API. This is not Bravado data.

Kalshi lists YES bids and NO bids. A NO bid at 44¢ is the same as a YES offer at 56¢. So the YES ask is 100¢ minus the NO bid. Dollar size is contracts times price.

The top of the book was tight. The best YES bid was 55¢. The best YES ask was 56¢. The spread was 1¢, and the mid was 55.5¢.

Only about $2,000 sat at the best price on each side.

Text version: asks were 3,710.08 contracts at 56¢, 2,841.20 at 57¢, 1,727.62 at 58¢, 2,976.00 at 59¢, and 1,832.00 at 60¢. Bids were 3,815.60 at 55¢, 1,476.82 at 54¢, 1,594.10 at 53¢, 2,406.10 at 52¢, and 2,317.11 at 51¢.

What would a large order cost?

We walked the ask side of this book. Fees are not included.

• $1,000 YES buy: about 1,785.71 contracts, all at 56¢. The average was 56.0¢. That is 0.5¢ (0.9%) above mid.

• $5,000 YES buy: about 8,788.82 contracts. The order fills at 56¢, 57¢, 58¢, and 59¢. The average was 56.89¢. That is 1.39¢ (2.51%) above mid.

The quoted price did not change. The cost did. A last-trade price would hide this gap.

Text version: 3,710.08 contracts at 56¢ ($2,077.64), 2,841.20 at 57¢ ($1,619.48), 1,727.62 at 58¢ ($1,002.02), and 509.92 at 59¢ ($300.85). Total 8,788.82 contracts for $5,000, average 56.89¢.

Text version: average price above the 55.5¢ mid was 0.50¢ for $1,000, 1.39¢ for $5,000, and 4.74¢ for $15,000.

Why does this matter for backtests?

Many backtests fill every order at the last trade price. That is a hidden assumption. It says the book had enough size at that price. In the snapshot above, it did not for a $5,000 order.

On a strategy with a small edge, a 1–2¢ gap can remove the profit. Trades alone cannot show this. You need the book as it stood at the time of the order. That is why historic depth matters as much as live depth.

Why does pulled liquidity matter?

Some orders exist to be seen, not to be filled. A large bid can appear, move other traders, and then vanish. In a trades table, this leaves no trace. Nothing traded.

With depth snapshots over time, you can see size appear and disappear. You can measure how often quotes fade near a fill. This helps you judge if a market's liquidity is real.

Why use Bravado?

Bravado is prediction market infrastructure. We provide licensed data across Kalshi, Polymarket, and Novig through one API.

What you get today:

• Historic orderbook depth. Backtest against the book that existed, not the last price. For Kalshi, this covers short captured windows.

• Trades. Fill-level history across venues.

• Live feeds. Our Live Trades WebSocket streams fills as they happen. For Kalshi, the live data beta has a public trades feed.

• UMA WebSocket. Track Polymarket resolution proposals and disputes in real time. See our UMA disputes WebSocket guide.

For Kalshi, our docs show what is live today:

• Native Kalshi market data. Current orderbooks, markets, public trades, and candles over REST.

• Kalshi current liquidity. Mid, spread, and depth within 1¢, 5¢, and 10¢ of the best price, from one full book.

• Captured Kalshi orderbook history. Stored books at a past time, for short captured windows. It is not full Kalshi history.

• Kalshi live data beta. Live books for a selected set of markets, plus feeds for public trades and market changes.

Start with the Kalshi overview. It lists what each product covers.

Trades and depth come from one API, so you can join them without venue-specific glue. For more on the data model, read our Polymarket data API guide and our Polymarket WebSocket guide. If you track performance, see how to calculate Polymarket PnL.

Bravado and tools like Dune can work together. Use hourly bars for broad research. Use depth when the question is about cost, size, or execution.

Conclusion

Pick the data type that matches your question.

• To study price history and volume, use trades and hourly bars.

• To estimate slippage, size, spread, or fill quality, use orderbook depth.

• To backtest a strategy you plan to run with real money, use historic depth.

A price is a promise only if there is size behind it. Depth is how you check.

Start at bravadotrade.com.