The most liquid board on the Prediction Alpha trending page ranks ten markets by depth, from $4.7M at the top down to $3.8M at the bottom. Read the titles down the column and every single one of them says the same thing, which is the 2028 Democratic presidential nominee.
Ten rows. One question. The board is not broken and the numbers are not wrong, but if you read that column the way you would read a list of ten different markets, you have just formed three separate false impressions about where the depth is, how much of it there is, and how many distinct things you could trade.
A nominee market is a stack of separate books
Anything with more than two possible outcomes has to be expressed on a binary venue as a set of individual contracts. There is a contract on candidate A becoming the nominee, another on candidate B, another on candidate C, and each of them is its own Yes and No pair with its own order book, its own price, its own spread and its own depth figure.
The question is the container. The candidate is the instrument. When a leaderboard ranks instruments by liquidity and a single container holds forty of them, the container takes the whole board. Nothing about that is a bug, it is what ranking contracts means, and it becomes a problem only at the moment you read the board as a list of ideas rather than a list of order books.
The board understates the depth and overstates the variety
Both errors run at once, in opposite directions, which is what makes this worth writing down.
Add the ten visible legs together and you get roughly $40M sitting behind that one question, against a headline tile of $4.7M. So the board understates by close to a factor of nine how much depth is actually attached to the nominee question, and there is no reason to think the legs stop at ten. Any candidate outside the top ten by liquidity is off the bottom of the board entirely.
At the same time it overstates variety by exactly the same factor. If you were skimming that board for markets to look at this week, you found one. The other nine slots that could have shown you a different question, a different category or a different venue are consumed by candidates in the same primary.
Notice also that all ten rows carry the same platform badge, and that the volume board on the left carries a different one on all ten of its rows. Neither board is a fair survey of the six venues in the filter above them. Each is showing you whichever venue happens to win on that particular metric.
Collapse to the question before you compare anything
The routine takes a minute and it is worth building the habit, because every venue comparison you make is wrong without it.
- Read down the titles and group anything sharing a stem. Multi-outcome questions almost always repeat the container name, which is what makes them easy to spot once you are looking.
- Use the search box on the markets tab to pull every leg of that question, not just the ones that made the leaderboard, because the tail matters for the total.
- Sum the liquidity across the legs. That sum is the question's depth and it is the number you compare against another venue's version of the same question.
- Then compare against the other venue's total, not against its top leg. Comparing your summed total to their single best row is the mistake that makes one venue look ten times deeper than it is.
There is a second check worth running while you have all the legs on screen, though I would treat it as a data quality test rather than a trade. Add up the Yes prices across every listed outcome. In a well behaved market that total sits a little above one hundred, with the excess being the venue's overround. If it sits well below one hundred, you are almost certainly missing legs, or the question has an unlisted outcome covering any candidate not named. Before you conclude you have found something, remember that acting on it would mean buying dozens of separate legs, each with its own spread and fee, and holding all of them until 2028.
Ten positions in one question is one position
The part that costs money is not the leaderboard reading, it is what people do next. Buying three candidates in the same primary feels like diversification. It is not. Exactly one of those contracts can pay, so the legs are not independent bets that offset each other, they are three slices of a single event whose outcome is determined on one day by one process.
The consequence for sizing is straightforward. Add the cost of every leg you hold in a question and treat that total as one position against your risk limit, not as three positions of a third the size. If you would not put four hundred dollars on the nominee question as a single bet, do not put four hundred dollars across four candidates and call it a spread.
The same applies to how you monitor them. Saving every leg to a watchlist with price alerts means one piece of news generates a wall of notifications about a single event, and after a week of that you will start ignoring the alerts, which is a worse outcome than not having set them. Track the two or three legs where you actually hold a position and let the rest go.
Where this shows up when it is not obvious
The nominee board makes the pattern easy to see because the titles are identical. Most of the time they are not, and the same structure is hiding behind wording that looks different enough to fool a quick read.
The volume board on the left of the same page has it too. Three of its ten rows are Natus Vincere against Fnatic, once for Game 1, once for Game 2 and once for the best of three. Those are not identical titles and they are not independent contracts either, since the match result and the individual game results are tied together by the same fixture. Sports series, tournament brackets, rate decisions at consecutive meetings and staged regulatory approvals all produce the same shape.
So the habit to build is not really about leaderboards. It is to ask, for every contract you are about to buy, what single real world event determines it, and then to check what else in your portfolio is determined by that same event. On a multi-outcome question the answer is sitting in the title. On everything else you have to go and look, and the leaderboard that put ten candidates in ten slots is a useful reminder that the screen will not do that work for you.