Manifold Markets is a prediction market platform where I've been trading since September. This post will compare it to other prediction markets that I've used.

Play Money

The most important fact about Manifold is that traders bet mana, which is for most purposes not real money. You can buy mana, and use mana to donate real money to charity. That's not attractive enough for most of us to treat it as anything other than play money.

Play money has the important advantage of not being subject to CFTC regulation or gambling laws. That enables a good deal of innovation that is stifled in real-money platforms that are open to US residents.

Play money misses out on an important advantage of large real-money markets: traders who want to manipulate a price face arbitrarily large costs to counter large investors (such as hedge funds) who notice the manipulation. So I trust Manifold prices less than I trust those of the best financial markets.

Actual real-money prediction markets are limited in size, and mostly too small to attract hedge funds. So Manifold is sometimes able to beat real-money markets.

Manifold is structured so that manipulation seems rare. Competent traders who have been active for a long time accumulate enough mana to dominate newer traders. So prices are mostly set by traders who have a significant track record of non-manipulative trading. That's not enough to make Manifold prices a good input for decisions that affect billions of dollars (how hard is it to bribe traders?), but it's good enough for a wide variety of less important decisions.

Liquidity

Another advantage of play money is that Manifold can create more mana whenever they want.

Manifold has taken advantage of that to enable traders to, in effect, buy on margin. In addition, they give daily handouts of mana to people who trade daily. Those two features ensure that I mostly don't end up in the situation that I got into on some other platforms of spending all my money and then waiting for markets to be resolved. My more regular trading on Manifold helps ensure that, at least for markets that get some minimal amount of attention, there are enough regular traders to keep the markets somewhat efficiently priced.

The most important way that Manifold provides liquidity is via an automated market maker. The normal trading mode is that I accept whatever price the system offers. Since Manifold uses play money, it costs them nothing to subsidize the market maker, thereby ensuring that markets have plenty of liquidity.

What's new about the market maker is that the amount of liquidity it provides increases as trading increases. We expect that the first trade on a market produces a less efficient price than the 100th trade. So we want the first few trades to move the price fairly easily. Whereas the hundredth trade is less likely to add new information about what the price should be, so the 100th trade should move the price less than would a similarly sized 2nd trade.

I advocated this kind of approach a decade ago, then forgot about it. I'm glad to see it being implemented well.

I traded for a few months before noticing a mention of limit orders. I had vaguely wanted to enter limit orders, but didn't see anywhere to enter them. (They're hiding under a button that says "%"). This reflects a UI that looks simple to beginners, but has enough power to maybe satisfy experienced traders.

50% Bias

Like many markets, prices on Manifold are too close to 50%.

E.g. as of January 30, Will NASDAQ TMC (The Metals Company) hit 2$ before 1st Febuary 2024? is trading at 10%. It has dropped about 55% in the past week, as the number of remaining trading days dropped from 9 to 2. I think it would ideally have dropped to something closer to 2/9 of last week's price. That would require more people to trade it on a daily basis, in order to overcome the price stickiness of the automated market maker. Attention is spread too thinly for that to happen.

A longer term contract with a similar problem is Will medicine master head transplants in the next 10 years?, currently trading at 7%. Betting NO here looks like a nearly guaranteed win, but it might be less than a 1% annual return on my initial investment (I need to put down 93 mana per 100 shares). That's not an impressive return compared to what I expect on shorter term contracts such as the one I mentioned above.

Why was the head transplant contract ever above 7%? It looks like the automated market maker is always initialized to 50%. One alternative that I've seen (in the original Good Judgment Project?) is to have that initial price set by an auction of traders, with the automated market maker only becoming active after the first trade. That makes the initial price more accurate. The downside is that there is much less incentive for anyone to make the first trade, so some (possibly many?) contracts won't start trading, and maybe traders will generally put less effort into searching for new contracts.

I selected these two examples because they showed the problem most clearly of markets I've been following recently. That's partly because they've hardly attracted any attention. Heavily traded markets show milder versions of this bias.

Addictiveness

Trading on Manifold is mildly addictive. Factors that contribute to this are:

• Having a potentially daily source of new mana.
• Having that mana supply depend on daily trading.
• Having prices that are inefficient enough that I can make frequent profits.
• Having a conspicuous and readable game-like score.
• Having a wide variety of markets, with an adequate UI for finding new markets that interest me.

Amateurs Resolving Markets

Most serious platforms try to ensure that all markets are resolved in accordance with an unambiguous interpretation of the wording of the contract. This usually involves someone who works for the platform acting like a lawyer when each market is first proposed, to rule out any important ambiguity in the wording. That person also acts as a judge when each market is due to be resolved. The overhead associated with this work likely limits the quantity of markets that can be created, while still leaving traders dissatisfied with some market resolutions.

Manifold rejects that approach: any user can create new markets, and resolve them when they come due, with approximately no requirement that market will be resolved fairly. The burden is on traders to check that market resolution criteria are clear, and that the creator is reputable enough to judge it fairly.

This policy generates some complaints. As a trader, I don't feel like I'm affected much by this choice. Other prediction market platforms never managed to handle this well enough that I was more eager to trade there due to a greater expectation of fairness.

One of my earliest memories of a prediction market is the Iowa Electronic Markets, where I lost all my investment ($500?) by betting that Bush would win the 2000 election. I apparently failed to read the fine print that said the market would be resolved based on who won the popular vote.

The difference between play money and real money makes me less worried about similar mistakes on Manifold.

The existence of a wide variety of markets on Manifold enables me to diversify more than I did on other platforms, which also reduces my concern over individual markets being resolved unfairly.

Concluding Thoughts

Manifold's approach of encouraging lots of markets with little thought about quality is working better than I expected it to work.

Manifold seems somewhat likely to become a respected source of predictions for a moderately important set of questions.

However, they don't seem to be close to being profitable. My intuition says they'll find a way to earn enough to sustain some sort of growth. I don't expect them to do well enough that I'll regret not investing in them.