[skybrian]

The prediction market itself is a ouija board. You're given a number. You don't know who's moving the needle or why. You don't know what you're paying for. Maybe you're paying for information from people who are breaking someone's trust by giving it to you? Or maybe you're paying them to make it happen?

Although, sometimes a market provides incentive to publish information that's associated with the market being influenced. For example, someone can do an investigation, short the stock, then publish it.

[testaccount28]

but like, here in the real world, farmers use weather derivatives. so the technology works, has a use case, is proven.

if your point is that one should not treat the market's number as some oracular probability, then... of course i agree! there is no such thing. the market provides a signal, like any other.

[skybrian]

Not a farmer, but I believe they use weather derivatives to hedge and weather forecasts for predictions? Going through markets for weather forecasts is adding a level of indirection that generates a noisier signal.

The idea when hedging isn't to win on expected value. It's to reduce risk. You're paying the market to provide insurance.

As a side effect, insurance does sometimes generate interesting data. The insurance industry generates good data about life expectancy. But it doesn't tell you when you're going to die.

[DennisP]

Good points, but commodity futures and stock prices take into account all sorts of information. They go haywire sometimes but given how hard it is to beat the market, they seem to do a pretty good job of aggregating it all.

[skybrian]

As we learn from reading Matt Levine, they might also be taking into account signals that are irrelevant to you for technical reasons, or just nonsense. Often it works well but sometimes you get meme stocks.

[DennisP]

Which is why I said "they go haywire sometimes." But in most cases, they work well enough that very few professional investors are able to consistently beat the judgement of the market.

[testaccount28]

just to be clear about what you're saying, you would predict that the implied probabilities from weather derivative prices are often very different than what is forecasted, and, when those differences occur, it is the forecast that is more likely to be right?

...would you like to make a bet to this effect?

[chii]

you dont need to pay to access the odds - it's public info.

There are people who pay to make bets on it (if they think the odds are wrong). But you don't have to be a betting participant to access the betting odds. You simply use the betting odds as a prediction of a future outcome, and you take your action/planning accordingly.

[BoorishBears]

But usually when prediction markets have shown interesting predictions, it's by odds taking large swings then collapsing to the correct outcome relatively shortly before the event right?

I assume because even if you know the future perfectly, putting up large lump sums early could cap your upside if people take your large sum as a signal (like OP is doing)

As a viewer you can take your own short-term "actions" (gambles) outside the market using the brief advanced notice I guess, but I'm not sure planning works like that.

In other words, what happens to the accuracy of prediction markets if we're including the discrete odds that occured along the way to the final odds? It's not better than random chance or public sentiment for large events is it?

[skybrian]

Sure, but I meant it in the sense that someone needs to lose money or there's no point in smarter or more well-informed people playing. Their profits have to come from somewhere.

These could be (a) people who aren't as smart as they think they are (b) people who subsidize the market in order to get good predictions (c) people who are hedging (essentially, buying insurance). Perhaps other possibilities.

[chii]

> the sense that someone needs to lose money

yep, and that's fine because they did so voluntarily.

If there were no stakes on the line, the information in the odds will also not have any real meaning.

[skybrian]

It’s good that it’s voluntary, but that’s not really what I’m getting at. People voluntarily spend money in Vegas and buy meme coins too.

This doesn’t tell us all that much about whether a price signal is a valuable source of information. Often, people have varied interpretations of what a price movement means. The price doesn’t tell you how to interpret it. The obvious interpretation can be wrong.