YES if the Dems win with a candidate other than Biden
NA if they do not switch candidates
NO if the Dems lose with a candidate other than Biden
1,000
3.00Fascinating that this market drops only after it looks like it won't resolve N/A.
This looks like strong evidence that N/A resolution criteria prejudicially excludes certain groups of people from betting, which seems obvious in hindsight, but isn't something I've heard anyone discuss as being potentially problematic for accuracy of conditional markets.
I wouldn't say there's strong evidence for that. There has been a lot of new information in the last day (the entire party rallying around Harris). Manifold thinks most of the other potential candidates would have had a higher chance to win, so not having them compete for the nomination would naturally drop this market.
Fair enough, and given that the market has rebounded since my original comment it's also entirely possible that the increase in traders was simply enough to increase market volatility.
I'd still be interested in some kind of analysis on whether there is such an effect in general though. It's not something I've seen anyone investigate, and it's piqued my curiosity. So if anyone knows of any sources or studies on the topic, I'd appreciate any reference.
(To somewhat formalize my curiosity: Assume a conditional market "If A occurs, then will B occur?" Assume that there is a positive correlation between believing A will occur and believing B will occur. Will such a market have higher odds than it accurately should have? In other words, would the occurrence of A naturally lower the market's predicted odds of B occurring?
If anyone can think of a good way to controllably test this via prediction market, then feel free. Otherwise I suspect I'll be trying to come up with a good market to test it myself.)
Two parameters that I expect to be relevant here are the size of the market and whether you can actually make money on it. As a person who has only put $25 into Manifold for the sole purpose of betting that the debate would result in Biden stepping down, I would have put a lot more money in if I could turn it back into cash. I was quite confident. But since I only have this much money to play with, I don't want to spend it on counterfactuals that won't resolve for months, since even if I'm right it doesn't pay the bills. Fix those two things and I would bet that much of this effect would go away.
If it's more like a real investment, parking my money somewhere where I might get zero returns or really far out returns would be a big disincentive to put money in some markets. E.g. there is a market for China succeeding when invading Taiwan. I've read a lot on the topic and would put the probability at 90%+. It's currently below 50%. The market has no close date. So it will likely stay open for years, if not decades. So I am very much limiting how much mana I put in there. If this was real money I can withdraw, I'd probably be better off putting my money in an index fund than keeping it bound up in this market for a very long time with the current probabilities.
Yes, time discounting is a very common problem in markets, but that's not what I'm talking about. Such effects affect all long-running markets. I'm interested in specifically conditional markets.
Presumably people would be less likely to invest in markets if they believe the market will most likely resolve N/A. The question is whether this can lead to prejudicial effects on the market if those who believe the market will likely resolve N/A are also more likely than average to lean one way or the other in predicting the actual resolution criteria. Note that I'm talking merely about correlation of beliefs here; there need not be any correlation between the actual two events themselves (although there may be, which would not need to be controlled for in any test).
The Democrats already revealed one candidate during the primaries (Dean Phillips) so i think if they switch now it goes from a 1 in 3 chance to a 1 in 2 chance
