In political campaigns, momentum also exists as an underlying phenomenon reflecting the gradual shift in public opinion. If a candidate picks up significant new support, it tends to take time for word of mouth, media to work. Polls definitely exhibit momentum. I'm not surprised there's a bit of that in prediction markets! Market efficiency means there's probably less momentum than polling alone, but the prices still have to track with polling (ie. impossible to know when a trend stops so the fair value is not obvious).
One reason you might see momentum is if no news is good news.
Consider a market on "will there be an earthquake above [magnitude] in [region] before [date]"
Every day that goes by without an earthquake, the probability drops slightly.
But occasionally, there will be a big quake, and the probability will jump to 1. Or there will be small tremors that could indicate a quake is coming. And the probability will jump upwards.
For politics, this might be "the only way this candidate loses is if there is a giant scandal, so each scandal free day is good news".
If this is what is happening, then momentum traders won't win or lose on average. Usually they will make a small win. Occasionally they will make a huge loss.
That’s really insightful - I imagine this is particularly true for sleepier races. A presidential race has the appearance of constant news and a long tradition of shakeups (e.g. the October surprise). A lightly polled race like the mayor of Milwaukee probably stays basically flat outside of polls (or insider trading)
It seems like the big problem here is that outcomes in different causal categories might differ in momentum’s impact.
Like, a solar event is pretty separate from human factors, while a revolution is entirely human. I’d put an election somewhere in between — it’s based on human factors, but also much better quantified than a revolution.
Absolutely - I’d speculate that something like a solar eclipse is MORE susceptible to momentum because there’s less to go on, but it’d probably be worth doing this again while filtering for e.g. natural events, etc to see how it holds
If Polymarket keeps growing in importance, and if your three basic theories are true, will we see politicians start spending campaign money to drive the outcomes on Polymarket? Is that even legal? Spending large amount of money might be very good for news making and push quite a bit of momentum trading
This actually happened! In 2012 someone put $4M on Mitt Romney in Intrade (the main prediction market back then). This one bet was 2/3 of all money put on him in the final two weeks of the campaign; there was speculation that it was an attempt to boost campaign morale, donations, and turnout.
Which it sort of did! I remember news stories about his increasing odds as "there's something happening the polls aren't seeing," which DID sort of change the narrative from "Obama is going to win."
In political campaigns, momentum also exists as an underlying phenomenon reflecting the gradual shift in public opinion. If a candidate picks up significant new support, it tends to take time for word of mouth, media to work. Polls definitely exhibit momentum. I'm not surprised there's a bit of that in prediction markets! Market efficiency means there's probably less momentum than polling alone, but the prices still have to track with polling (ie. impossible to know when a trend stops so the fair value is not obvious).
One reason you might see momentum is if no news is good news.
Consider a market on "will there be an earthquake above [magnitude] in [region] before [date]"
Every day that goes by without an earthquake, the probability drops slightly.
But occasionally, there will be a big quake, and the probability will jump to 1. Or there will be small tremors that could indicate a quake is coming. And the probability will jump upwards.
For politics, this might be "the only way this candidate loses is if there is a giant scandal, so each scandal free day is good news".
If this is what is happening, then momentum traders won't win or lose on average. Usually they will make a small win. Occasionally they will make a huge loss.
That’s really insightful - I imagine this is particularly true for sleepier races. A presidential race has the appearance of constant news and a long tradition of shakeups (e.g. the October surprise). A lightly polled race like the mayor of Milwaukee probably stays basically flat outside of polls (or insider trading)
It seems like the big problem here is that outcomes in different causal categories might differ in momentum’s impact.
Like, a solar event is pretty separate from human factors, while a revolution is entirely human. I’d put an election somewhere in between — it’s based on human factors, but also much better quantified than a revolution.
Absolutely - I’d speculate that something like a solar eclipse is MORE susceptible to momentum because there’s less to go on, but it’d probably be worth doing this again while filtering for e.g. natural events, etc to see how it holds
Indeed, the line is pretty amorphous, but a spectrum of causality with an imperfect correlation to momentum absolutely does exist.
And those sorts of correlations CAN be modeled!
If Polymarket keeps growing in importance, and if your three basic theories are true, will we see politicians start spending campaign money to drive the outcomes on Polymarket? Is that even legal? Spending large amount of money might be very good for news making and push quite a bit of momentum trading
This actually happened! In 2012 someone put $4M on Mitt Romney in Intrade (the main prediction market back then). This one bet was 2/3 of all money put on him in the final two weeks of the campaign; there was speculation that it was an attempt to boost campaign morale, donations, and turnout.
Which it sort of did! I remember news stories about his increasing odds as "there's something happening the polls aren't seeing," which DID sort of change the narrative from "Obama is going to win."
https://www.theatlantic.com/politics/2013/09/one-trader-bet-millions-romney-win/310726/