Version that is still open: https://manifold.markets/WilliamKiely/will-pennsylvania-be-decisive-in-th?play=true

I just found this question, but too late to move the price. I would have bought yes on both the mana and sweepstakes markets at 28% and 36% respectively. My credence is ~50%.

I made this (duplicate, without realizing) question, that is still open: https://manifold.markets/WilliamKiely/will-pennsylvania-be-decisive-in-th

Made into sweepstakes!

Bump

I was wondering why @MaximLott bought yes so I looked at more real-money prediction markets and I found PredictIt has 10% for dems by 15-34, while polymarket has 5% for the same contract? This seems big enough to actually profit in real money while accounting for fees. Am I wrong? Can someone who actually likes dollars and isn't afraid to make real money bets fix this for me?

@BoltonBailey I don’t think it makes sense to arb anything using Manifold after taxes. Buying sweeps cash doesn’t seem like something that could be counted as a loss to offset income from the arb.

@JakeLowery I am not a financial professional and this is not advice, but I don't see why losses on one prediction market wouldn't be a capital loss offsetting a capital gain on another, as described here. Maybe it's different for sweeps with this whole loophole model they're going for, but here I'm purely talking about PredictIt and Polymarket.

@BoltonBailey Yes I think the difference is exactly because of the sweeps loophole. If you buy mana and get sweeps cash for free, what is the cost basis of your sweepscash? Zero. If you lose your sweepscash it would not hold up in an audit as a capital loss IMO. In fact if you buy $100k of mana+sweepscash and break even on your sweepscash betting, then you cash out $100k, I think the IRS will see that as $100k of earned income. I am not a CPA and this isn’t advice of any sort.

Arbitrage on Poly/Kalshi is a different story. PredictIt, account for the fees on profits and withdrawals too.

What am I missing here? It seems like this market keeps being significantly overpriced based on its description.

For the market to resolve YES, the electoral college margin needs to be no more than 40 points. The only elections where that's happened since Alaska and Hawaii became states were 2000 and 2004. So if we count the past six elections, that's a 1/3 chance, and the base rate is lower if you use any other window. It then also needs to hold that the winner wins Pennsylvania, which seems like it would not be much more than 70/30. (This is distinct from the unconditional probability of the winner winning Pennsylvania since this is in the case of a very close election.) The odds say it's a very close election given polls to date, but we're still almost two months out, and errors are likely to be correlated. (Also, I don't have Nate Silver's forecast, but at least recently it had the electoral college outside this bandwidth.)

I'd also think the odds of Pennsylvania being the tipping point state should be higher than this market, and Nate Silver recently had those at 32%.

I suppose I should also be betting on this and other markets.

It looks like maybe there's a miscalibrated bot automatically keeping this market high?

@ZacharyFreitasGroff No idea why @acc is so active here, as I recall it's just supposed to assess whether you are underconfident or overconfident and trade based on that.

My current best explanation is that people were perhaps betting on this market without carefully considering whether the numbers were correct first, or just mistakenly thinking this was about PA being the tipping point state.

As far as I can tell, 538 has the <40-point EV margin at about 20% and polymarket at about 30%.

oops misread the rules then Manifold gave stale odds when making a bit

oh well...

Potential arbitrage with https://manifold.markets/ManifoldPolitics/which-state-will-be-the-tipping-poi

@GustavoLacerda Yes, probably very correlated. More direct arb would be the 38 option I just added on this market (which also seems too high to me relative to 538 and polymarket).

@BoltonBailey It's a parlay on both, right? This is true when Pennsylvania is the tipping point AND the margin is 38 or fewer. Not true, this can resolve yes even if PA isn't the tipping point.

When the margin is under 38 in general, Pennsylvania being the tipping point implies the winner won PA, so yes, this will resolve YES when PA is tipping point and margin is under 38.

I don't see why it couldn't be the case that the margin is under 38, Pennsylvania is not the tipping point, and the winner still wins PA. It could just be the case that there is another state which also would have flipped the election which has a lower in-state margin than PA.

And conversely, PA could be the tipping point and this market could resolve no if the EV margin is >38.

They are still correlated though, I think it's just not a risk-free arb either way.

@BoltonBailey Yeah, you're right that this can resolve YES even if Pennsylvania is not the tipping point. It feels weird to say that Pennsylvania decided the election in this case, but it is true. (If Kamala wins, there's a good chance the "California decided the election", which is really weird.)

This just seemed massively overpriced to me on the basis of the 289 criterion alone.

@BoltonBailey Not to mention that the winner of PA losing the election is trading around 10% and is anticorrelated with the 289 thing. Am I crazy?

Upgraded to Basic