One tension in this model (we'll see how it plays out) is the fish aren't exactly the same as in the sportsbook. You're inviting in all the "fish" in the sea to trade on the exchange, some of which are salmon (uninformed) while there are also piranhas (sharp flow). If the exchange listed products are tighter/deeper than sportsbook lines, creates opportunities for the sharp flow which makes it more difficult for MM on an exchange to tighten vs a sportsbook since the exchange won't ban/limit. MM need to be sharp.
I find this exchange model to be a fundamentally more fair one overall.
This is a great point - everyone has been comparing the prices of the two but it’s a fundamentally unfair comparison bc you get to ban adversity on one, but on the other the spreads are necessarily long-run slightly blown out by the piranha.
Good question. When I calculate the hold, I'm using the filtered data, so it's a clean representation of the win rate. But I can see how that's confusing in the table.
> easier to do a privileged-market-maker sweepstakes model like Novig
~1/3 of the US population cannot participate in Novig's model: https://www.novig.us/ so it's not disingenuous to imagine their exchange was their most feasible path to operating something nationwide sportsbook-esque (whether directly through Kalshi Trading or indirectly through partners like Susquehanna International Group)
Great post as usual Adhi!
One tension in this model (we'll see how it plays out) is the fish aren't exactly the same as in the sportsbook. You're inviting in all the "fish" in the sea to trade on the exchange, some of which are salmon (uninformed) while there are also piranhas (sharp flow). If the exchange listed products are tighter/deeper than sportsbook lines, creates opportunities for the sharp flow which makes it more difficult for MM on an exchange to tighten vs a sportsbook since the exchange won't ban/limit. MM need to be sharp.
I find this exchange model to be a fundamentally more fair one overall.
This is a great point - everyone has been comparing the prices of the two but it’s a fundamentally unfair comparison bc you get to ban adversity on one, but on the other the spreads are necessarily long-run slightly blown out by the piranha.
Doesn't this underrepresent hold though if you're:
(1) Excluding PNL for unsettled trades.
(2) Including volume/handle.
Maybe I misunderstood.
Great post though. Love to see the data!
Good question. When I calculate the hold, I'm using the filtered data, so it's a clean representation of the win rate. But I can see how that's confusing in the table.
> easier to do a privileged-market-maker sweepstakes model like Novig
~1/3 of the US population cannot participate in Novig's model: https://www.novig.us/ so it's not disingenuous to imagine their exchange was their most feasible path to operating something nationwide sportsbook-esque (whether directly through Kalshi Trading or indirectly through partners like Susquehanna International Group)