Bet the Process: Prediction Markets and Market Making

Episode Overview

The hosts welcome Henry Karens, affiliated with Novig and the trading group Manhattan Athletic Group, to discuss prediction markets, sports market making, pricing, risk controls, and how market efficiency is changing.

Key Discussion Points

• A costly golf-market mistake: Rufus’s team lost about $185,000 after quoting golf parlays while unaware that the tournament’s final round might be canceled. The results were ultimately settled using 54-hole scores, as some bettors had anticipated.
• Automated safeguards stopped further quotes and limited the loss.

• The incident prompted the team to reconsider edge cases, overnight quoting, and which markets a small team can safely manage.

• Risk controls must be clear and usable: Henry cautions that overly sensitive safeguards can cause “alarm fatigue,” leading traders to ignore alerts. He and Rufus emphasize that controls should be understandable and consistent—for example, if a trading interface shows a quantity of zero, the team should not make exceptions by accepting small bets anyway.

• How market makers decide what to quote: Henry’s group trades across multiple venues and can both make and take bets, including through request-for-quote systems. Its approach varies by market:

• It may post limited amounts and wait to see how the market responds.
• It can time quotes around events such as kickoff.
• It adjusts aggressiveness based on the expected activity and behavior of bettors.

• It may use its own pricing view while keeping quotes close enough to market that its opinion is not obvious.

• Market data is not always available: Some markets, such as alternate lines and custom combinations, lack reliable external prices. Teams must decide whether to quote using their own models, require outside market data, or limit exposure. The guests stress that adding exceptions and extra safeguards can make systems harder to operate.

• Trading across venues helps develop skill: Henry’s group maintains accounts across exchanges. The guests argue that different fees, order books, and market structures teach traders how markets behave—and help prepare them for competition as prediction markets evolve.

• Prediction markets and efficiency: Henry believes markets have generally become more efficient as more activity moves into open, two-sided trading venues. However, some opportunities may remain less efficient.

• Prediction-market RFQs can let traders express an opinion without fully revealing it through a prominent market bet.
• Over time, competitors tend to find hidden edges, making those opportunities less effective.

• A durable approach is to build strong baseline systems, exploit new opportunities early, and expect them to become more efficient.

• Live and second-half markets: The guests discuss how prediction markets may create more opportunities to act on in-game opinions before halftime. As a result, some information that once appeared mainly in second-half lines may now be reflected earlier.

• Rufus’s preliminary college-football testing suggests that market prices move partway toward his model’s numbers during halftime, with further movement later in games.
• He stresses that the sample is small and that observed price movement does not necessarily equal realized model value.
• Measuring market movement is difficult: the number of bets, liquidity, price changes, and related markets all affect how much signal a trade represents.

Other Episode Notes

• The hosts’ weekly pick is Charlotte +10 at home against Tulane, scheduled for Friday, which is also Tom’s birthday.
• The episode is sponsored by Novig. Listeners are encouraged to visit novig.com and use promo code BTP.