A specific $2,500 threshold for a 'longshot bet' feels too rigid; it's like setting a universal price for a luxury item without considering local purchasing power.
For instance, a £500 bet in a market with an average stake of £50 represents a 10x deviation, which is a much stronger signal than 2,500 in a 10,000 average market, which is only a 0.25x deviation.
This fixed boundary could miss 30% to 50% of relevant signals in smaller markets by setting the bar too high.
We should rank these bets by their relative impact, perhaps using a multiple of the average transaction value.