So I get the conventional wisdom that the sportsbook's goal is to get equal money on both sides of the bet by adjusting the odd and ensuring that the odds are such that the sportsbook either makes money or does not lose money regardless of outcome, presumably because the sum of the implied probabilities behind the odds offered for both sides exceeds 100%.
However, my experience with DraftKings today in betting on point spreads on NFL games seem to defy this conventional wisdom. I will use a more extreme and recent case that I encountered to illustrate:
As of Sunday night before the Monday Night Game, DK had the LA Rams as 2.5 favorites (-2.5) with odds for the bet at -112. DK had the Dolphins as underdogs by the same spread (2.5), offering -108 odds. Ok so far. However, according to DKN DraftKings Network, https://dknetwork.draftkings.com/dra...orrow&tb_emt=0, the "handle," that is the split, in the amount of money wagered was 71% for the LA Rams and 29% for the Dolphins.
If you look at a hypothetical of this, and say that the total amount bet on the game is $100, then the following presents itself:
$71 on the Rams, $29 on the Dolphins.
The payout on a $1 Rams bet (i.e., Rams to beat the 2.5 spread) would be $1.89. The total payout to the Rams' bettors would be $136.74 [i.e., 71 one-dollar bets x $1.89], meaning that DK would have a loss of $36.74 and would need to dig that much into its pocket to cover the Rams payout. (If the Dolphins bettors won, i.e., the Rams did not win by more than 2.5 points, the total payout to Dolphins bettors would be $55.85, the payout on a $1 bet being $1.93 multiplied by 29. So DK would have a profit of $44.15 under that scenario.)
So how can DK go ahead with the money so imbalanced and why doesn't it adjust the odds further to try to balance the money on both sides? I would love to have someone explain the math and logic of this.
My answer would be that the DK handicappers must be concluding that the probability of the Rams winning by more than 2.5 points is sufficiently low that the expected value of that bet does not exceed $100. In other words, if the game were played 100s of times, the Rams' bettors would be on the losing side enough times to make the expected total payout less than $100. Here is how I figure the math: To avoid a loss, the expected value payout cannot be more than $100, that is breakeven for DK. To calculate the discount from 100% or the highest probability of a successful LA Rams bet without going over $100, you would solve this equation: $100 = Probability of Rams Bet x 136.74 = > Probability = 100/136.74 = 73.13%.
So the DK handicappers must have some high level of confidence that the probability of a successful LA Rams bet would not exceed 73.13%, in order to have the confidence to allow the book on these bets to be so imbalanced. In fact, if you use the implied probability derived from the odds on these bets as a rough proxy for what the DK handicappers think the true probability is on these bets, then you see that a -112 odds on an LA Ram bet translates to a 52.83% implied probability of success on the bet. And if you plug that number into the EV calculation, you see that expected payout to Rams' bettors would be $72.24, leaving DK with an expected profit of $27.26. (My reckoning is that DK would never be at risk of losing money on a Dolphins bet with this type of imbalance, because even if the probability of success were 100%, the house would still show a profit.).
Of course, the DK handicappers must have some sense of the variance here to allow this type of imbalance to persist. If the variance is not huge, the risk of being wrong about the true probability such that the true probability exceeds 73% is low and this imbalance is defensible.
****
That's my laypersons thinking about this. I would welcome more educated views and analysis on this.
However, my experience with DraftKings today in betting on point spreads on NFL games seem to defy this conventional wisdom. I will use a more extreme and recent case that I encountered to illustrate:
As of Sunday night before the Monday Night Game, DK had the LA Rams as 2.5 favorites (-2.5) with odds for the bet at -112. DK had the Dolphins as underdogs by the same spread (2.5), offering -108 odds. Ok so far. However, according to DKN DraftKings Network, https://dknetwork.draftkings.com/dra...orrow&tb_emt=0, the "handle," that is the split, in the amount of money wagered was 71% for the LA Rams and 29% for the Dolphins.
If you look at a hypothetical of this, and say that the total amount bet on the game is $100, then the following presents itself:
$71 on the Rams, $29 on the Dolphins.
The payout on a $1 Rams bet (i.e., Rams to beat the 2.5 spread) would be $1.89. The total payout to the Rams' bettors would be $136.74 [i.e., 71 one-dollar bets x $1.89], meaning that DK would have a loss of $36.74 and would need to dig that much into its pocket to cover the Rams payout. (If the Dolphins bettors won, i.e., the Rams did not win by more than 2.5 points, the total payout to Dolphins bettors would be $55.85, the payout on a $1 bet being $1.93 multiplied by 29. So DK would have a profit of $44.15 under that scenario.)
So how can DK go ahead with the money so imbalanced and why doesn't it adjust the odds further to try to balance the money on both sides? I would love to have someone explain the math and logic of this.
My answer would be that the DK handicappers must be concluding that the probability of the Rams winning by more than 2.5 points is sufficiently low that the expected value of that bet does not exceed $100. In other words, if the game were played 100s of times, the Rams' bettors would be on the losing side enough times to make the expected total payout less than $100. Here is how I figure the math: To avoid a loss, the expected value payout cannot be more than $100, that is breakeven for DK. To calculate the discount from 100% or the highest probability of a successful LA Rams bet without going over $100, you would solve this equation: $100 = Probability of Rams Bet x 136.74 = > Probability = 100/136.74 = 73.13%.
So the DK handicappers must have some high level of confidence that the probability of a successful LA Rams bet would not exceed 73.13%, in order to have the confidence to allow the book on these bets to be so imbalanced. In fact, if you use the implied probability derived from the odds on these bets as a rough proxy for what the DK handicappers think the true probability is on these bets, then you see that a -112 odds on an LA Ram bet translates to a 52.83% implied probability of success on the bet. And if you plug that number into the EV calculation, you see that expected payout to Rams' bettors would be $72.24, leaving DK with an expected profit of $27.26. (My reckoning is that DK would never be at risk of losing money on a Dolphins bet with this type of imbalance, because even if the probability of success were 100%, the house would still show a profit.).
Of course, the DK handicappers must have some sense of the variance here to allow this type of imbalance to persist. If the variance is not huge, the risk of being wrong about the true probability such that the true probability exceeds 73% is low and this imbalance is defensible.
****
That's my laypersons thinking about this. I would welcome more educated views and analysis on this.
Comment