You Want Me To Pick A Team That’s Expected To Lose?

On some occasions, we will recommend a team that we believe has a less than 50% chance of winning, due to other strategy reasons.

If you’re seeing a game winner pick with win odds of less than 50%, or a point spread pick that is not listed as a “Model Pick,” then yes indeed, it’s true. We’re telling you to make a pick we think is more likely to lose than win.

Why on earth would we do that?

To make a long story short, it all depends on your goal and pool details. Just picking all the most likely teams to win is the most conservative strategy you can employ. It will typically serve you well in small pools, or when there are lot of games to pick, such as in a season-long football pool.

Beyond that, however, strategy gets significantly more complex. Here are a couple examples:

Point Spread Pools

In point spread based pools, a fair number of picks will be very close to 50/50 propositions, especially if your pool uses spreads that are close to current Vegas odds. Most spreads are pretty “efficient” — that is, they are as good as a prediction as anyone could come up with, given the information known before a game starts. (There are exceptions, of course, but spreads have generally proven to be good predictors overall.)

In this situation, anticipating how your opponents will pick a game can make a big impact on our recommendation. For instance, our models may think a certain popular team, playing at home, has a 52% chance to cover a big spread against a small, non-conference opponent. That’s pretty close to a coin flip.

However, what if we then find out that 80% of the public is picking the big name team to cover the spread in national pick’em pools? A huge bias like that happens sometimes, especially in these types of situations.

In this case, the no-name team — technically the “expected loser” according to our models, with less than a 50% chance to cover the spread — may well be the better pick to make. With only 20% of opponents picking them, but a 48% chance to win, the no-name team is a great value. That’s because the benefit of a correct pick of the no-name team is you gaining ground on 80% of your pool, while the benefit of a correct pick of the popular team is you gaining ground on only 20% of the pool. The penalty for a loss is similarly higher, but in a pick’em pool where only a small percentage of entries win prizes, maximizing your chance of a very good outcome can sometimes be more important than minimizing your chance of a very bad one. In this case, a 48% chance to cover is close enough to a coin flip that the slightly higher risk in picking the no-name team can be easily justified.

Defending a Lead

If you’ve built up a lead in a pool and it’s getting late in the season, your primary goal is to defend your lead. That means you want to minimize opportunities for your opponents to gain ground on you.

Let’s say in this position, you have to pick a game where our models think the home team only has a 47% chance to win, but 75% of the public is picking them. As it turns out, a solid strategy here could be to ignore our model prediction, and just follow the crowd and pick the home team too.

If that sounds weird, here’s the logic. Even though we technically expect the home team to lose, your decision to pick with the crowd means that 75% of your opponents immediately lose any chance to gain ground on you. Either you and 75% of your pool are both going to win, or you’re both going to lose. Just like that, you’ve neutralized three quarters of your opponents.

In addition, the 25% of people who picked our expected winner, the away team, are only slightly over 50% likely to get the pick right. If they get it wrong, then literally no one in your pool makes up a point on you with this game.

In total, by following the extreme bias of the crowd with this pick, only about 1 in 8 people in your pool are expected to gain ground on you. There’s a decent chance that none of your closest competitors in the standings are one of them.

Especially over the course of a season-long football pool, there can be so many potential cases like these that the complexity gets to be mind blowing. That’s why we built computer algorithms to figure it all out for us, and that’s why we trust their pick recommendations even when some of them look a little crazy to the human eye. Our software can process all the variables in play in order to figure out the best picks to make each week in order to maximize your odds to achieve your goal; our minds can’t.

In conclusion, picking expected losers makes total sense sometimes. Every opponent you have that laughs at you for doing so is just another sucker we are going to beat in the long run!