The Greatest Angles Ever - #2: Under Every NFL QB Prop

In 2020 NFL props are an accepted mainstream bet, and if anything, I tend to see more prop picks than full game side/total wagers on the typical tout programming. Despite the stigma around props with some known professional bettors, if booked competently props should be a windfall for the sportsbooks. Most players, who are squares who bet more or less at random, are happy to lay -115 or even -120 on a prop, a price they would never pay on a full game side or total. Even better, props give the books the opportunity to take in far more volume, as players can wager multiple times on the same game while still remaining within their personal risk tolerance on any single bet. While the equation of handle * hold looks amazing, the challenge is setting up lines good enough that you don't lose much to sharps, who will only bet your props when it is bad for you.

If I were starting a book in the US I would kick out/limit no one but protect myself by only booking major US sports, only the majors in golf and tennis, Euro Big 4/MLS/Mexican soccer, and the absolute biggest e-sports events, because only sharps bet other stuff and small soccer, tennis and e-sports is all fixed. To further protect myself I would only offer good limits the day of the game and in some smaller "major" sports, only after lineups. Finally I would hire the sharpest prop people in the business to set my props and book good limits on them, the sharpest live/halftime people I could find for that part of the business and offer a leading product there, and just copy CRIS for everything else.

I think if any book follows this simple advice, they can take on all comers with a world-class betting product and have almost no chance of losing to all but the most elite bettors and only slightly to them, while making more money from the vast majority of wannabes and recreationals. I would also note that this is not an original idea but instead is more or less what the once-decent books in Vegas did for some time barring the live/prop aspect. In any case, this may be a pipe dream as most people competent enough to set up good lines on the props/live probably wouldn't work for what a small book could afford to pay and unless I had a large operation I could never book enough action to make it worth it anyway.

The major challenge to booking player props is not pricing out the line itself based on past stats, but that they only involve one player, whose role in the game may change drastically from week to week. In a full game side or total, other than basketball or the quarterback in the NFL, an injury to a single player is not going to bury a book. The line might move 20-30 cents for a really big time injury, and of course you are only going to get limit bets on the "good" side when said injury is known, which will always be to those in the know rather than you, but these types of injuries or lineup changes are not all that common. Just looking at historical line moves in most sports it is readily apparent that most games don't move much more than a standard -110 point spread barring very early action and many games get great two-way action allowing books to chop up bettors on both sides.

With some NFL player props it is totally different. While books are somewhat protected by "no snaps - no action" a simple change like a running back splitting time or the WR4 being promoted to WR3, the type of thing that is super hard to always know about no matter how much you follow the news, can easily move fair value on a prop line 25% or even more. I was never big on these props for that reason. The full-time prop bettors can make big money, assuming you find places to bet, but I always either had a real job or was betting other sports at the same time and did not have the entire week to sift through NFL practice reports and the like. I prefer betting stuff that can be modeled, and while you can and should build a model for these types of player props, information on playing time and role will always be far more important. This is even more true in the NBA.

For me the best player props of all were the NFL quarterback props. Barring injury the quarterback usually plays the entire game and in situations where this is not 100% the case, there usually are no props offered anyway. This means it is more of a static event where modeling over a large sample of games based on past stats alone is much more effective. In fact for some props, the identity of the QB barely matters at all with the pregame spread and total carrying most of the predictive value.

I am not sure how long these were offered other than on the Super Bowl, but I know it was the early days of these props in 2011 when I first bet them as props were constantly added at various books around this time in the following years, and the lines were absolute garbage. Five units won on 50 QB prop plays was a typical week. The books really paid the price for the semi-decent prop lines they have today.

Back in those days there were occasionally expanded offerings for SNF/MNF, but there were four main props that everyone had on every game:

Total Completions and Total Passing Yards: These lines were very weak, but there were no dead obvious angles meaning they to be handicapped. It was apparent that the books set the lines based on past QB averages. But as any DFS player knows, matchup and game flow are far more important at the QB position. In pricing these props I didn't use past QB stats at all initially other than team passing rate as past QB stats are only weakly correlated to future outcomes in these categories. Instead spread, total, team passing preference, and weather conditions were king.

The best plays were always under everything on big-name QBs, preferably in games where they were big favorites, in a low total game due to wind. It was quite fun to bet, because the nature of the angle meant that unders usually won right around the number due to teams running out the clock on the last two drives, so it always felt like you "escaped" with a miracle win. What few overs there were were always on bad QBs facing favorable conditions, which often won on trash touchdowns on the final drive, so those were fun also.

Total TD Passes in the game: This was one of the strongest prop angles ever and it was always on the under. For whatever reason, books did not price the distribution of TD passes correctly, in a similar mistake to what they made with the Under 10.5 angle described earlier in this series. Usually this prop was lined at 1.5 with plus to the under if the QB was a big name on a good team, and a slight minus to the under otherwise. Even before I built models for the other stuff, just from going through past box scores by hand, it was obvious that under 1.5 covered way more often than the prices reflected.

Once again, past QB stats had almost no correlation to this prop. It was all about team total with slight adjustments for team pass tendency and wind. Just betting every single one blind would have easily won, but there was little value on bad QBs in higher team total games, so I passed those. Usually I would bet around 10-15 of these each week, always on the under.

Yes/No Interception in the game: This was the best prop of all, but for a long time I did not bet it, because I thought I had no model that could predict it well. It turns out that is the secret to yes/no interception betting, that interception rate highly regresses to the mean.

Rather than any stats related to the QB or actual ability of the QB, the biggest predictor of whether a QB will throw a pick in the game is the pregame point spread. Interception rate drastically rises when teams are chasing down in the game late; in 4th and long late it is around 4-5 times more likely that an INT is thrown than in a neutral game state.

There are other stats that predict INTs in addition to the spread, such as average depth of target and sack rates. While sack rates tend to be highly correlated to the spread anyway, average depth of target is not; I did not have that stat back in those days, and it is only a minor factor anyway. There is some slight additional lift to using past QB stats and experience levels as well. In particular, running QBs tend to be good for the under, since they often get away without throwing a pass on 3rd or 4th and long, where interception rates are highest.

Some of the other stats that seem obvious do not matter for this prop. For example, you would think that teams that threw a lot of passes in past games would throw more interceptions. But, teams that pass more tend to do this because they know they have good quarterbacks, who throw fewer picks, and it ends up being a wash. In addition, interceptions are far less common on early downs, which is where most of the difference between "pass-heavy" and "run-heavy" teams lies.

All this meant that for the most part, this was a brain dead obvious angle that required no handicapping after two or three weeks of learning how the lines were set. If teams were heavily favored, you bet the under, unless it was a "good" quarterback known for not throwing picks such as Rodgers; then the line on the under would be too high, so you didn't bet it. And you bet the under for all "bad" quarterbacks pretty much regardless of the spread, even if they were the type that threw a million picks in past games. Often the "no" would pay +160 or more, when it rarely should be more than +125. So mostly, you bet the under.

The most memorable bet on this prop I made was in 2014, when lines were somewhat better than they were when I first started, but still very beatable. Johnny Manziel made his starting debut for the Browns in week 15 against the Bengals. For those that don't recall the 2014 NFL season, this was a year that the Browns somehow got to 6-3 and 7-4 through the time-honored strategy of winning every close game and forcing a million turnovers, before losing out to miss the playoffs because they were still the Browns.

I am not sure why Manziel started for a team that was 7-7 in the NFL, but the Browns were a slight home favorite in the game and in this special case there was no "yes/no" interception betting. Instead, there was an over/under, and at Bovada at least I recall the total was actually 2 albeit heavily shaded to the under. It is very hard to throw three INTs in an NFL game where your team is a one-point favorite, and the under 2 was the play by a million percent, but sadly it pushed and the 1.5s dealt elsewhere all lost, only for Manziel to go pickless in the next game, where he was taken out to an injury.

These days, openers on QB props and probably other player props are still bad, but get pounded into the ground by sharps at every PPH and offshore as soon as they are released. Just try it, you can mash refresh on any major PPH or offshore book as props are posted and by the time the props come up on the next game the +EV props that were just posted on your last refresh will all be -160. However, with the rise of new sportsbooks, some of which may not always kick people out after two days, and increased popularity of more exotic props, there will always be opportunities in the prop space for those willing to hustle hard for news, models, and accounts.