Propsheet FAQ

Where do the projections come from?

Derek Carty has developed THE BLITZ over the past 10+ years and 10,000+ hours, using skills and methods developed from his time in the sports analytics community working with some of the best spots minds in the world, many of whom are now making decisions for championship-winning franchises.  THE BLITZ accounts for the basics (player quality, home field advantage, etc.) plus loads of cutting-edge factors like offensive lines, game scripts, formations, weather, scouting combine data, and adjusted-defense based on individual player inactives. The secret weapon? It’s aided by football-loving data-scientists to account for subtleties like coaching changes and team philosophy.

How often do THE BLITZ Projections update?

Unlike THE BAT for MLB, things don’t change as quickly for NFL, so updates happen on a looser schedule.  When they are needed, it usually has to do with player injuries, which can create big shifts in which players will be on the field and how much they will be utilized in their team's offense.  THE BLITZ establishes an automated baseline in these scenarios, but a manual touch is always required to ensure that usage is being divided optimally among teammates.

While the goal is always to update projections as quickly as possible after news breaks, here is a rough schedule of the bare minimum updates you can expect each week. Our pages display the last time projections were updated, so you can check around these times and see if there’s been a recent update:

– Tuesday evening (or Wednesday afternoon): First preliminary run of projections
– Wednesday (or Thursday): First run of projections with manual usage adjustments
– Thursday evening: Adjustments for TNF inactives
– Friday and/or Saturday: More updates as for-sure Sunday inactives get announced
– Sunday morning: Several updates as official inactives get announced
– Sunday by 12:50 PM ET: Final main slate update
– Sunday by 3:50 PM ET: Final late slate update
– Sunday night: SNF update for inactives
 – Monday evening: MNF update for inactives

When is the best time to make to reference the model and the projections?

The projections are the most accurate right before the game has started.  HOWEVER, so are the sportsbook odds, as they become more efficient after bettors have shaped them.  Because opening odds are more inefficient, we generally recommend referencing the model and projections as soon as odds come out.

The implied projections on the Player Propsheet look wrong.   For example, how can the line be 10.5 receiving yards +100/+100 and the implied projection is 16.0?  Shouldn’t it be 10.5?

This would be true if NFL statistics were normally distributed, but most are not.  Betting lines are meant to represent a median, where half of the time the player will go over, and half of the time he will go under.  But if a player averages 10.5 yards per game (or projects for an average of 10.5 yards), he would go under 10.5 yards far more than 50% of the time.

As a basic example of how this could happen, let’s say this player has played 4 games.  In one of the games, he catches a deep pass and finishes with 25 yards.  In the others, he only catches shorter passes and finishes with 2, 6, and 9 yards.  His average is 10.5, but he goes under 75% of the time.  This is just one made-up example, but it’s an easy illustration of what actually happens in reality across all players.

So why is the implied projection 16?  Because in order for a player at this position to go under 10.5 half of the time and over 10.5 half of time, his average would need to be 16.

Why don’t THE BLITZ projections at EV Analytics match the ones at RotoGrinders?  One of them must be incorrect, right?

So both are actually correct.  Essentially, comparing the base projections at RotoGrinders to the lines directly can lead to some misleading assumptions.  Almost all projections (including THE BLITZ) are means (which are usually higher), while sportsbook lines are medians.  If you compared them directly, it would tell you to bet the over far too often.

Additionally, THE BLITZ is not directly projecting the chance a player gets injured, which means they're skewing further too high.  (ex. Patrick Mahomes will project for 100% of the Chiefs QB snaps, but in reality, his true snap rate accounting for the chance of injury may be closer to 95%).  For DFS purposes at RotoGrinders that doesn't really matter; everyone's chance of injury is more or less than same, at least as far as we can project it.  But for betting over/under a concrete number, it matters a lot.

So the number you see on EVA is taking the original projection you see at RG and converting it into a more apples-to-apples number with the betting lines.  At its core, it's the same projection; the version at EVA is just a more accurate representation of the projection for betting purposes.

Can you explain what the Expected Value is and how you came up with it?   And can you provide a numerical version of Expected Value?

On our player prop pages, our Expected Value is a 0-to-5 “Plus” visual depiction of potential value.  The more Plus marks, the more potential value we believe that bet offers.  You can sort the Expected Value column.  Although we provide this column and a suggested side to bet, we are not explicitly telling you to bet it.   We want to empower users with data tools that they can use to save time and make more educated bets.  As such, we refrain from telling you exactly what to bet and how much, and although our data tools provide valuable insight, we are not specifying which bet has the most value if there are multiple with the same Expected Value “Plus” mark score.  This is actually done with the user in mind.  Otherwise, if everyone is betting the exact same bets that have the highest numerical Expected Value, the lines would move quickly and only a select few users would get them before they move.  Plus, this is a good way to get limited by a sportsbook.  As a result, we came up with the “Plus” value system as a better way to present the data.  It’s worth noting that anytime you see a betting market with any Plus marks for its Expected Value, the projections believe that betting market offers positive value.  And if you could bet that same market over and over again, the projections believe you would end up being profitable over a large sample size.  That being said, we encourage users to evaluate all the data and not blindly wager everything that shows Plus EV.  Sportsbooks frown upon bettors who bet a ton of volume and betting player props at scale is a good way to flag yourself as a sharp bettor and get limited or cut off, regardless of whether you are up or down.

Can these projections and models make me money?  

Historically speaking, yes.   However, it’s important to realize that sports betting isn’t just about having a good projection system.  Sports betting is a grind where edges are typically small, but these small edges add up over time.   Most bettors struggle with the rigors that are associated with this, so to make money using these projections, you must be price-sensitive, able to effectively manage a bankroll and time bets, and withstand roller-coaster like emotions. As such, it’s important to realize sports betting is not for everyone.  If you’re looking to make a quick buck, we advise you to look somewhere else… or simply acknowledge that sports betting can be a fun and legal form of entertainment if it’s handled responsibly.  

If the system is profitable, why share it with me?

An inconvenient truth for aspiring bettors is that sportsbooks are in business to make money.   As such, sportsbooks are quick to restrict sharp action or lower limits on markets that don’t help their bottom line.   This creates a glass ceiling for professional bettors who have worked hard to develop profitable models but get limited after showing profitability.  At some point the challenge of profitability becomes two-fold.  Not only do you have to develop profitable models, but you must find legal outs or sportsbooks that will take your action.  This glass ceiling has led to the formation of EV Analytics, a bootstrapped data company that has pivoted its focus to offer valuable and innovative data in the sports betting space.  We want to empower fans and companies with valuable data.