FantasyLabs’ mission is pretty simple: deliver the best data and provide the tools necessary to transform that data into winning daily fantasy sports lineups.
And we offer a lot of data. Some of the stats are pretty self-explanatory, but others either aren’t widely used or were created by us, meaning you might have no idea what the hell they mean.
We have tool tips across the site that you can hover over to learn more about specific stats, but I wanted to create a guide to all of the numbers we offer and just fill you in a bit more about the philosophy behind each stat, why we use the numbers we do, and how to get the most out of the data.
About +/- (Plus/Minus)
A lot of our DFS-focused stats—the stuff that’s sport-agnostic—are centered around a concept we call “Plus/Minus” (+/-). Simply put, a player’s plus/minus is his actual points minus his expected points. So if Bradley Beal scored 300 points over the past 10 games and his expectation was 250 points, he would have a Plus/Minus of 50 total points, or +5.0 points per game.
Cool. So how do we know what to “expect” from each player? We know based on our database of historic salaries and fantasy performances on each daily fantasy site. Instead of using a fragile $/point system (or, even worse, sorting players into completely arbitrary tiers), we use historic performance data to help calculate exactly what to expect out of a player based on his cost. So if Beal costs $6,000, we know he should produce X points, on average.
Using Plus/Minus, we can calculate all kinds of really cool stats and identify league-winning trends. Our Consistency stat, for example, shows how often a player has exceeded his expected points. Instead of using a “well-how-often-does-he-reach-4x-his-salary?” system that naturally overvalues cheap players, we put every player on a level playing field since all of our stats are adjusted for cost.
Our Bargain Rating is a historical percentile rank representing how much of a bargain a player is on one daily fantasy site versus the other. We look at the typical difference in site salaries at a position and then rank a player based on how much of a bargain he is in a particular game relative to the historical data.
If a player costs the same on DraftKings as he does on FanDuel, for example, he might have a high Bargain Rating for DraftKings and a low Bargain Rating for FanDuel since the latter site has a smaller salary cap and thus tends to price certain players much lower, on average.
The Bargain Rating stat is extremely powerful and useful in a number of ways. First, there’s a strong link between Bargain Rating and player value (Plus/Minus). That shouldn’t be surprising since Plus/Minus is determined based on price; the cheaper a player, the more potential value he can offer (assuming the same skill level). It is smart to use Bargain Rating in your player models, especially in a sport like basketball in which it pays to be price-sensitive.
Second, Bargain Rating is an excellent way to determine where to get exposure to certain players. If you play daily fantasy sports on both DraftKings and FanDuel, you should get exposure to the players you like where they’re the cheapest. A big part of finding value is leaving yourself a cushion to soften the negative impact of assessment errors, and Bargain Rating does that better than any other stat.
The percentage of games in which a player has produced within a standard deviation of his expected points based off of historical scoring and pricing
To identify high-floor players for cash games
Upside figures show the percentage of games in which a player has finished at least one-half standard deviation above his salary-based implied total
To identify high-upside players for tournaments
Our Dud stat calculates the percentage of games in which a player finishes at least one-half a standard deviation below his salary-based implied total
To identify low-floor players to avoid in cash games
Our Trends product lets you leverage our massive database of historical salaries and fantasy performances to determine in which situations players traditionally offer value. You can create your own trends or utilize our DFS-pro-created ‘Pro Trends,’ which already show up in models and player cards.
Our Pro Trends are very strongly linked to value, and they allow you to truly customize your models based on angles you find.
Moneyline percentage, or the percentage of bets coming in on each team in a game. We aggregate all our Vegas data from seven different sportsbooks.
There’s a positive correlation between public betting trends and player value. You can also use public betting trends to help predict player/team ownership in tournaments.
A player’s change in salary over a given period of time
To help identify players whose price might be artificially inflated/deflated due to variance
The Daily Fantasy Basketball Glossary of NBA Stats
We have a typical median projection, but then we also calculate a ceiling and floor for each player. Those numbers are what we expect the player to reach with 15 percent probability, i.e. a 15 percent chance of exceeding the ceiling and a 15 percent chance of finishing under the floor (and thus a 70 percent chance of falling between the two numbers).
We calculate ceilings/floor numbers based on predictors of volatility. We look at things like the usage rate of players, on-off statistics, players’ shot charts in general and opposing shot charts, etc. Then we assign a volatility rating to each player and project his range based on that.
To find value in either GPPs (ceiling) or cash games (floor); it’s also useful for visualizing each player’s range of outcomes so you can get exposure to certain players who have access to the type of range you need.
Simply put, this is median projection minus salary-based expectation. We just compare how we think a player will perform most of the time versus what he should do (historically) based on his cost.
A great way to identify pure value for all leagues
This is the typical “points-per-dollar” you see across the DFS industry. It has its uses and can be valuable for cash games, although we think it should be just one component of your model. It is simply projected points for every $1000 of salary.
A way to identify pure value for all leagues
Opponent Plus/Minus Allowed
This metric measures a team’s fantasy production allowed versus a specific position. However, unlike traditional, standard defense-versus-position metrics, we adjust for playing pricing. It’s a nuance that’s very important — if a team is letting up a lot of fantasy points to point guards but they’ve gone against Chris Paul, John Wall, and Damian Lillard to start the season, their Opponent Plus/Minus won’t be as poor as a team that got torched by TJ McConnell.
An extremely predictive way to judge matchups for all leagues
Simply the points we expect a player to score (historically) based on his cost.
More as a way to visualize expectations
PER (Player Efficiency Rating)
PER is the overall rating of a player’s per-minute statistical production; the league average is 15.00 every season. This is the basic (and first) one-number metric to rate a player’s on-court efficiency. While it doesn’t adjust for teammates or other factors like newer one-number metrics like RPM, in DFS it doesn’t matter. RPM and similar statistics measure how good a player is in his specific role — in DFS, we’re merely concerned with overall production, independent of that.
A way to rate players on a per-minute basis
Usage is the percentage of team possessions used by a player while he was on the floor. Possessions can be “used”, or finished, in three specific ways — a made shot, a missed shot and a resulting rebound, and a turnover. Both usage rate and team possessions per game are relatively stable statistics from game-to-game, so usage is perhaps the most important statistic outside of per minutes played to judge a player’s fantasy potential.
A foundational way to judge a player’s fantasy potential, combined with minutes
B2B (Back to Backs)
NBA is unique in that it is a fairly daily sport (like MLB) but it’s also very taxing on a nightly basis (like NFL). As a result, players typically have very strong splits when playing in back-to-backs and especially when they play a lot of games in a short period of time, like four games in five nights, for example.
A way to predict whether players will underperform
As hinted above, rest is really important for NBA players because of the frequency of their games and the toll that each one takes. In our models, we have a column that shows how long it has been since the player’s last game.
A way to integrate environmental and physical factors in player evaluation