Accurately Predicting Minutes in NBA DFS
Minutes are as important to a basketball player in the NBA as touches are to a running back in the NFL. Or, if you are a fan of daily fantasy baseball, they are equivalent to quality at-bats for a baseball player. In daily fantasy basketball, everything is about opportunity and minutes directly provide that for a player. This should be obvious, but there is a strong linear relationship between minutes and fantasy production.
Logically, this makes sense, as the longer a player is on the floor, the more opportunities they are going to have to score points, grab rebounds, dish out assists, and rack up blocks and steals on the defensive end of the floor. Minutes aren’t the only statistic that we should use in our daily research routine, but they are certainly one of the most important. If you can accurately predict minutes, you will be able to accurately predict fantasy production over the course of the season. The key to that sentence is the word accurately. This course is designed to help you get better at predicting minutes.
To be able to predict minutes, we have to understand that there are 48 available minutes for each position, so 240 total minutes for a team. There are basically three main factors that change a team’s rotation throughout the course of the season: injuries, role changes, and blowouts. If we know how to factor each of these into our minutes’ equation, we will have a more accurate prediction of playing time for each player. Predicting minutes may sound difficult, but once you get the hang of it and know what to look for with rotations, it becomes a fairly easy process.
Injuries
I’m not going to spend a ton of time on injuries in this lesson, because I have devoted an entire course to the subject. That course can be found here. However, I will hit on the high points of that course, because injuries affect minutes more than anything else in the NBA.
As with every daily fantasy sport, our ultimate goal is to create a lineup of players that can out-produce their respective salaries. The general rule of thumb in daily fantasy basketball is that we want to have players that can reach 5x or 6x their salary. For instance, if a player is $5,000, then we want them to score at least 25 fantasy points if we are using the 5x method. The best way to find players that can out-produce their salary is by taking advantage of players that are expected to have an increased role due to injuries.
Injuries are very prevalent in the NBA. Every single day, there are going to be players that are ruled out, players that are probable and doubtful, and players that are game-time decisions. Whenever a player (especially a starter) is forced to sit out a game, the rest of the lineup will pick up their minutes. If Kyle Lowry gets injured and misses a game, his 35 minutes are going to get picked up. This provides an opportunity to target players that are expected to see an increased role.
In general, I like to use Rotoworld during the day to monitor injuries and then switch over to Twitter once we get within an hour of lineup lock. The reasoning is that Rotoworld does a nice job of placing all of their injury information in one place (which makes it easy to track during the day). Then, when we get close to lineup lock, Twitter is preferred because this is time-sensitive information. You should also have the RotoGrinders app downloaded on your phone. Make sure to allow push notifications as well. I can’t tell you how many tells I’ve been busy making lineups and was saved by the alert on my phone.
After an injury is announced, there is generally follow-up information, such as whom the starter will be or who is expected to see more minutes. However, sometimes the information is announced so close to lineup lock that we have to do the research on our own. This is where checking depth charts and past game logs comes in handy. For instance, if a team only has two active point guards on the roster and one of them is ruled out, we can immediately project the other to play heavy minutes. We saw this earlier in the season when Damian Lillard and C.J. McCollum were both ruled out for Portland. Tim Frazier ended up playing the entire 48 minutes because the Blazers simply didn’t have any other point guards available that night.
Role Changes
This probably shouldn’t surprise you, but very few teams have the same starting lineup for the entire season. Injuries obviously play a big factor here as well, but players’ roles constantly change throughout the season. The hardest part is identifying an actual role change from an aberration.
Role changes are hard to identify because, often times, game logs can be misleading. A player may see a short-term boost in minutes, but that can be from a number of factors. The biggest role changes are when a player is awarded for his strong play and is inserted into the starting lineup based on merit. As we mentioned earlier, injuries also cause role changes.
Another way to identify role changes is to compare recent minutes, usage, and fantasy production to the player’s season long averages. Being a spreadsheet junkie, I am constantly looking for trends in player statistics. I like to use NBA.com for all of my statistics. Not only are they accurate, but also easily exportable into Excel. Here is the main webpage that I use to get my player statistics:
http://stats.nba.com/league/player/#!/
The best part about this page is that you can sort by any time period that you want. Generally, I like to compare a player’s season averages with their averages over the last five games. To do this, just change the filters in the settings. My reasoning for the five-game sample is that 1) it’s big enough to have merit and, 2) the five-game sample is recent enough to show how well a player has been playing as of late.
In case you can’t find the setting, here is the same page, but with statistics only from the last five games for each player:
http://stats.nba.com/league/player/#!/?Season=2015-16&SeasonType=Regular%20Season&LastNGames=5
Using these two pages, I export the statistics into Excel and create six different columns in my spreadsheet: recent minutes, season minutes, recent usage rate, season usage rate, recent fantasy production, and season fantasy production. We are looking to identify big spikes in all three of these, especially when it comes to minutes. It’s easy for most people to spot an injury or a change in the starting lineup, but this allows us to find trends that most people won’t see.
One example from earlier in the season was Dwight Powell. He quickly established his role on the Mavericks and people didn’t seem to notice until after a couple of weeks into the season. However, if you were to compare his recent production with his prior season production, you would have started utilizing him much earlier.
Before we get into the next section, I want to point out that game logs can be a little misleading. One of the biggest mistakes new DFS players can make is trusting game logs without looking at the reasoning behind why a certain player saw so many minutes. Bench players that see a major spike in minutes in strictly thanks to blowouts aren’t actually seeing a change in their role. Don’t fall for the trap when a bench player has one or two big games but they were the byproduct of blowouts.
Blowouts
The last adjustment that we have to make is for potential blowouts. There isn’t anything worse than feeling great about how your players are performing, only to see them sit out the entire fourth quarter of a blowout. Blowouts are terrible for fantasy production, because if the game is out of hand, both teams typically end up emptying their bench in the fourth quarter.
Trying to predict each and every blowout can be a daunting task, but there are general guidelines that I like to follow. Barring an extreme value play, I try to avoid targeting players on the expected losing end of blowouts. Any game that has a spread of at least eight points qualifies in my eyes, although most people like to use 10-point spreads. Players on the favored team are less risky though, because if their team is ahead, then they probably played well to some extent.
Minutes Equation
Now that we have identified all of the factors that go into predicting a player’s minutes, the question becomes – how do we put it all together? There are a number of ways to predict a player’s minutes and most people don’t like to get as precise as I do. That’s just fine by me though, as I will take any small edge that I can get on the field. Here is my daily process for predicting minutes for each player:
As with any projection, you have to start with a baseline on which you can add or subtract to based on other factors. For my baseline projection, I use the following equation:
(Average minutes per game for the season * 0.75) + (Average minutes over the last five games * 0.25)
This gives us a good place to start, because it takes into consideration both season long and recent averages. The next step is to adjust for injuries. This is more of an art than a science, but you will quickly get used to it. When a starter is injured, the backup that gets the start will obviously see a major boost in minutes and in some cases, so will the other players in the lineup.
The next step is to adjust for potential blowouts. I haven’t found a perfect way to do this, but I feel my method is better than most. Rather than assigning an arbitrary number of, say 10 points for a game to be a potential blowout, I have developed a sliding scale. The scale starts with games that have spreads of at least seven points. For each point higher than that, the minute’s projection is reduced by 1.5%. Here are three examples:
Player 1: 32 minutes projection with an eight point spread = (8-7 = 1 * 1.5%), so multiply his projection by 98.5% = 31.52 minutes
Player 2:* minutes projection with an ten point spread = (10-7 = 3 * 1.5%), so multiply his projection by 95.5% = 30.56 minutes
Player 3:* minutes projection with an 14 point spread = (14-7 = 1 * 10.5%), so multiply his projection by 89.5% = 28.64 minutes
I’ve found that this method works well because the larger the spread, the lower the minutes projection becomes for each player.
After I’ve adjusted each player’s minutes for injuries and potential blowouts, I go through each player’s projection and make sure it passes the eye test. Generally, there will be a few adjustments based on matchups. For instance, if LeBron James is playing against the Warriors, we should expect him to see a couple extra minutes in what is expected to be a close game throughout.
The more you practice it, the better that you will become at predicting minutes and rotations. While many people just like to use minute projections posted on any number of fantasy sites, you will be surprised how much better you will be at spotting value on a nightly basis.