Our Blog


Predicting Multi-Touchdown Games in NFL DFS, Part 1

Introduction

In daily fantasy football tournaments, you need upside at every position of your lineup. But how is upside defined? Is it surpassing expectations — a positive Plus/Minus? Is it doubling expectations — as our Upside metric defines it? Or is it even more nuanced than that?

Football is largely an event-based sport, and fantasy contests accentuate that: In NBA DFS, you can take down a tournament with players who do a little of everything. In football, you need touchdowns. And you need lots of them.

The question is this: Just how predictable are multi-touchdown games? That is exactly what this three-part study is set to find out.

Some Details

To clarify, we are talking solely about rushing and receiving touchdowns. There have been 324 multi-TD games in the last couple of years — 320 if you exclude QBs from the sample, which we will. In order to predict multi-TD games, we need to know prior information. As such, I took every multi-TD game in our sample and, using the FantasyLabs database, looked at the data before the game started.

The first part of this series will look at projections and FantasyLabs-unique metrics. I want to know what the average salaries, projected points, projected ceilings, projected floors, Projected Plus/Minus values, Pro Trends, Bargain Ratings, and Opponent Plus/Minus values were for a player going into his multi-TD game. Hopefully we will see some trends emerge that can help us identify potential situations to target in the future.

Parts 2 and 3 will look at Vegas data (totals, line movement, betting percentages, etc.) and more traditional football metrics (rushes, receptions, yards per target, etc.). I might even throw in some bonus data like weather.

Here’s Part 1: Projections and FantasyLabs-unique metrics.

Salary

Here are the average salaries for all players with multi-TD games.

avgsalarymultifinal

Overall, the salaries for WRs were the highest across the two sites. In general, if you want multi-TD upside, you have to roster the pass-catching studs. Looking through the entire data set really accentuates this. In the sample are some instances of cheap receivers, but many of those instances are of young receivers who 1) were underpriced at the beginning of their careers and 2) would become more expensive relatively quickly. Guys like Martavis Bryant and Odell Beckham had multiple multi-TD games priced under $4,000 at DraftKings. They’ve also had multi-TD games as expensive wide receivers. Beckham has had six in his pro career, ranging from $3,800 to $9,100 at DK.

There is a big natural difference between running backs and pass catchers: The latter aren’t guaranteed volume. Whereas running backs get the ball in the backfield — most of the time before being touched by the defense — pass catchers must 1) get open and 2) have the ball thrown their way accurately. It is much easier to have opportunity as a running back than as a pass catcher. As such, talent evaluation is much more important for wide receivers and tight ends. For running backs, you can largely follow opportunity. For the pass catchers, you probably want to identify the studs and freaks and roster them as much as possible.

The relative rarity of multi-TD wide receivers is seen in this graph showing the percentage of unique multi-TD performances. (That is, multi-TD performances by wide receivers who accomplished the feat only once in the data set.) It is much lower for wide receivers than for running backs, suggesting that talent is much more important for the former than the latter. The same studs have the big games, over and over again:

uniqueperformancesmulti

And when you look at the breakdown of how often these multi-TD games have occurred for each position, it further shows how important the great wide receivers are:

games1

And, just for fun, here’s the distribution of these games by NFL week.

week1

A couple of items to note: The non-bye weeks (when more teams play) on average have more multi-TD performances. That makes sense. Also, Week 17 sees the fewest multi-TD games. That also makes sense, as most teams rest players and thus give fewer players on the whole the requisite snaps needed to have a decent chance of getting at least two touchdowns in the first place.

Projections

Here were the average projections, ceilings, and floors for players going into their multi-TD games.

dkproj1

fdproj1

This data allows us to have a sense of pre-game expectations through a perspective other than salary. Although we might generally expect players with lower salaries not to have multi-TD potential, that is not always the case. Players from a wide range of salaries and projections have recorded multi-TD games. Here are the marks for running backs:

dkrbsal1

Within our data set, we don’t see a lot of low-salaried running backs projected for a lot of fantasy points. This might lead you to believe that the spot-start running backs — the cheap ones filling in for an injured lead back — are overrated in terms of multi-TD upside. However, in order to measure that, we need to look at the whole sample of running backs from the last couple of years. While there aren’t a lot of players in our sample with low salaries and high projections, perhaps that is just a function of their being relatively rare in the first place.

To measure this, we can look at the entire two-year sample of all running backs projected to score at least 10 DraftKings points and compare that to the multi-TD data:

RBMulti3

Here’s the percentage difference of the graph above:

RBMulti4

This data shows that those players — the cheap running backs with high projections — definitely have high upside in terms of multi-TD potential. In fact, they have the highest percentage difference between the overall RB sample and the multi-TD sample. Basically, they outperform expectations (in terms of multi-TD games) more than any other group does.

The upper ranges are still positive as you can see. This likely coincides with our salary data above: In general, players with higher salaries (which are likely proxies for talent and matchup) have more multi-TD upside. The negative middle ranges are interesting, too. The data could be hinting here that players with good situations and volume potential do well regardless of salary. And if you think about it, you’ll probably find most of those guys in the high range (obviously) and low range (guys filling in for those higher range guys who become injured). The middle range likely consists of either mediocre running backs or ones in less-than-advantageous situations, such as a team that doesn’t run often in the red zone. If you want multi-TD upside, target the extremes with running backs.

Bargain Rating

bargain1

A couple weeks ago, I looked at Bargain Rating, a FantasyLabs-unique metric that identifies value by measuring the difference between a player’s FanDuel and DraftKings salaries. What I found was that in MLB, the sites price players fairly similarly. In NBA, DraftKings has historically had tighter pricing. And, in NFL, FanDuel has historically been tighter in its pricing. This . . . does not show here at all.

The study linked showed that players in the top 75 percent of Bargain Rating at DraftKings have a much higher Plus/Minus than their counterparts in the top 75 percent of Bargain Rating at FanDuel. So if players with high DK Bargain Ratings are more valuable overall, why do multi-TD players have higher Bargain Ratings on FD?

I’m not sure this question is 100 percent answerable, but a theory is that on DraftKings we might find value via Bargain Rating — we might find a larger percentage of players who exceed their salary-based expectations — but on FanDuel the metric might be a greater signal of upside. In general, Bargain Rating is intended to help us measure relative value and to determine on which site we should gain exposure to players — it has never been intended to be a measure of upside — but it’s hard to ignore the fact that FanDuel (relative to DraftKings) has consistently priced down players (especially tight ends) when they’ve had multi-TD success.

Pro Trends and Opponent Plus/Minus

protrends1

opp1

Be careful with Pro Trends: Not all of them are created equally. They are, however, a good general way to gauge the overall situation of a player: The more Pro Trends a player has, the better his situation. Interestingly, tight ends have the most average Pro Trends going into multi-TD games. I think this ties in directly with the Opponent Plus/Minus data.

Opponent Plus/Minus is a stat that shows the points above or below expectation a defense has allowed to a particular position. Naturally, it adjusts for opponent strength. And this is where tight ends come in: The data suggests that matchup is much more important for tight ends than other positions in terms of multi-TD games. The average Opponent Plus/Minus — positive is good for the offensive player — is nearly double for tight ends.

From an ownership perspective, this information is likely has implications.

Ownership, especially in large-field guaranteed prize pools, typically follows matchups. A quarterback facing the Saints last year was always incredibly high-owned, regardless of who he was. In general, if we roster tight ends in obviously good matchups, we might be rostering players who are likely to have high ownership. That’s unfortunate, but if you want a multi-TD game from a tight end, increased ownership might just be something you need to accept. You can always differentiate your lineups at the running back and wide receiver positions, which is convenient since matchups (though important) aren’t nearly as important for those positions as they are for tight ends.

All three positions had positive Opponent Plus/Minus values, showing that 1) it is a predictive metric and 2) matchups still matter, even if the degree to which they matter varies. However, running backs in particular have shown that they do not need to be in vastly superior situation to record multi-TD games. In general, matchups are just less important for running backs, probably because (as mentioned in the ‘Salary’ section above) volume determines so much of their production.

Wrapping Up

We’ll stop here for now, but I do want to leave you with a couple takeaways. (I’ll break down what we learned overall in depth in Part 3.)

– Emphasize volume over price or matchup for running backs.

– Emphasize talent over all else for wide receivers.

– While high DK Bargain Ratings typically lead to value, high FD Bargain Ratings are a better indicator of multi-TD upside.

– Focus on matchups more with tight ends than running backs or wide receivers.

– Potentially fade running backs and wide receivers with great matchups; or, at least, don’t be scared to roster those positions even when they don’t have great matchups.

Introduction

In daily fantasy football tournaments, you need upside at every position of your lineup. But how is upside defined? Is it surpassing expectations — a positive Plus/Minus? Is it doubling expectations — as our Upside metric defines it? Or is it even more nuanced than that?

Football is largely an event-based sport, and fantasy contests accentuate that: In NBA DFS, you can take down a tournament with players who do a little of everything. In football, you need touchdowns. And you need lots of them.

The question is this: Just how predictable are multi-touchdown games? That is exactly what this three-part study is set to find out.

Some Details

To clarify, we are talking solely about rushing and receiving touchdowns. There have been 324 multi-TD games in the last couple of years — 320 if you exclude QBs from the sample, which we will. In order to predict multi-TD games, we need to know prior information. As such, I took every multi-TD game in our sample and, using the FantasyLabs database, looked at the data before the game started.

The first part of this series will look at projections and FantasyLabs-unique metrics. I want to know what the average salaries, projected points, projected ceilings, projected floors, Projected Plus/Minus values, Pro Trends, Bargain Ratings, and Opponent Plus/Minus values were for a player going into his multi-TD game. Hopefully we will see some trends emerge that can help us identify potential situations to target in the future.

Parts 2 and 3 will look at Vegas data (totals, line movement, betting percentages, etc.) and more traditional football metrics (rushes, receptions, yards per target, etc.). I might even throw in some bonus data like weather.

Here’s Part 1: Projections and FantasyLabs-unique metrics.

Salary

Here are the average salaries for all players with multi-TD games.

avgsalarymultifinal

Overall, the salaries for WRs were the highest across the two sites. In general, if you want multi-TD upside, you have to roster the pass-catching studs. Looking through the entire data set really accentuates this. In the sample are some instances of cheap receivers, but many of those instances are of young receivers who 1) were underpriced at the beginning of their careers and 2) would become more expensive relatively quickly. Guys like Martavis Bryant and Odell Beckham had multiple multi-TD games priced under $4,000 at DraftKings. They’ve also had multi-TD games as expensive wide receivers. Beckham has had six in his pro career, ranging from $3,800 to $9,100 at DK.

There is a big natural difference between running backs and pass catchers: The latter aren’t guaranteed volume. Whereas running backs get the ball in the backfield — most of the time before being touched by the defense — pass catchers must 1) get open and 2) have the ball thrown their way accurately. It is much easier to have opportunity as a running back than as a pass catcher. As such, talent evaluation is much more important for wide receivers and tight ends. For running backs, you can largely follow opportunity. For the pass catchers, you probably want to identify the studs and freaks and roster them as much as possible.

The relative rarity of multi-TD wide receivers is seen in this graph showing the percentage of unique multi-TD performances. (That is, multi-TD performances by wide receivers who accomplished the feat only once in the data set.) It is much lower for wide receivers than for running backs, suggesting that talent is much more important for the former than the latter. The same studs have the big games, over and over again:

uniqueperformancesmulti

And when you look at the breakdown of how often these multi-TD games have occurred for each position, it further shows how important the great wide receivers are:

games1

And, just for fun, here’s the distribution of these games by NFL week.

week1

A couple of items to note: The non-bye weeks (when more teams play) on average have more multi-TD performances. That makes sense. Also, Week 17 sees the fewest multi-TD games. That also makes sense, as most teams rest players and thus give fewer players on the whole the requisite snaps needed to have a decent chance of getting at least two touchdowns in the first place.

Projections

Here were the average projections, ceilings, and floors for players going into their multi-TD games.

dkproj1

fdproj1

This data allows us to have a sense of pre-game expectations through a perspective other than salary. Although we might generally expect players with lower salaries not to have multi-TD potential, that is not always the case. Players from a wide range of salaries and projections have recorded multi-TD games. Here are the marks for running backs:

dkrbsal1

Within our data set, we don’t see a lot of low-salaried running backs projected for a lot of fantasy points. This might lead you to believe that the spot-start running backs — the cheap ones filling in for an injured lead back — are overrated in terms of multi-TD upside. However, in order to measure that, we need to look at the whole sample of running backs from the last couple of years. While there aren’t a lot of players in our sample with low salaries and high projections, perhaps that is just a function of their being relatively rare in the first place.

To measure this, we can look at the entire two-year sample of all running backs projected to score at least 10 DraftKings points and compare that to the multi-TD data:

RBMulti3

Here’s the percentage difference of the graph above:

RBMulti4

This data shows that those players — the cheap running backs with high projections — definitely have high upside in terms of multi-TD potential. In fact, they have the highest percentage difference between the overall RB sample and the multi-TD sample. Basically, they outperform expectations (in terms of multi-TD games) more than any other group does.

The upper ranges are still positive as you can see. This likely coincides with our salary data above: In general, players with higher salaries (which are likely proxies for talent and matchup) have more multi-TD upside. The negative middle ranges are interesting, too. The data could be hinting here that players with good situations and volume potential do well regardless of salary. And if you think about it, you’ll probably find most of those guys in the high range (obviously) and low range (guys filling in for those higher range guys who become injured). The middle range likely consists of either mediocre running backs or ones in less-than-advantageous situations, such as a team that doesn’t run often in the red zone. If you want multi-TD upside, target the extremes with running backs.

Bargain Rating

bargain1

A couple weeks ago, I looked at Bargain Rating, a FantasyLabs-unique metric that identifies value by measuring the difference between a player’s FanDuel and DraftKings salaries. What I found was that in MLB, the sites price players fairly similarly. In NBA, DraftKings has historically had tighter pricing. And, in NFL, FanDuel has historically been tighter in its pricing. This . . . does not show here at all.

The study linked showed that players in the top 75 percent of Bargain Rating at DraftKings have a much higher Plus/Minus than their counterparts in the top 75 percent of Bargain Rating at FanDuel. So if players with high DK Bargain Ratings are more valuable overall, why do multi-TD players have higher Bargain Ratings on FD?

I’m not sure this question is 100 percent answerable, but a theory is that on DraftKings we might find value via Bargain Rating — we might find a larger percentage of players who exceed their salary-based expectations — but on FanDuel the metric might be a greater signal of upside. In general, Bargain Rating is intended to help us measure relative value and to determine on which site we should gain exposure to players — it has never been intended to be a measure of upside — but it’s hard to ignore the fact that FanDuel (relative to DraftKings) has consistently priced down players (especially tight ends) when they’ve had multi-TD success.

Pro Trends and Opponent Plus/Minus

protrends1

opp1

Be careful with Pro Trends: Not all of them are created equally. They are, however, a good general way to gauge the overall situation of a player: The more Pro Trends a player has, the better his situation. Interestingly, tight ends have the most average Pro Trends going into multi-TD games. I think this ties in directly with the Opponent Plus/Minus data.

Opponent Plus/Minus is a stat that shows the points above or below expectation a defense has allowed to a particular position. Naturally, it adjusts for opponent strength. And this is where tight ends come in: The data suggests that matchup is much more important for tight ends than other positions in terms of multi-TD games. The average Opponent Plus/Minus — positive is good for the offensive player — is nearly double for tight ends.

From an ownership perspective, this information is likely has implications.

Ownership, especially in large-field guaranteed prize pools, typically follows matchups. A quarterback facing the Saints last year was always incredibly high-owned, regardless of who he was. In general, if we roster tight ends in obviously good matchups, we might be rostering players who are likely to have high ownership. That’s unfortunate, but if you want a multi-TD game from a tight end, increased ownership might just be something you need to accept. You can always differentiate your lineups at the running back and wide receiver positions, which is convenient since matchups (though important) aren’t nearly as important for those positions as they are for tight ends.

All three positions had positive Opponent Plus/Minus values, showing that 1) it is a predictive metric and 2) matchups still matter, even if the degree to which they matter varies. However, running backs in particular have shown that they do not need to be in vastly superior situation to record multi-TD games. In general, matchups are just less important for running backs, probably because (as mentioned in the ‘Salary’ section above) volume determines so much of their production.

Wrapping Up

We’ll stop here for now, but I do want to leave you with a couple takeaways. (I’ll break down what we learned overall in depth in Part 3.)

– Emphasize volume over price or matchup for running backs.

– Emphasize talent over all else for wide receivers.

– While high DK Bargain Ratings typically lead to value, high FD Bargain Ratings are a better indicator of multi-TD upside.

– Focus on matchups more with tight ends than running backs or wide receivers.

– Potentially fade running backs and wide receivers with great matchups; or, at least, don’t be scared to roster those positions even when they don’t have great matchups.