DFS Strategy: Following Trends & Analyzing Statistics

Following trends and analyzing statistics are important pieces in anyone's daily fantasy sports strategy. Luckily, no one who hates statistics participates in this hobby. After all, statistics are what daily fantasy sports are all about.

On this page, we offer a broad overview of how to follow trends and analyze statistics that can affect winning percentages and profitability.

Tracking Your Own Statistics

Most people reading this page are probably looking for advice about how to analyze trends and statistics for various sports. We'll get to that, but tracking your own statistics is the place to start. If you're not keeping up with your own numbers, you're doomed to fail.

We once read a great book about poker. We came away from the book with only one solid piece of advice, but it changed our thinking forever.

This was a light bulb moment. You can go from being an amateur fish to a serious player by making a single change.

Making that change and tracking results will improve anyone's game dramatically.

The same holds true for fantasy sports. If you want to be a serious player, track your results. Here's why.

The key to profiting consistently is making predictions based on your theories. When those predictions pan out, players can gain confidence that their theories are accurate. More positive results add to a player's level of certainty.

It's easy at most sites to collect data. Major daily fantasy sports sites usually offer the option of downloading your entire history. The spreadsheet they provide usually includes the following data.

• Which contests you entered.
• How well you placed.
• How much was scored.
• How much was won.
• How much the entry fee was.

We like to use these statistics in multiple ways.

Tracking our average scores is what enables to see if we are making a gradual improvement in our scores over time. Eventually this improvement levels off, but everyone should always be trying to improve.

Keep in mind, though, that rule changes can affect scores. Sites change their scoring systems and their salary caps. When they do, the statistics change, too.

A more important statistic to track is win rate. We like to play a lot of head-to-head and 50/50 contests. In fact, we'll even play in a large number of low buy-in contests just to see how strong our lineups are. Anything can happen in a single contest, but when you're looking at the results of 100 or 200 contests a day (or per week, in football), you're starting to get some statistical accuracy.

If your scores are consistently in the bottom half of the field, you're doing something wrong. Figure it out. Make changes to your drafting strategy until you start winning a larger percentage of contests. In daily fantasy sports, the winning percentage for break-even players is 55.56%. Set that as your first target.

A common misconception is that someone has to be winning in order to track their return on investment. That is not correct. Even losing players have a return on investment—a negative return.

The math isn't complicated, as it's just one simple calculation.

The result then needs converting into a percentage.

Here's an example.

This player is a net loser. So his return on investment of -7.5% is a negative number.

If that overall number changes this week to -5%, then he's on the right track. On the other hand, if it goes up to 8% or 9%, he's doing something wrong. Or maybe he's just gotten lucky or unlucky.

That's something else everyone should keep in mind when looking at statistics. In the short term, anything can happen. 72 contests is a small sample size. You're not looking at any kind of statistical accuracy until you start looking at results of 1,000+ games.

The Value of Overlays

An overlay is when there's more money in the prize pool than that generated by the players. Here's how that works.

Daily fantasy sports tournaments with guaranteed prize pools don't always "fill up". Sites count on a certain number of players entering, but when too few players enter, the site covers the difference. Here's an example.

It's easy to calculate what an overlay is actually worth to you. Take the amount of the overlay and divide it by the actual number of entrants. That's the dollar amount of equity that each person has in the contest on top of their entry fee. In this example, the overlay is worth $8.33 per person.

When a player consistently puts himself into overlay situations, he increases his upside. In the example above, if the contest had filled, his chances of winning—assuming everyone were exactly the same skill level—are 472/2298, or about 20.5%. But in this overlay situation, his chances of winning increase to 472/1500, or about 31.5%.

That's a huge difference.

Projecting Players' Fantasy Points

Most readers are probably more interested in knowing how to use statistics and trends to project players' points. In fact, that's a critical aspect of strategy, too.

Take multiple statistics into account when making a projection. For one thing, you need to have an idea of how many points a player scores on average, regardless of competition. You can calculate that as a long-term average, but it's as much an art as it is a science—especially for new players who don't have much historical data.

Also, look at who the player's opponents are. If a player is facing a strong opponent, he's liable to score fewer points than average. On the other hand, if his opponent is weak, he'll probably score more points than you might expect.

Those two factors determine your projections, but the actual score for that player is affected by a third factor, variance.

In any given game, a player might perform better than or worse than expected. That's the random nature of the game. The good news is that this variance is only a short term factor. It matters in tomorrow's game, but over a large number of games, assuming your estimates are correct, your projections will start to near the averages.

One place to look is historical data. In some sports, like baseball, you can look at an average player's performance over the last 50 games or so to get an idea of what you can expect. In other sports, like football, you have to take into account the competition in those historical games—that's because there are so many fewer football games each season.

Another place to get data is from the Vegas sportsbooks. If you know the point spread for a game and who the favorite is, you can project each team's total performance, point-wise. Combine that with the playing tendencies for that team, and you can get an idea of how the strength of their opponents might affect an individual's performance.

Dozens of sites on the web estimate how each player will perform in any given game. You can use those sites as a starting point, but make informed decisions based on your own educated opinion. Otherwise, you'll never be able to get an edge over other players who are using those same sites for their research.

Here's an example.

In this case, Cabrera and Harper are more or less equal, statistically. Cabrera's a little better, but he also costs a little bit more, so it's hard to make a decision based purely on value.

But since Cabrera is playing against the weaker pitcher, he's the obvious choice.

Players Don't Become Due

Something else to keep in mind is that players don't become due for a good game just because they've had poor performance in the previous game or games. By the same token, just because a player is on a losing streak doesn't imply that he's a bad choice.

Don't count on variance to help you make your predictions. Variance is just the level of randomness in the game. Over time, players are going to perform in a statistically predictable manner. But narrowing that down into how a player is going to do in a specific game is folly.

On the other hand, situations can change what you expect.

Here's an example.

An NFL team suffers a humiliating loss on the road. This week, they're playing at home versus a reasonably weak opponent. They have something to prove. Their home game advantage takes some of this into consideration, but not the "something to prove" part.

One might use this data to inform his decision on whether or not to take that team's RB1 versus another team's.

These are the kinds of situations which take into account player psychology. They're not the same thing as assuming that a team is due for a win because they lost last week.