Beyond our fantasy football content, be sure to check out our award-winning slate of Fantasy Football Tools as you navigate your season. From our Trade Analyzer – which allows you to instantly find out if a trade offer benefits you or your opponent – to our Waiver Wire Assistant – that allows you to quickly see which available players will improve your team, and by how much – we’ve got you covered this fantasy football season.

In this post, we will look at last season’s box scores and DraftKings salaries to explore the relationship between salary and points scored. In this post, we did the same thing for pitchers. We should expect salary to be a better predictor of points scored on the pitching side compared to the hitting side since there is much more randomness involved in four or five plate appearances compared to a starting pitcher outing.

Get expert advice during your draft with our fantasy baseball draft software

The Data

To do this type of analysis, I had to compile a huge data set of every single box score from last season along with each player’s salary and DraftKings points total. I now have one giant table that has all of this information. Here is a quick subsection to show you what we’re working with:

These are Mike Trout’s first ten games of the 2019 season, you can see we have his salary and DK Points there on the right.

When we look at the distribution of DraftKings points scored on the hitting side as a whole, it looks like this:

I took out all pitchers and pinch hitters before running that chart. You can see the most common points total is zero, with more than 12,000 starting hitters scoring less than two DraftKings points. It is pretty rare to see a hitter go over 20 points, and you have very few occurrences of a hitter going over the 30 mark.

Salary vs. Points Scored, as a Whole

Here is what the plot looks like when we do a scatter of every single point total against the hitter salary:

There are too many data points here to really learn much from, but you can start to see that generally the higher DraftKings points totals go to the higher-priced hitters.

I then aggregated the data to find the average points scored by each price point. Here is that plot:

The far-right gets a little weird because there were so few hitters priced near $6,000. There were only three cases of a hitter being $6,000 – and less than 40 occurrences of hitters priced at every price point above $5600.

I rounded each salary to its nearest $500 mark and ran the plot again to get a more general idea of salary vs. points scored:

You can see that when you look at the data as a whole, it is a pretty darn linear relationship. You can nearly draw a straight line those points, which means the pricing algorithm is doing a good job. At a league-wide level, hitters will score more points on average the higher they are priced.

It is worth taking a look at the specific numbers for this last graph to see what those differences are. I took the average for each price points and subtracted the previous price point from it to see how big of a jump happens when you move up to the next salary bucket:

You can see the biggest jump in average points come for hitters going from $5,500 to $6,000. As we mentioned before, however, there were so few of these $6,000 hitters that we should not take that too seriously. The most interesting insight from this is probably the jump between being a $5,000 hitter (in this case that actually means being between $4,700 and $5,200 since we rounded to the nearest $500) and being a $5,500 hitter. Those hitters averaged 0.8 more points. A similar jump was seen between the $3,500 and $4,000 price point.

I cannot say for sure this will be repeated every season, but considering how much data we have going into this (38,702 data points to be exact) – there is a good likelihood that this is meaningful. The way to apply this to your lineup building would be: if you have a few hundred left over and you want to spend it, look to upgrade your players that are priced around $3,500 and $5,000 first.

Salary vs. Points Scored, Individual Player Level

I performed this part of the analysis using only players that saw 200 or more plate appearances last year (360 different hitters), and I am still only considering players that were in the starting lineup to avoid having averages brought down by games where a player only saw one plate appearance as a pinch hitter.

This was a bit of a convoluted process, so I will do my best to explain it.

Let’s take Nolan Arenado as an example. What I did was find Arenado’s average salary for the season, his average point total for the season, and then looked at his average point productions for all of his different price segments. The results look like this:

In 2019, Arenado’s average DraftKings salary was $5,100. His average DraftKings points output was 7.82. Rounding each of his salaries to the nearest 500, we come up with five different salary buckets for him, from $4,000 up through $6,000. You can also see the frequencies of each salary there. He was only priced at $5,800 or above six times, and in those six games, he averaged 12.8 DraftKings points per game, five points above his season average.

I generated this table for all 360 hitters, and then I found the average change in points for each average change in salary. I limited this to salary changes with frequencies of 10 or more so that +5 for Arenado at the +$900 price point won’t be reflected. This was necessary to avoid the noise of small sample sizes.

Long story short, if the hitter pricing algorithm was perfect, you would see another straight line going up from left to right in the plot. By this, I mean that you would expect to see hitters score fewer points than their average when they are priced below their average, and vice versa.

However, this is not at all what we see:

Starting from the left of the plot, we can see that when hitters were priced $800 or more below their average, they scored significantly under their season averages. This is an extreme occurrence though, being that far below your season average price – so that is a bit subject to small sample size. The rest of the data points come in very close to 0.0 on the y-axis, which makes sense given the volume of data we are looking at here.

The interesting thing to note is the trend. Hitters actually scored above their averages when they were priced below their average, although it’s a very slight difference. Being priced well above their average salary did not mean they would score above their average either, in fact, they were slightly below for most of the positive price changes there.

All of those data points come in close enough to the average there for this not to be evidence that the DraftKings pricing algorithm is all wrong. However, it is pretty clear that paying for a player when he is priced much higher than normal does not come with an advantage – you really cannot expect the player to score more in those situations in general terms.

Let’s take a look at some more individual players:

You can see some different stuff going on with those players. The algorithm pretty much had it right on Khris Davis, who really struggled when he was priced down but scored a ton of points when he was priced up. Brett Gardner, on the other hand, did his best work at his lowest price. You gained a lot by paying the premium price for JD, but Gary Sanchez really disappointed at his highest prices.

If you looked at every player individually you would find all kinds of different things, but the overarching averages show that there is really not much to be gained for paying for a player when he is priced above his usual salary.

Conclusion

We all pay very close attention to pitcher matchups, ballpark environments, and recent performance when selecting hitters for our DFS lineups. You would be silly not to. However, sometimes we forget that the pricing algorithm considers all of this as well – and often times it puts too much weight on those variables, raising the price on a player too much given the situation.

This seems to be the case. Expected ownership is quite possibly baked into the salaries as well, which means they could add on a couple hundred dollars to a price just because they expect the player to be popular – super high ownership on a guy isn’t a great situation so they will try to steer away from that. A lot of DFS players preach the importance of being contrarian and zigging when everyone else is zagging, and the numbers seem to bear this out as a money-making strategy.

Slumps and lack of popularity can be a smart DFS player’s best friend. While everyone else is ignoring a guy because he is facing a tougher than average pitcher and hasn’t had a hit in a few days, the data suggests that it is a great idea to take advantage of that discounted price – because you should not actually expect a much different outcome than you normally would.

Hope you enjoyed this, check back daily for more MLB DFS content as we get ready for the 2020 season!

Practice fast mock drafts with our fantasy baseball software