top of page
Writer's pictureAlex Cates

2020 Auction Draft Strategy: What Worked?

Updated: Sep 4, 2022

If you've played fantasy football for a long time you have probably heard of an auction draft. While the traditional snake draft means everyone takes turns choosing a player to add to their team, auction drafts embrace a capitalistic approach. You want Patrick Mahomes on your team? Pay up for him! Want Christian McCaffrey? He can be yours. Since every manager has the same budget, everyone has the same opportunity to get the players they want, so the question is, how should you spend your money?


Today we are going to look at what spending strategies worked last season and which didn't. Should you blow all your money on 3 players and then hope to mine the waivers? Should you wait and draft for value on every pick? Should you pay up for a QB or a TE? Today we will find out.


Note: Same as with my snake draft analysis, this data will be specific to the 2020 results. This is obviously a small sample size and should be treated as such, that doesn't mean that any conclusions should be dismissed, but they should be taken with a grain of salt. I have an updated version which looks at 3 years of auction drafts here.


The Data

We will be using the results from ESPN public leagues (as with many of my posts). In this case, Auction drafts are a little less popular so we will have the results of ~500 leagues leading to ~5000 teams. I normalized all prices to a $200 budget league which is how all my analysis will be framed here. In all cases, we will be comparing how their draft decisions affected their odds of winning the championship. This obviously ignores all the decisions during the season that make a significant impact on how your team does but it will provide insight into how to set up your team for success. Finally, I binned the price data such that all prices or values are rounded to the nearest $5. This is just to make sure the championship odds assessment has enough data to be accurate. Now let's look at what draft strategies worked.



How To Divide Your Budget Between Picks?

The first question we will look at is how much should you spend on your top players. A common strategy is stars and scrubs, spending up on a couple of elite guys, and then mining the waiver wire to fill in the rest of your lineup, but is that the best?


Scatter plots and trend line of the total money spent on the top x most expensive players. The redline represents random chance. The shaded area represents the 95% confidence interval of the trend line


For each graph, we can see how much someone spent in total on their top X most expensive player. Anywhere the trendline (blue) is above the random chance line(red) we know we are helping our team. If we focus in on the peak, we see an interesting trend. The best odds seem to be taking a balanced approach, such that the best players spent $50-$60 on their top player, $90 on their top 2, $120ish on their top 3, $130 on top 4, $140 on top 5, and $150 on top 6.


Working iteratively we can create a pricing chart for individual picks like this:

I've always thought of this approach as following a snake draft in an auction draft. Essentially you are picking a 1st round player ($55ish), a 2nd round player ($40ish), a 3rd round player ($30ish), and so on. You just get your pick of players in each round. This isn't really stars and scrubs (at least not the extreme version) but still has you allocating most of your budget (80%) on your top 6 picks.


A second interesting note from the above graphs comes with the top 4 and top 5 picks. Here the graph seems to look bimodal, with a peak at 110 and 150 for the top 4 and a similar peak at 110 or 170 for the top 5. I am not sure if this is just a quirk of the small sample set or a trend of 2 divergent strategies (likely stars and scrubs vs balanced value), we will have to see how these trends hold for 2021. Note I did try fitting a 4th order polynomial, but it didn't change the fit much due to the limited dataset.



How Much To Spend On Each Position?

Another common question is how much to spend per position, while this will almost certainly depend on the year and the strategy, seeing how money was spent on the presumed starters affect championship odds. For the following analysis, I am assuming a traditional lineup, 1 QB, 2 RBs, 2 WRs, 1 TE, 1 Flex.


Quarterback

Scatter plot with regression line on price spent on QB vs championship odds. The red line represents odds due to random chance.


The QB chart is the noisiest we will see. In the end, I think that means that it comes down to if you hit on your QB or not. There are high championship odds for cheap QBs, a $25 QB, and a $40 QB. Unsurprisingly, you should not go higher than that, but the high odds at $40 surprise me a little and push back a little on the focus on late-round QB. All that said, the trend definitely supports spending your money on non-QB positions


Runningbacks

Scatter plots with regression lines on price spent on RB1 (left) and RB2 (right) vs championship odds. The red line represents odds due to random chance.


Unsurprisingly you want to get a stud RB and preferably 2. Really this aligns with the pick allocation above such that you should spend in the 50's to get a top RB and then get a solid rb2 in the 30s. I also think the blip of a second $50 RB is interesting as it supports committing to a stars and scrubs strategy.


Wide Receivers

Scatter plots with regression lines on price spent on WR1 (left) and WR2 (right) vs championship odds. The red line represents odds due to random chance.


WRs is also noisy, but at some level, it fills in the gaps of the RBs. Specifically with the WR1 costing in the 30's/40s or being cheap (10s) while the WR2 should be in the 10s. This fits around the RB chart above to align with the pick pricing we first looked at. The only interesting blip is the high championship odds for a $40 WR2. I wonder if this is a zero-rb strategy, but I can't say for sure.


Tightend

Scatter plot with regression line on price spent on TE vs championship odds. The red line represents odds due to random chance.


With TE we see conventional wisdom coming forward. Basically, you either want to spend up for a top TE ($30-40) or wait for a cheap one (0-10) while getting the one you want. I think an interesting trend will be if those top TEs shift farther to the right (become more expensive), especially considering the current price for Travis Kelce ($53 on ESPN as of today)


Flex (RB, WR, or TE)

Scatter plot with regression line on price spent on the Flex slot vs championship odds. The red line represents odds due to random chance.


Those of you that have done auction drafts know that the flex position following an auction draft can have some odd results. There's always that person that grabs a 3rd RB or WR for top dollar such that the flex slot is their 3rd best player. Interestingly, this graph seems to support that strategy with a $40 flex player being a solid choice. Additionally, the more common strategy of having a weaker (but still starting) player in the flex position (a $10 player) works as well, aligning better with the pick analysis we started with.


Putting it all together, a successful breakdown of positions and picks last year may have looked like the following:


Now there are clearly other successful strategies as shown above, but this would be a pretty traditional breakdown that aligns with the pick analysis above.

 

 

Drafting For Value

One of the most common suggestions you hear in auction drafts is focusing on value. Here, you eschew targeting specific players or positions and instead focus on drafting players who are available for less than their average price. In this case, I am defining value as the price you paid for a player compared to the average auction value of that player in the entire dataset. So if a player normally goes for $40 and you got him for $35, that is $5 worth of value. We can assess the total value gained by a team to see how that impacted their championship odds.

Scatter plots with regression lines on total value generated among the top 3 (left), top 6 (middle), and all picks (right) vs championship odds. The red line represents odds due to random chance.


Unsurprisingly, we see that gaining positive value improves your odds of winning the championship. More surprising to me is how little the value impacts your odds. For each dollar saved on your first 3 picks, you increase your odds of winning by only 0.07% and this only goes down as you add more players into the mix. I think more interesting is that negative values (-10 for top 3, -20 for top 6, and -33 for total) are the break-even points in terms of odds of a championship.


My takeaway is that similar to snake drafting, you should try for value, but don't be afraid to pay up a little for your guys if you want. Just be careful not to get sucked into a bidding war.


Limitations

The main and obvious limitation has to do with the underlying data. First, it is only looking at 2020. As we know 2020 was a weird year for many reasons and it is entirely reasonable to wonder if this data will hold for 2021, we will have to wait and see. A second limitation is the number of auction drafts. ESPN defaults to snake drafts and while I had 500 leagues in this dataset, that is nothing compared to the few thousand leagues that made up the snake draft analysis. Given this, please do take these results with a grain of salt, but I think some of the conclusions will hold and are interesting.


Conclusions

In the end, the main conclusion is that you should try to mirror a snake draft composition when forming your team, striking a balance across positions and prices. You should focus on getting 2 good RBs, though there may be opportunities for other strategies if need be. Finally, you should try to draft for value but not at the cost of getting the stud players you want.


Questions? Comments? Let me know at ac@alexcates.com. Want to read more breakdowns like this? sign up for my newsletter. Finally, like what I do? Consider supporting me on buy me a coffee.

7,094 views

Recent Posts

See All

Comments


bottom of page