Running Back Draft Strategy:
By: Nick Zylak
A complete record of success by year since 2015 can be found by clicking here
This is the first of the positional draft strategy articles that I’ll be writing, so I’ll begin by going over the data. If you don’t care about how the data was collected and what that means for our understanding, then you can skip ahead to “When to draft your first RB”.
How did I get the data?
It was miserable. If any of you are from a data science background, you’ll know that data discovery and cleaning typically takes 80% of your time…with the other 20% dedicated to the actual analysis of the data. I’ve been on project teams where that split is 60/40 and others where it’s 90/10…but typically an 80/20 split should be expected. I would estimate that I spent 97% of my time on this project (which took nearly every day from January-April) on data discovery and cleaning. The analysis is the easy (and fun!) part that I’ll share with you through a number of different articles. But oh my goodness was the data cleaning a nightmare. Here’s why…
Underdog is an absolutely incredible platform and their commitment to data transparency is unparalleled. If every company was run as well as them, the world would be so much better off. Sadly…none are. And when I say none, I mean none. It is almost impossible to find historical draft data from any other platform. These companies (ESPN, Yahoo, NFL, Sleeper) do not make draft data available for the general public. And if you ask for it (and are even willing to pay for it) you’ll be met with radio silence. These companies DO NOT GIVE UP THEIR DATA…so hats off to Underdog for being the best!
Unfortunately, Underdog is a best ball platform, and while I highly recommend everyone uses their platform, I know that the vast majority of you only play in redraft leagues. So while I have written analysis on the optimal Underdog draft strategy, I will not be using that data for redraft.
So what data will I be using? I spent this offseason building and running code to scrape ESPN public leagues over the last 7 years. I collected draft and standings data into a beautifully formatted dataset. I did my absolute best to removed drafts that:
Had incomplete data:
This was the easiest data to remove. If the league only pulled 3 rounds of draft data then I’m obviously not going to use it for analysis. Other issues could be not pulling the data correctly from 1-2 teams. Again, these leagues are easily removed from the dataset.
Were clearly from keeper leagues:
This one is also fairly easy, and it ties into the final removal you’ll see. It’s pretty clear sometimes when leagues are keepers because everyone gets insane value while also skipping over stud players early. It’s not good to include keeper leagues because your draft strategy will change depending on who you keep…and also not everyone starts on the same playing field. In a typical draft everyone starts at zero. In some keeper leagues people started off with Davante Adams and Jonathan Taylor. Kind of tough to evaluate a draft strategy when one team could win drafting 4 QB’s to start.
Were from in-person drafts where the order is clearly incorrect:
ESPN is trash at accounting for in-person drafts. I am genuinely shocked at how little they’ve done to improve this over the years. Plenty of people do online drafts and yet many leagues (my league included) are forced to use third-party software to conduct and upload the draft to ESPN. Because of this, offline drafts have a high rate of inaccurate data. Basically, the order of the draft is all jumbled up. That doesn’t help us at all, so I’ve done my best to remove these leagues.
Are from two QB leagues:
I know a lot of you play in 2 QB or SuperFlex leagues. However, using two different formats would skew the data. You can’t evaluate strategy when some leagues are heavily weighted towards one position and some aren’t. I figured the best thing to do would be to remove these leagues, conduct the research, and then circle back to think about how things might change if we transitioned to SuperFlex/2 QB leagues (also side note – These are the same thing. Sometimes people ask if I have 2 QB rankings as opposed to SuperFlex. If you’re not using a QB in the SuperFlex spot then you’re playing with one arm tied behind your back. I assume that everyone is starting 2 QB’s in my SuperFlex rankings and you should as well. If you run into teams that don’t think this way, then just be thankful for the edge you’re gaining over them).
Are TE Premium Leagues:
Same idea as above. I know a lot of you like these leagues, but they skew the data. To remove these leagues I just counted all the tight ends taken in each draft. If a billion were taken then it’s obviously a TE premium league.
Had impossibly high value levels:
This ties into offline drafts and keeper leagues. I just removed leagues where people got insane values. And I’m not talking 2-3 rounds of value, I mean like CMC went in the 10th…or a 15th round pick was taken in the first. Just getting lucky did not remove your league from the dataset. I did this by importing historical ADP and matching that to every draft.
Where teams clearly traded draft picks:
It’s tough to evaluate a draft strategy when someone gets three first round picks, and another team gets 0! This is impossible to eliminate entirely, but I ensured that everyone had the same number of picks and that everyone only had one pick in each round. This honestly only comes into play with keeper and offline drafts…so many of the leagues where picks were traded had already been eliminated in prior steps.
So how much data did I collect?
My computer was running nonstop from January through the beginning of April. Once a run was complete the data was saved, I would begin a new run, and repeat the process. This was a tough process to perfect and I did lose a lot of time figuring out the most efficient way to gather and save the data. After all the cleaning and filtering out of leagues, I was left with over 110,000 teams for each of the 7 years I gathered data for. Some years have more data than others, but the lowest total is 110,000. The dataset that I use for this analysis includes a random sample of 110,000 teams from each season. That’s just done to make sure we don’t skew towards one season. As you’ll see in different articles/tables, some years are wildly different than others. 2015 was the death of the RB position, whereas RB was necessary to win in 2016. Using an uneven amount of data from one of these years would give us inaccurate results.
This means that we have a random sample of 770,000 teams from 2015 through 2021, with an equal representation in each year! I hope to continue this study next season, and since the code is already written…I think it’s possible I can reach a sample of over 2 million teams next year!
Is 770,000 teams a large enough sample?
In short, yes. But barely if we’re being honest. As I collected the data I was attempting to answer this question. I would take random samples of the data and track if the numbers were stabilizing. Once we hit around half a million teams the data started to really stabilize. No matter what sample was taken, success rates of different strategies remained relatively constant. At 770,000 the results seem to be very consistent, with success rates barely moving between samples. Obviously every data analyst always wants more data. That’s a given, and I’ll be pumped if I can expand the dataset to 8 years and over 2 million records next season. But for now, I do believe that the results can be trusted with 770,000 teams.
So! Now that you know where the data came from…what can we learn?
When to draft your first RB:
This is the first question that popped into my head, and I’m sure that’s the same for many of you. What round should we be drafting our first RB? Or better asked – What’s the playoff and championship rate of teams who’s first RB came in each respective round? That’s answered through these first two graphics. The green line is the average for all teams among the sample.
You might be surprised that nearly every round is below the average. Don’t be! Understand that the vast majority of teams take at least one RB in the first three rounds, so even though the later rounds are well below average…there are less teams that fall into these buckets (but still a large enough sample to maintain consistent results). The playoff and championship rates of teams that take their first RB in round 1 doesn’t have to be much higher than average to offset the later rounds. Looking at these graphics makes it clear that you should be targeting RB early. Teams that drafted their first RB in round 1 had the highest average playoff and championship rates. Teams that took their first RB in round two had the second most success – performing average overall. Teams that drafted their first RB in round 3 had the third most success – performing slightly below average overall.
Starting in round 4 the results seems to flatline. It does become worse and worse as you continue to avoid drafting RB…but the differences are minor and are likely not impactful. This gives credence to the idea of the RB dead zone. After the first three rounds, RB talent historically falls off. So if you’re taking your first RB in round 4+… you likely just missed out on the elite options… and might be hurting your team by forcing a pick at the RB position (while also opening up value for your opponents who not only took RB early… but now get to draft other positions at optimal times…while you force a pick based off positional need).
But What About Value?
Now…this is holding everything else equal, but as I’m sure you know – every draft is unique. You might have every intention of taking RB in the first or second rounds…but sometimes the value is terrible! So when we’re answering the question of when to draft your first RB, value has to come into play, right? What if we’re forced to reach for RB early? Or what if a WR becomes an incredible value? Do the numbers still hold?
This graphic might give us some level of insight into the answer! Column 1 indicates a RB’s value compared to ADP. Lucky enough to draft the 1.01 at pick 12? Congrats on 11 spots of value (12-1)! Decided to reach on the 12th ranked player at pick 1? Well you get -11 spots of value (1-12)!
Column 2 compares the playoff rate of teams that drafted the respective value level at RB in round 1, to the average playoff rate of teams that skipped RB in round 1, opting to take their first RB in round 2 or 3. Why those two rounds? Because we determined earlier that it’s definitely optimal to get RB early…with a cliff in success when you leave the 3rd round without one. And since we’re trying to determine when NOT to take a RB in round 1…I’m assuming that RB early is still in the mind of whoever’s drafting.
Finally, column 3 is the exact same as column 2, just comparing championship rates instead.
There’s of course going to be randomness, especially when breaking off data into smaller samples…but it’s very interesting that we seem to get a distinct cut off at -2 spots of value for both Playoff and Championship rates. Once a team reached 2+ spots for a RB in the first round, they performed worse overall than teams that drafted their first RB in rounds 2 or 3. Is this a perfect way of looking at things? No…nothing is perfect. But this does prove that you can’t just blindly draft RB just because the data says you should! I also think it’s worth noting that the data makes sense logically. ADP’s are extremely efficient in the first few rounds. So when you start reaching too much in the first round, you’re opening up clear value for your opponents, while simultaneously drafting someone that the community deems less valuable. I do not view reaching 2 spots as a hard cutoff…but rather a general idea that we should be aware of. It should be in our minds, but not something we strictly follow. At the end of the day, if a RB is a value in the first round, DRAFT THEM. If not, you can still draft them, but it’s more than ok to wait and take your first RB in round 2 or 3. Just understand that the longer you wait, the worse off your team becomes historically.
What About Your Second RB?
When to draft your second RB follows the same story as with RB #1! As early as possible! In fact, redraft draft strategy for your second RB is distinctly different than in best ball formats. In best ball (as you can see in the Underdog specific article) you should actually wait and grab your second RB in rounds 8-11 (though taking your second in round 2 is also viable). In redraft however, the data is very clear. You should be drafting your second RB as soon as possible! Every round that you wait to take your RB 2 decreases your chances of making the playoffs and winning the championship.
Also side note: I’m just going to show the graph for playoff rate for now on. As you can see, and as will continue being the case, championship rate and playoff rate graphs follow the same trend. We really only need to look at one of them. Since championship rates can be slightly skewed by one player having a big week 16 or 17 (depending on the year), I’d rather use playoff rates.
What About Your Third RB?
Things start to get interesting when analyzing your third RB selection. For the first time we see above average success for a massive range of rounds! Because of how optimal RB early is…taking two early affords you flexibility in selecting your third. You still shouldn’t wait too long, but as long as your third RB is taken in rounds 3-8…you’re in good shape.
What’s really interesting is the flexibility that a RB/RB start can offer! This next graphic shows the same thing…playoff rate by round each team took their third RB. Only this time I only included the teams that started off the draft with a RB in each of the first two rounds! As you can see, the only consistent trend is that every single round besides round 13 produces an above average playoff rate! If you start off RB/RB…you basically can’t go wrong with when you decide to take your third RB. That’s incredibly valuable to know because it allows you to be patient for the entire draft. You can afford to wait until a RB you like becomes a value, not needing to worry that you’re missing out on the position. You can start attacking other positions and just wait for that third RB to come at a discount! And what happens if you do that? Success continues to skyrocket! If you start off RB/RB, and take your third RB at an ADP value, your playoff percentage will be significantly higher than average. Also yes, I know the scale for the Y axis should expand on this graphic. But I wanted to display how much your playoff chances go up when you do this. The data does seem to indicate that there are better options than drafting your third RB in round 3 – and I tend to agree with that from a strategy perspective. The whole point here is that going double RB allows you to wait for the best value when drafting your third RB. By drafting one in the very next round, you’re not really taking advantage of that edge. A great value might show up in the middle rounds, but you’ve already taken your third, and so the positional advantage starts to quickly dissipate. This will be evident again in the next section.
A Better Way to Display Specific Draft Decisions:
The graphs above do a good job of showing when it’s best to take a RB. But I wanted to create graphics to highlight what’s optimal for every decision you might have at the RB position. Below are the 5 graphs I created to help guide you throughout your drafts!
Let’s start by explaining the first one (Right: Drafted 0 RB – PR Diff Drafting 1st). This graph shows teams that have drafted zero RB’s entering each round. The y axis (vertical axis) shows how much playoff rate changes by drafting your first RB in that respective round. As you can see with this first graph (again… Drafted 0 RB – PR Diff Drafting 1st) every single round is positive. Your chance of making the playoffs goes up if you choose to draft your first RB in any of the displayed rounds (only 2-8 because the data for 9+ wasn’t large enough for my liking…and we already know drafting a RB in round 1 is optimal).
Here’s how you would utilize this graph. Let’s say you’ve drafted 0 RB’s in the first two rounds. You’re in round 3 and you want to know if drafting your first RB would be optimal. Simply look at the graph that says Drafted 0 RB’s and locate round 3. The value at round 3 is positive (green) meaning your playoff chances increase if you decide to draft your first RB in that round!
Here’s another example. Let’s say you started off your draft with RB/RB/WR/RB. This means you’re in round 5 and you already have 3 RB’s. Simply locate the graph titled Drafted 3 RB’s and find round 5 on the x axis (horizontal axis). As you can see, the value is negative (red)! This means that your playoff chances actually decrease if you draft your 4th RB in round 5!
This is a VERY important note I need to make. These graphics DO NOT show you how long you should wait before drafting each respective RB. For example…You see a spike in playoff rate difference at round 9 in the “Drafted 1 RB – PR Diff Drafting 2nd” graph. This DOES NOT mean that the best time to draft your second RB is the 9th round. These graphics depict single moments in time. What it’s actually saying is: If you happen to find yourself with 1 RB in round 9…you should definitely draft your second RB. It’s not saying to wait until round 9 to draft your second RB. Another way of looking at it is by saying that the most dangerous time to avoid taking your 2nd RB is in the 9th round. That’s because these graphs are displaying the difference between drafting a RB in that exact round, and not doing so. This is NOT showing you the playoff rate of waiting until each respective round. PLEASE read this paragraph again if that does not make sense. Again, each round should be treated as a single moment in time. Look at your current situation and evaluate what the graph says is optimal. The highest points on each graph indicate the times when you’d do the most damage to your team by not doing what the graphs says (i.e. not drafting a RB if it says to). The same is true for the lowest points. You do the most damage to your team if you draft a RB at a round where it’s negative on the graph. This is because your playoff rate goes down by drafting a RB in a round that’s red on a graph.
Also, please make sure you use the correct graph! Once you use one and draft a RB…you move onto the next. So you’d start by using the one that says Drafted 0 RB’s…then once you draft one you move on to the one that says Drafted 1 RB and so on. Looking at the wrong graph is going to mislead you!
I’ll post them all below here in order, so you don’t get confused.
All 5 RB Draft Decision Charts:
What About Value?
The one caveat to the graphs above is when value comes into play. We’ve all been in a draft before where we want to take a RB…but we’re just not in a good position to take one. So the question pops into our minds – Do I reach? And if reaching is ok, how far can I reach? While there’s no cut and dry answer here, I attempted to answer this with the chart below. What you’ll see is the value cutoff for drafting a RB by round and by RB status.
Let’s go over one of our examples from above. We learned that if it’s the 5th round, and we already have 3 RB’s, that we DO NOT want to take a 4th. But…that’s an average. What if there’s a screaming value! Then what should we do? According to this chart, if the value is greater than 10 spots in ADP then it’s actually ok to draft the RB! This is depicted with a +10 under round 5 value for the 3 RB’s so far row. Had we only drafted two RB’s, then our value would actually be -2. This means we are fine to reach up to 2 ADP spots!
Value/Reach Chart:
You can use this chart in conjunction with the 5 above to make those difficult draft decisions, but I’d personally recommend studying these graphics and understanding them. If you understand what situations result in drafting and avoiding the RB position, then you’ll be better off on draft day.
The one value that’s going to stand out is round 4 when you don’t have a RB yet. You’d think it would display a negative number, but it doesn’t. My take on that is that the 4th round is just a historically bad spot to get your first RB. The elite options are gone and there are some great WR/TE/QB options. And even if you get a value, you’re getting a value for a fringe elite RB talent (third rounder). It’s just not a great spot to attack RB, which you can see in a lot of the graphs above as well.
One last note here is a reminder that reaching is never ideal. This chart does show you when it’s ok to do so, and by how much. But we always want to be getting a discount in our drafts. So while reaching 8 spots for a RB in the second round is fine…it’s much better if you were to end up getting that same player at a discount in the third round. Sometimes it’s worth taking a risk that a player will still be there in the next round, instead of reaching too much. But again, the above chart shows the cutoffs for how much it’s still ok to reach in each round.
How Many RB’s Should You Draft?
The final question we’ll answer today is how many RB’s you should be drafting. All else equal, aim for 5-6 total RB’s. Unless you have abnormal settings (1 or 3 starting RB spots (not including FLEX spots), or like 10+ bench spots) this is the way to go. If you expand or reduce the starting RB spots in your league, then adjust these numbers by that same amount.
The range of 5-6 total RB’s also changes depending on what you do in the early rounds. If you decide to start off your draft RB/RB…then it’s now still optimal to draft 5…but anywhere from 3 to 6 still gives you an above average chance of making the playoffs.
And this makes sense, right? If you use two extremely high draft capital spots at RB, then each additional RB carries less and less value to you. At the same time, you’ve also avoided every other position for the first two rounds. That means you’ll need to take more shots on them in the middle and later rounds!
One Last Note For the RB Zero Crowd:
I want to note that this sample obviously includes people who are bad at fantasy football. It’s entirely possible that some people avoided the RB position by mistake, started the season horribly, and gave up. So while we do see a big drop off in success if you wait too long for RB…I want to note that this does not invalidate strategies that wait on the position. If you consider yourself an above average fantasy player, and you know you’ll be active all season, then you can afford to deviate from optimal drafting strategies. Especially if that reason is because you see insane value somewhere during the draft. If you’re a subscriber here, then I think you fall into this camp. You’re much more likely than the average user to make good waiver and trade moves…thus affording you some flexibility to take chances in the draft. That’s not a pass to take 4 TE’s to start a draft…but it does mean that you don’t need to freak out if you do something deemed sub optimal. The odds are you’ll more than make up for it in other areas, and there’s likely a good reason you decided to deviate. Every single draft will be different, so study these draft strategy articles, stay up to date with the latest news, and I’m sure you’ll feel great about your draft when it’s all said and done!