Diversifying In Best Ball

Feb 19th 2022Hayden Winks

The always trusty Wikipedia says diversification “is the process of allocating capital in a way that reduces the exposure to any one particular asset or risk. A common path towards diversification is to reduce risk or volatility by investing in a variety of assets.”

In best ball, we typically think about not being over exposed to a single player. Is 50% of Mike Williams too much? What’s the most exposed we should be to a player? These are the debates, and I’m paid to have thoughts on them, so let’s dive in.

The first thought is that diversification isn’t only about player stances. Those are the easiest to grasp because we literally have a player exposure page on Underdog Fantasy, but diversification can be about players, roster construction, positional value, offense projection, player archetypes, and more. The point of my initial tweet on the subject wasn’t only about player stances. It was about these things, too:

But the debate on diversification in fantasy sports really comes down to risk preference and confidence.

Diversification doesn’t inherently make you more money. It just improves your downside risk and allows you to play the game for longer within your bankroll. … If you don’t go 100% Josh Allen in the 2nd or 3rd round this year, then you aren’t going to lose all your money this year if he gets hurt in Week 1. If you don’t want massive anxiety every week sweating Bills games, or if you want to be able to enter NFL Playoffs Best Ball when the season is over, or if you don’t want to get divorced, then yes, diversification is your friend. But that alone doesn’t mean you’ve gained expected dollars. You just sleep better.

For the same reason, the size of the bet matters. For us, that means the round of the draft pick. Because most points are added in the first 12 or so rounds, being highly exposed to those early-round players will dramatically swing your profitability (in either direction). This is obvious, but being 40% exposed to a third round pick is a whole lot different than being 40% exposed to a 15th rounder. If the thesis of diversification is simply to not expose yourself to the lowest outcomes, then keeping the same maximum exposure on players throughout the draft doesn’t make sense. The consequences of being wrong on a 17th round pick is close to zero, so taking stands on later round picks isn’t the problem it’s made out to be.

Confidence is the key second part here, and I’ll set things up with an extreme example before getting to the real life examples later. … Let’s say your uncle is Bruce Arians, and he told you behind closed doors that Tom Brady is actually not retiring and is instead showing up to July training camp with Chris Godwin and Rob Gronkowski re-signed. In this scenario, you shouldn’t diversify much (if at all) on selecting Brady and Gronk in the last rounds of the current best ball tournament. There is a lot of confidence in this bet being a good one (especially given the low stakes), so the exposure should be much higher here than if we had zero idea if Brady is unretiring. … When you have confidence, the exposure climbs.

But enough of the nonsense. Let’s get to the actual best ball debates.

Should I have 75% of a particular player?

No! But it’s not because of “diversification”. 

In order to accomplish having an exposure that high, you’d have to reach ahead of ADP and that in itself is assuredly -EV. Even the most optimistic drafter should be maxing out on 30-50% of a player solely because we want to pay good prices on each player we draft. 

If I’ve drafted Mike Williams in Round 5 every single time because I thought the odds were good that he re-signs in Los Angeles, then I’m still losing to the people who are currently drafting Mike Williams at his Round 7 ADP. In other words, we should still be waiting until we get very close to ADP to be drafting “our guys”, which at the same time will mean we get sniped on occasion, lowering our exposure to that player. That’s fine!

The other part of this is player confidence, which is the obvious counterargument to my initial tweet. And I don’t disagree with it.

We really shouldn’t be that confident in any particular player. Period. There are so many injuries (especially with Miss Rona still around) and the sample sizes in the NFL are too small to be confident in the actual result playing out as expected, even if we have confidence in a player’s talent, scheme fit, or whatever. 

Just as an idea of how far player stances can get us, I attached the results of my 2021 overall best ball rankings. I had about a 3% advance rate edge over the market (pretty damn good imo!), but this is as a full-time best ball analyst while grinding my ass off with building models, reading news as a Rotoworld alum, grinding the tape (as an amateur tbf), and consuming fantasy football content like a mad man. Player stances alone will not make you a profitable player. It’s just not the game we’re playing.

Is it better to be overexposed to something other than players?

Yes! And here are a few examples.

Roster construction: Myself and others have grinded the pick-by-pick data on Underdog Fantasy to produce a lot of solid content on when and how often we should draft each position, and there are other analysts (myself included) who think about this from a game theory standpoint, too. The theories on each build are out there and if you think there is an edge to a particular one, then you can be highly exposed to that build without being super confident in one particular player stance. For example, a full Zero RB hardo could split his or her first round exposure somewhat evenly between Cooper Kupp (30% exposure), Ja’Marr Chase (20%), Justin Jefferson (20%), Davante Adams (15%), and Tyreek Hill (15%) without having to be 60% exposed to a single one of them.

Positional value: Sometimes a certain site is just not valuing an entire position because of unique scoring. For example, I wrote a column about how QBs are Undervalued on Underdog Fantasy, and for me that means being slightly above ADP in my top-250 overall rankings on 10 of the first 11 quarterbacks. If I max entered using these rankings, I’d have 12-30% exposure to each one of the quarterbacks but wouldn’t be at 50% on a single one of them.

Player archetypes: This can be about certain incoming prospects, but I’m mostly thinking about specific NFL player roles (like insurance RBs, downfield WRs, or thicc committee backs). These types of profiles are ones I believe are undervalued in a vacuum, so I want lots of exposure to all of them without having to go all in on just Alexander Mattison (who is a total stud at ADP for the record). If people keep taking Chase Edmonds types ahead of James Conner types, then I’d win over time based on my research on player archetypes.

Offensive projection: We are getting back to micro takes which can be dangerous, but betting on a coaching philosophy, strength of schedule, offensive line skill group, or another form of a team is a way to being overexposed to a defined group of players without attaching to one player. For example, last year’s Rams (to me at least) were going to be a high-scoring passing offense because of Sean McVay's neutral pace and neutral pass rate. And a bet on the collective Matthew Stafford, Cooper Kupp, Robert Woods, Tyler Higbee, Van Jefferson, and DeSean Jackson was extremely profitable, even if some players listed above busted and Cooper Kupp happened to be the best pick in the entire draft.

Conclusion

Diversification isn’t necessarily making you more money. It’s about risk preference, confidence, and the size of the bet. If you are okay with losing your bankroll due to an injury, then your exposure can be higher. If you are pretty confident in an outcome, then your exposure can be higher. If the stakes are low, then your exposure can also be higher. None of this alone makes you a profitable or losing player.

But for best ball, diversification isn’t just about individual player takes. We can diversify players within the things we have more confidence in (like roster construction, positional value, player archetypes, offensive projections, and more) because the sample sizes are larger, which makes these things more predictive year over year.

We should not be scared to compete when we are exploiting an edge just for the sake of diversification. We just need to understand how confident we should be in each identified edge and make sure we are still getting good prices when we’re getting “our guys”. But putting money into the identified edge is actually how you profit off of it. And for that reason, it’s okay to make identifying edges more of a priority than diversification.

But if you just want my 100% infallible player takes, then watch my videos on YouTube: