The Draft Rankings Built By 3 Million Voices

How AI met the Wisdom of Crowds to create the most original draft board ever

Who Knows Best?

There’s a crowd gathered around a prize ox at a fair. They watch as butchers go up one by one, all experts in their craft, analyze the creature, and propose their best estimate of the weight. Meanwhile, the crowd itself is hazarding a guess. They’ve decided to take the average of absolutely everyone’s individual estimate. And, to great surprise, they’ve come far closer than any of the experts. It’s not magic – it’s math.

This is exactly how James Surowiecki introduces his book, ‘The Wisdom of Crowds’. The theory suggests that the average prediction, forecast or estimate of a group of random people, when subject to specific criteria (a very important point that we will return to later), tends to be more accurate than any group of experts on the topic.

As we approach the peak of fantasy draft season, this seems more relevant than ever. There are scores of supposed, and actual, experts vying for your attention, and peddling you yet another draft cheat sheet.

So, let’s try and find a new way of doing things. How can we leverage this idea of crowd wisdom to optimize fantasy outcomes? Spoiler: we already did it, and all you have to do is enjoy the fruits of our labor. I’m not suggesting that this should be your be-all-and-end-all draft ranking (although, to be honest, it’s pretty good), simply that this was an interesting experiment that we wanted to share.

---

Choose Your Crowd Wisely

There are four conditions that must be met for this phenomenon to come to life. If met, we are greeted with an incredible forecasting tool.

The first is diversity of background. Varied experiences fuel better group estimates – they cancel out bias and add range. This is exactly what is missing from the expert opinions. For example, the butchers may think they have seen an ox which is a carbon copy of the one on the stage. They remember its weight when they broke it down. But, in reality, this idea led them down the complete wrong path, because – unsurprisingly – this ox is not simply a clone of another.

The second criteria eliminates many crowds from accessing collective wisdom, especially in a digitally driven world. It is the requirement of independence. One crowd member’s opinion cannot influence those of other members. Otherwise, it quickly spirals into groupthink, and we lose the diversity that was the very fuel of our prediction vehicle. This is incredibly challenging to avoid. You can see the damage it causes through the case of financial bubbles, when greed and mania take over. We’re confident we avoided it in our draft rankings though…

Next is the need for decentralization. This simply means that information is coming from multiple sources, as opposed to one dominant source that is influencing everyone’s minds in the same way. In the case of fantasy football, the only shared source between everyone is the football itself. As every person has a different reaction and opinion on what they’re watching, this condition is pretty much automatically met. It’s not like everybody is reading the same football analyses all the time.

And, finally, is aggregation. Everyone’s opinions must be measurable, so that the average one can be discovered. This is incredibly easier than it used to be – with the wealth of data that can be found online.

So, does fantasy football have a crowd that fits? We think we found it.