The Rundown: FantasyScore Week 3

Photo: Kirby Lee, USA Today Sports

The past two weeks, I gave guidance related to game selection and bankroll management when playing FantasyScore's Draft-N-Go (DNG) games. Today, I'm going to pivot towards an analysis of strategies you should (or shouldn't) use while actually drafting.

For redraft leagues, there's a debate that goes on every preseason wherein people make their case for why you should adopt any number of position-based strategies at the top of your draft. Given the parameters that exist in DNGs (i.e., PPR, with three wide receivers and a flex), the most popular of these strategies are

• Three running backs, one wide receiver
• Two running backs, two wide receivers
• One running back, three wide receivers
• Zero running backs, four wide receivers

For this experiment, I'm going to make the two basic assumptions I did in last week's column:

1. All players are using the same draft list.
2. All players select the best player available unless they've already filled that player's position in their lineup.

week 3 draft lists

Because the results of my experiment depend on everyone working off of the same cheat sheet, let me first give you my optimal value-based drafting lists for this week:

the zero-rb experiment

The 2-player DNG is too simple of a simulation and the 8-player DNG is too complex, so let's focus on Baby Bear's porridge, the 5-player DNG.

In addition to what I wrote earlier, every experimental result needs a baseline to compare against, which in this case means that we need to find out what happens when the draft proceeds without anyone adopting the 0RB strategy. In other words, how does the draft play out when everyone perfectly adheres to the "same draft list," "best player available," and "wide receiver as flex" assumptions only? Well, here's how:

Just so we're all on the same page, and because this kind of table shows up another four times in this article, it's worth clarifying how you should read it. In a 5-person DNG, which snakes and assigns draft slots at random (and given our earlier assumptions), the player with the first pick will select Antonio Brown (i.e., the highest-VBD player on my draft list), the player with the second pick will select Julio Jones (i.e., the second-highest), and so on until the player with the fifth pick, who will select Rob Gronkowski and LeVeon Bell when the draft snakes. The draft proceeds this way until a player has filled a certain position in their wide receiver flex lineup (e.g., Pick 3 had to take Seahawks defense instead of Adrian Peterson at No. 18 because they already had two running backs). Once the draft is over, we can use this week's Footballguys projections to see how many points each team would have scored if the projections were perfect. For instance, the player picking first would have ended up winning the 5-person DNG with 157.0 points (italics is for winners), while the player picking fifth would have ended up finishing last with 152.0 points.

OK, so now we know the baseline result when we've programmed five robots to play a DNG using the same information and the same strategy. The only variable in play here was the randomness with which draft slots were assigned.

And because of this random assignment, our experiment actually requires five mini-experiments. Namely, because we have an equal chance of being slotted into any particular pick (i.e., we should expect to have each pick once in every five 5-person DNGs we play), we're not always going to be using the 0RB strategy from Pick 1 (or Pick 2 and so on). Rather, we're going to be using it from the first pick 20 percent of the time, using it from the second pick 20 percent of the time, etc. All of this means that we need to find out how much -- if at all -- using the 0RB strategy affects each pick's total projected points, all else equal.

Below are the draft results for using the 0RB strategy from Pick 1:

I've highlighted in green how Pick 1's team differs from baseline when adopting the 0RB strategy. The major inflection point in Pick 1's draft occured at No. 11, where they selected Cobb as their WR3 rather than Jamaal Charles as their RB1. That change affected two later picks, with the end result being (in essence) a trade of Charles, Aaron Rodgers, and Steve Smith for Cobb, Forsett, and Luck. And if you look at the bottom line, that "trade" increased their projected points by 0.2, thereby increasing their expected margin of victory. Of note, and as you'll see going forward, the typical case is that adopting the 0RB strategy primarily has a "nearest neighbor" effect. For instance, in this case, the extra 0.2 points added to Pick 1 came directly from Pick 2, while projections for Picks 3 through 5 remained exactly the same.

Now, let's see what happens when we use the 0RB strategy from Pick 2: 

Because the baseline team at Pick 2 selects three wide receivers in the first three rounds anyway, the major inflection point was in the fourth round, where they selected Hilton as their WR4 rather than Adrian Peterson as their RB1. That change only affected one other pick later on, and so using the 0RB strategy at Pick 2 was tantamount to "trading" Peterson and Donte Moncrief for Hilton and Ingram. As it turns out, that's a negative expected value "trade," and so Pick 2's projection was 0.8 points less than their baseline result. Also of note is that, once again, Pick 2's loss was Pick 1's gain (and vice versa), except this time the projected gap between them actually doubled from 1.1 points when Pick 1 used the 0RB strategy to 2.2 points when Pick 2 used it.

Let's see if Pick 3 benefits from the 0RB strategy:

Pick 3's inflection point occured immediately because they were no longer selecting Marshawn Lynch at No. 3. Unlike what happened to Pick 1 and Pick 2, going 0RB had a cascade of effects on the rest of Pick 3's team when compared to baseline. But was it worth it? On the surface, no: Their 0RB team ended up with a projection that was 0.7 points worse than baseline. Given that they weren't projected to cash (i.e., finish in the Top 2) to begin with, adopting a 0RB strategy seems to have made them even less likely to do so.

That said, what's most interesting about these results is how Pick 3's choice to go 0RB was particularly proficient at throwing everyone else's draft boards into dissaray. By having so many different players under the two strategies, every other team ended up with a points projection that's different than baseline: Pick 1 still won, but scored 0.7 more points; Pick 2 still cashed but scored 0.5 fewer points; Pick 4 scored 0.3 more points; and Pick 5 scored 0.2 more points. Taken together, the results of this mini-experiment for Pick 3 seem to suggest that, although the 0RB strategy is bad for them, it's incredibly disruptive to the draft as a whole, and that can be an asset in high-variance situations. If you're new to the game, use it. If, despite my previous warning, you've elected to play a 5-person, winner-take-all DNG, use it. Otherwise, don't use it.

(Because Pick 4 selected four wide receivers in their first four picks under baseline conditions, the full-draft 0RB results were identical to the baseline results, so I'm going to skip to 0RB results for Pick 5.)

Finally, here are the results of a 5-DNG draft where only Pick 5 uses the 0RB strategy:

Alrighty then! As you can see, Pick 5 ended up with almost an entirely different team when they used the 0RB as compared to when they didn't. The end result? Like all previous picks aside from Pick 1, their projected scoring decreased; this time by 0.8 points. Unlike what happened when Pick 3 adopted the 0RB strategy, however, Pick 5's overhauled lineup didn't gum up the works for the rest of the draft. Rather, while Pick 4 benefited considerably (+0.6) and Pick 3 benefited slightly (+0.2), the teams for Picks 1 and 2 were unchanged from baseline.

bottom line

Based on the results of my experiment investigating whether or not to adopt a 0RB strategy in FantasyScore's 5-player DNGs, I've come to the following conclusions:

1. If you're assigned to the first pick, use the 0RB strategy. It will make you slightly more likely to win.
2. If you're assigned to the second pick, don't use the 0RB strategy. It will make you far less likely to win.
3. If you're assigned to the third pick, 0RB is a particularly high-variance strategy. You're unlikely to cash anyway, so use it if you're an underdog or are playing a winner-take-all DNG. Otherwise, don't use it.
4. If you're assigned to the fourth pick, it doesn't matter whether you use the 0RB strategy or not.
5. If you're assigned to the fifth pick, definitely don't use the 0RB strategy. If will make you far less likely to win. And although you increase the variance in your own lineup, you don't end up increasing the variance in the game as a whole.

All of that said, I'd be negligent if I didn't stress that these recommendations may end up being idiosyncratic to this week's Footballguys projections and only apply to 5-person DNGs. Although I went to great lengths to make this as "clean" of an experiment as I could, the fact that it's the first attempt (I know of) at investigating the question means we shouldn't generalize the results just yet.