Finding Upside in Fantasy Drafts: When Each Position (QB/RB/WR/TE) has Produced "Upside" -- Historical Statistics Study

“Upside” gets tons of talk for drafting strategy.

I wanted to dig in deep and find out where to find it.

I’m analyzing “where to find Upside” on 2 parameters, in this study:

1. (1) Fantasy position (QB/RB/WR/TE) and

2. (2) Draft rank order.

Bear in mind there are other complex ways to look for Upside, not part of this discussion— for example certain backfield situations, or dependence on rookies, etc.

So this study is only about draft position (ADP) and its effect on potential upside.

Inspiration.

There’s a fantasy narrative out there, that you should draft an RB early because they have “upside”.

And that narrative is sometimes used to imply: you shouldn’t grab TE because the same upside isn’t there.

I thought this was bogus. Wasn’t McBride a great example of “upside” last year? Hasn’t Kelce been a league-winning draft pick, in years before?

Now to be fair, it might be semantics. What do people really mean when they refer to “upside”?

For me, Upside means points scored beyond expectation.

But maybe people are debating something different. Maybe their claims of RB-Upside simply meant “LOTS OF POINTS”. Or more nuanced, “more EXPECTED points above Replacement”.

To me, these latter concepts don’t represent “Upside”— because those are just about getting what you ask for: You draft them early because you expect more points.

So on reflection, I might have avoided all this work— everything that went into this analysis— just by accepting we had different meanings.

Nevertheless, isn’t a bit interesting to know where Upside-Beyond-Expectation occurs in the draft?

Defining Upside.

Let’s get assumptions out of the way.

As I wrote, my meaning of “Upside” is: getting more points than you expected.

More specifically, Upside is when you get more points as compared against draft value.

So if you’re drafting the 01.01 running back, his point expectations are already HUGE.

And his expected-points above replacement is also HUGE.

You might say, he “can only go down”.

Now, that’s not necessarily true: the RB1 could exceed expectations. And it goes without saying, of course you should still pick him, because of the draft value. But the point is: “upside” is different from “Value”— and it’s a different mathematical approach.

The concept I’m addressing here is: Upside is the player’s fantasy scoring that goes above his expected scoring, as seen before the season began. Any outcome that isn’t above expectation just gets measured as a ZERO, to represent “No Upside”. (These are players you might have dropped. Or they’re players who stayed on your team and you did “okay” but not the “League-Winning” type of okay.)

Since you’re still with me, there’s one further nuance to the definition: Everything is taken with respect to Replacement Value. Because this is what determines draft order AND because it’s also what determines player “worth”. To make sense of the study, we also have to consider later drafted players as having a ZERO draft value starting point— instead of counting them for their “negative VORP”. We only want upside to count beyond Replacement Value. Otherwise we’re just counting “extra scoring they had to do, to catch up to being just Average”.

With that definition of Upside, let’s look at the actual calculations!

Methodology

My big advancement to this analysis is that I gathered the stats connected to specific players— including Rookies:

1. All actual player Fantasy Score OUTCOMES, tied to their name each season, and

2. All actual player Fantasy Score PROJECTIONS from each season— as viewed from each pre-season.

This is powerful, because there’s no approximation about it (such as counting last year’s RB1 result as the next year’s RB1 projection).

We’re using the real vision of what players were EXPECTED to do, beforehand. And we can directly compare the RESULTS to EXPECTATIONS.

My steps were like this:

1. Convert projections into VORPs.

2. Convert outcomes into Upside.
   This is similar to VORP, done by subtracting (outcome - MAX(projection, replacement level) ).
   To express it differently: MIN(Outcome Over Expectation, Outcome Over Replacement).

3. The above values are floored at 0, for the purposes of this study described above.

4. Plot Upside vs. VORP, for each position.

5. Look for best-fit curves that fit the scatter— but only if they are validated by a Running Average trendline (15-20 samples).

In other words, this procedure converts all past real DATA into scatterplots that represent the true historical occurrence of Upside, as a function of Approximate Draft Position.

I know there will be objections— some people think the floor should be different, some people think only the “ highest percentiles of upside” should count. But I think these all average out in the end, and give meaning to the results.

Results

Here are the scatterplots for the different positions.

As you can see, in some cases (not all) there are apparently real trends through the scatter, as validated by some of the Running Averages.

These results imply that certain “tiers” of each position have a higher / or lower chance of breaking out for Upside.

Some initial things to note:

1. The highest AVERAGE amount of upside is 25 season points— About 1.5 ppg.

2. From all the scatter points at the top of each graph, it’s clear that UPSIDE can come from ANY pick number. It’s not only the top-tier, or only the bottom-tier players who can give these excess points.

3. Running average gives an especially bad fit to WR in particular.

Interpretation

Here’s my explanation of what I see in the graphs.

First of all, we should shift the Upside trends down to the same baseline.

Why?

Well, the RB graph currently shows RBs as having the most Upside almost ALWAYS.

But you can’t just take RBs at every turn of the draft.

The question really is “Is it better to find Upside earlier or later with RBs?”, and we’re not looking for an answer like “All the time!!”.

Here’s the adjusted graph, without all the noise:

Since VORP was used for the x-axis, it already approximates the draft order of players! (From right to left.)

This is why we can compare positions in the same graph— the dots line up approximately in their ADP from right to left. And the approximate ADP numbers are labeled, where the dots represent the positional average VORPs for each rank, across all the seasons.

Here are the trends we can pick out:

1. Quarterbacks have increasing Upside towards later in the draft. Upside peaks around QB7, and then plateaus around QB20.

2. Running backs. Most upside comes from the second tier of RB3 - RB7— though typically not RB1 and RB2 as we all know (injury effect).
   RB upside reaches another high point again after RB36 or so.

3. Wide Receivers have the most noise of all— as revealed by the Running Average trendline: It converges very poorly. The curve fitting reduces the noise but cannot be trusted.
   WR3 - WR6 bring the most upside on average.
   After the low-points around WR10, Upside increases through mid-draft (though erratically), and then increasing again into the late draft especially.

4. Tight ends have some upside peak around TE2-TE4
   Then TE6 - TE12 fall in trough It seems significant, since the Running Average trend reveals such a WIDE low point.
   Finally most TE upside exists around TE18-TE32.

Upshot

I hope the visual is graphed in a way so you can see clearly when each position has its best chances at upside.

Don’t take this serious, but it means that for a pure Upside-targeting approach, you would:

1. Start with Early RBs. RB3-12.

2. If the above RBs are gone, then WR3-8; otherwise wait.

3. If they fall to you at draft cost, get TE2-4 but don’t take any mid-round TE.

4. During the mid-rounds, get mostly WRs, before a QB.

5. Get later-round QBs. QBs 5-9 are best, otherwise just wait.

6. Get a late TE if you need one.

7. Get back-up RBs after ADP 100.

8. Fill anything else with WRs.

Finally don’t forget that you can’t just value upside alone. 
The whole point is that you need to grab Draft Value (Value-over-replacement firs)t, and then consider upside as extra.  

Total potential = Value + Upside

Thanks if you followed along!  I hope this helps you as you approach your draft strategy.  For a lot of us—myself included—it has been enormously difficult to find a way to judge draft order positional value, and I hope this paves the way to better analysis and understanding, in the future.

/Subvertadown