In the Trey Sermon thread, some folks offered they’d probably take the cheaper option (Raheem Mostert) a few rounds later since their rankings/projections are pretty close. That led to a good discussion revolving around The Ambiguous Backfield study by JJ Zachariason (NumberFire, Late Round QB podcast.) Turns out the data points to a sort of wisdom of crowds result, that the higher ADP (player drafted first) is more likely to be a breakout league winner. You can check it out in the Sermon thread, read about it 
here, or give it a 
listen.
Yesterday my favorite FF YouTube channel had a discussion on a less extensive (but still interesting) study they did, specifically to see if it might give some clues on the Gordon/Williams backfield situation.
2nd Round RB Production, 2011-2020
29 players drafted in the 2nd round, 3 never got on the field bc if injury, so sample size = 26
looking at those 26 rookies & the incumbent RB on the roster:
- the veteran outproduced the 2nd round RB 8 times (30.8%)
 
- the rookie finished higher in FF 18 times (69.2%)
 
when the rookie received at least one touch in 10G
 (19/26)
- 14 times (73.7%) the rookie finished higher
 
- 5 veterans finished higher (2011 Reggie Bush over Daniel Thomas, 2012 3rd stringer Knowshon Moreno over Ronnie Hillman after McGahee injury, 2014 Frank Gore over Carlos Hyde, 2016 DeMarcus Murray over Derrick Henry, 2020 Darryl Henderson over Cam Akers)
 
the next one is insightful for the folks who have said I’ll just draft both
- only twice have both RBs been useful, each finishing inside the top 30. Both times it involved Gio Bernard - with Jeremy Hill In 2014, with Joe Mixon
 
Top 20 when the rookie received touches in at least 10G:
- Veteran RB - 4 times (21.1%)
 
- Rookie RB - 8 times (42.1%)
 
source (starts at 5:15)
		
 
I appreciate the post. Its intruiging.
At the same time I am skeptical of these narrow definitions and the selection bias that happens when people do this. The 2nd round is a pretty narrow sample.
I went into stathead and did a search of the criteria. Here are the RB who come up in the search:
1	2020	2	35	D'Andre Swift	RB	21	DET	2020	2020	0	0	0	6	13	4	Georgia	College Stats
2	2020	2	41	Jonathan Taylor	RB	21	IND	2020	2020	0	0	1	10	15	13	Wisconsin	College Stats
3	2020	2	52	Cam Akers	RB	21	LAR	2020	2020	0	0	0	4	13	5	Florida St.	College Stats
4	2020	2	55	J.K. Dobbins	RB	21	BAL	2020	2020	0	0	0	7	15	1	Ohio St.	College Stats
5	2020	2	62	AJ Dillon	RB	22	GNB	2020	2020	0	0	0	2	11	0	Boston Col.	College Stats
6	2019	2	53	Miles Sanders	RB	22	PHI	2019	2020	0	0	2	14	28	22	Penn St.	College Stats
7	2018	2	35	Nick Chubb	RB	22	CLE	2018	2020	0	2	3	24	44	37	Georgia	College Stats
8	2018	2	38	Ronald Jones II	RB	21	TAM	2018	2020	0	0	2	15	39	22	USC	College Stats
9	2018	2	43	Kerryon Johnson	RB	21	DET	2018	2020	0	0	1	10	34	16	Auburn	College Stats
10	2018	2	59	Derrius Guice	RB	21	WAS	2019	2019	0	0	0	2	5	1	LSU	College Stats
11	2017	2	41	Dalvin Cook	RB	22	MIN	2017	2020	0	2	3	37	43	42	Florida St.	College Stats
12	2017	2	48	Joe Mixon	RB	21	CIN	2017	2020	0	0	2	27	50	41	Oklahoma	College Stats
13	2016	2	45	Derrick Henry	RB	22	TEN	2016	2020	1	2	3	44	78	47	Alabama	College Stats
14	2015	2	36	T.J. Yeldon	RB	21	JAX	2015	2020	0	0	2	20	60	30	Alabama	College Stats
15	2015	2	54	Ameer Abdullah	RB	22	DET	2015	2020	0	0	2	13	74	22	Nebraska	College Stats
16	2014	2	54	Bishop Sankey	RB	21	TEN	2014	2015	0	0	1	6	29	12	Washington	College Stats
17	2014	2	55	Jeremy Hill	RB	21	CIN	2014	2018	0	0	3	21	55	43	LSU	College Stats
18	2014	2	57	Carlos Hyde	RB	23	SFO	2014	2020	0	0	4	31	90	58	Ohio St.	College Stats
19	2013	2	37	Giovani Bernard	RB	21	CIN	2013	2020	0	0	2	41	115	30	North Carolina	College Stats
20	2013	2	48	Le'Veon Bell	RB	21	PIT	2013	2020	2	3	5	62	88	81	Michigan St.	College Stats
21	2013	2	58	Montee Ball	RB	22	DEN	2013	2014	0	0	0	7	21	3	Wisconsin	College Stats
22	2013	2	61	Eddie Lacy	RB	23	GNB	2013	2017	0	1	3	36	60	51	Alabama	College Stats
23	2013	2	62	Christine Michael	RB	22	SEA	2013	2018	0	0	1	9	38	9	Texas A&M	College Stats
24	2012	2	50	Isaiah Pead	RB	22	STL	2012	2016	0	0	0	1	30	1	Cincinnati	College Stats
25	2012	2	61	LaMichael James	RB	22	SFO	2012	2015	0	0	0	2	18	0	Oregon	College Stats
26	2011	2	38	Ryan Williams	RB	21	ARI	2012	2012	0	0	0	1	5	3	Virginia Tech	College Stats
27	2011	2	56	Shane Vereen	RB	22	NWE	2011	2017	0	0	1	25	79	9	California	College Stats
28	2011	2	57	Mikel Leshoure	RB	21	DET	2012	2013	0	0	1	5	17	14	Illinois	College Stats
29	2011	2	62	Daniel Thomas	RB	23	MIA	2011	2014	0	0	0	11	52	3	Kansas St.	College Stats
	While the results of that study may be accurate I think presenting the information as a percentage is somewhat misleading because in many cases the rookie may have scored the most fantasy points, but not by that much.
In the case of Swift Adrian Peterson still had more rushing attempts than Swift did his rookie year.
In Taylors case Marlon Mack got injured early. Not saying Taulor wouldnt have outperformed Mack anyways, and perhaps by a lot, but if you are going to exclude the rookies who got hurt, it seems to me the same treatment should be given to the incumbents.
Can Akers outscored Henderson by 20 points in standard scoring while Henderson outscored Akers in PPR by 29 points. Close enough that I would call this more of a tie.
I could keep going but perhaps you see where this is going.
As far as Williams vs Gordon perhaps one should narrow the criteria to how many times has a 2nd round rookie outperformed a veteran RB who was a 1st round pick? Because thats more what this situation is than just drawing from a ton of different teams and circumstances.
In the RB study I did I found that the rookie season is the least productive of a RBs first 6 seasons in the league on average and there are a lot of different reasons for that. One of the main ones being that the rookie often has to time share with other RBs already established on the roster. The mileage on this varies a ton because some of these RB have stronger competition than others.
A 2nd round pick is a pretty strong investment into a RB and teams usually are not doing that unless the RBs they have are not part of their long term plans. If they were the team would use that 2nd round pick on a different position most of the time.
In any case I would hate for people to read your post and come away from it thinking Williams has a 70% chance to outscore Gordon this season.
A better way to approach this might be to look at the teams draft history instead of looking at it league wide.
Since 2000 here are the RB Denver has selected in the first 4 rounds:
1	2018	3	71	Royce Freeman	RB	22	2018	2020	0	0	0	9	46	8	Oregon	College Stats
2	2016	4	136	Devontae Booker	RB	24	2016	2020	0	0	0	15	77	7	Utah	College Stats
3	2013	2	58	Montee Ball	RB	22	2013	2014	0	0	0	7	21	3	Wisconsin	College Stats
4	2012	3	67	Ronnie Hillman	RB	20	2012	2016	0	0	1	17	56	15	San Diego St.	College Stats
5	2009	1	12	Knowshon Moreno	RB	22	2009	2014	0	0	3	33	63	45	Georgia	College Stats
6	2005	3	101	Maurice Clarett	RB				0	0	0				Ohio St.	
7	2004	2	41	Tatum Bell	RB	23	2004	2008	0	0	1	22	54	22	Oklahoma St.	College Stats
8	2003	4	108	Quentin Griffin	RB	22	2003	2004	0	0	0	6	16	5	Oklahoma	College Stats
9	2002	2	51	Clinton Portis	RB	21	2002	2010	0	2	7	72	113	108	Miami (FL)	College Sta
	I see a lot of miss here.
Freeman Ball and Hillman ended up being outperformed by UDFA RB Lindsey and CJ Anderson.