The Long Game: Keeper League Draft Value Chart

Written by

Updated on February 20, 2018 1:16AM EST

From the time I first proposed a column about keeper league strategy back in 2013, the focus has been primarily on auctions. This wasn't intended to be a slight on leagues that still use drafts to stock their rosters; it's just that leagues with auctions and a salary structure are easier and simpler to talk about, especially when it comes to adapting economic concepts like profit and inflation to fantasy baseball. If a player on a $5 salary is projected to earn $20, a decision on whether to protect him – or whether he should be traded for a player with a $20 salary projected to earn $25 – is obvious. Draft leagues, however, abstract those decisions in a less obvious manner. Is a player valued as a third-round pick in a re-draft format, but protected in a sixth-round slot worth more or less than a projected second-round pick in a fourth-round slot?

In fact, I even threw down a challenge in my very first Long Game piece, which to my knowledge no one ever picked up:

"And if you want to get really fancy with it, you can always try to adapt a version of the draft slot value chart that Jimmy Johnson pioneered in the NFL to determine whether those two fifth round picks are worth more or less than those three eighth rounders..."

(Incidentally, the answer is that you definitely want the fifth-rounders).

Well, that challenge has been nagging at me ever since, and this year I finally decided to put it to rest.

I present to you the initial version of the Draft Keeper League Value Chart:

	1st	2nd	3rd	4th	5th	6th	7th	8th	9th	10th
x.1	1700	1100	800	650	500	370	300	220	150	70
x.2	1600	1080	790	640	490	370	290	220	140	70
x.3	1550	1060	780	630	480	360	290	210	140	60
x.4	1500	1040	770	620	470	360	280	210	130	60
x.5	1460	1020	760	610	460	350	280	200	130	50
x.6	1420	1000	750	600	450	350	270	200	120	50
x.7	1380	980	740	590	440	340	270	190	120	40
x.8	1340	960	730	580	430	340	260	190	110	40
x.9	1310	940	720	570	420	330	260	180	110	30
x.10	1280	920	710	560	410	330	250	180	100	30
x.11	1250	900	700	550	400	320	250	170	100	20
x.12	1220	880	690	540	390	320	240	170	90	20
x.13	1190	860	680	530	390	310	240	160	90	10
x.14	1160	840	670	520	380	310	230	160	80	10
x.15	1130	820	660	510	380	300	230	150	80	10

The chart is formatted for a 15-team league, though you could simply rearrange the cells for larger or smaller leagues. Draft rounds are along the top and draft slots are down the left-hand side. The basic idea is to assign values to each slot in the draft, which can then be compared to the projected rankings of various players to determine their relative worth, either as keepers or trade assets. The point values were derived from the average of the last three years of 5x5 roto earned auction values in a 15-team mixed league, with the margins between picks smoothed out. Roughly 150 players earn positive values each season in that format, which is why the chart features 10 rounds of 15 draft slots. These averages are from end-of-season (not preseason) values: the points for slot 1.1 were derived from the average value of the most valuable player at year's end, whoever it happened to be that season — not the average value of the specific player at the top of that season's ADP list.

For instance, let's say you have four potential keepers but only three keeper spots, and the sixth slot in the draft. You can protect Jose Altuve as a first-rounder, Jose Abreu as a fourth-rounder, Jose Ramirez as a sixth-rounder or Jose Quintana as an eighth-rounder. Ramirez certainly jumps out as the best option when you eyeball them, but where do the other three stand?

Using the chart and RotoWire's overall 5x5 roto rankings, or whatever rankings you prefer, you can assign a relative "profit" to each player by subtracting the value of their keeper slot from the value of their projected ranking:

• Altuve (3rd overall projection, 6th slot): 1550 minus 1420 = 130 points
• Abreu (34th overall projection, 51st slot): 770 minus 600 = 170 points
• Ramirez (31st overall projection, 81st slot): 800 minus 350 = 450 points
• Quintana (68th overall projection, 111th slot): 430 minus 200 = 230 points

As it turns out, in that scenario Ramirez is indeed far and away your best keeper, but Quintana is second best by a not-insignificant margin. In fact, if you stick purely by the chart which you hadn't even seen until five minutes ago, Abreu would be your third keeper, but there are certainly strategic reasons to keep Altuve instead of him, and the values are close enough (the equivalent of a 10th-round pick) that you can probably justify fudging the decision a bit.

Here's another scenario. Let's say you have pick 1.1 in a keep-forever dynasty league where protected players don't take up draft slots, and you have made it known that you're willing to part with it. One of your competitors offers you Anthony Rizzo and pick 1.7. Another offers you picks 1.13, 3.8 and 3.13 to move up to the top of the draft. Which offer is better, and should you take either one?

Using Rizzo's RW rank as the 20th player in 5x5 roto, the first offer is worth 2500 (1380 plus 1020) points, a lot more than the 1700 value assigned to pick 1.1. The second offer, however, is worth 2600 points (1190 plus 730 plus 680). Unless you're a Cubs fan and really want Rizzo, you should probably take the second offer, as over the long haul it will add the most value to your squad. Of course, you still need to hit on those draft picks, but that's true in any league.

Note that while the chart may have been derived from a specific format, its application should be universal. Regardless of what league size you are in or what format it uses, or how many players are frozen and kept out of the player pool, elite talent separates itself from the pack, making top-end draft picks the most valuable. There's also hidden value in picking earlier to get the player(s) you want from a tier that might have similar projections, which is also priced into the chart.

Now, I freely admit that this is a first stab at constructing a chart like this. While the various test scenarios I came up with seem to produce reasonable results, there are likely ways it can be improved. Feel free to make suggestions in the comments.