KEEPER STRATEGY

Since we've accumulated a few n00bs over the past couple of years, and because of certain conversations I've had with league mates (and trade negotiations I and others have had), I thought it would be fruitful to examine exactly what makes a good keeper in our league.

Some might say I'm giving away secrets or a competitive advantage, but considering that keeper trades that totally should have happened in the past haven't happened, I think it will benefit the overall health of the league to do this analysis. So pay attention.

Keeper Round is More Important Than You Think
The general point of this post is that where you get to keep a guy is almost more important than the quality of the guy in the abstract. We all want to keep stars on our teams, but there are only so many of those to go around. Keeping a quality rotation guy in a late round is an underused option. Keeping a star in a late round is the utmost in keeper value. (And keeping a star right around where he'd be drafted anyway isn't really doing much for your team.)

To illustrate this, I'm going to pull some figures from last year's ESPN player rater (which you can find here; keep in mind, these are totals, not averages -- in averages, for instance, Anthony Davis would be a super stud and Russell Westbrook would be a borderline first rounder, but they both missed a significant number of games). If you're not familiar, the player rater uses standard deviations and league averages in each stat category to determine how valuable each player is above the "norm" in each category. The total player rater score is simply all the category scores added together. (Yahoo's ranking system uses a similar, but not exactly the same, formula.)

These figures can give us a sense of the expected value in each round of the draft (and other tiers). I've created this chart as a handy reference. Pay more attention to the numbers than the names:

THE SUPER STUDS				SIXTH ROUND				TWELFTH ROUND
Name	Pick	PR Score		Name	Pick	PR Score		Name	Pick	PR Score
Durant	1	21.31		R Foye	51	6.52		SLivingston	111	3.8
Curry	2	17.53		R Jackson	60	6.01		J Sullinger	120	3.3
LeBron	3	16.55
				SEVENTH ROUND				THIRTEENTH ROUND
FIRST ROUND (REMAINING)				Name	Pick	PR Score		Name	Pick	PR Score
Name	Pick	PR Score		R Hibbert	61	6		D Green	121	3.22
K Love	4	14.54		DWilliams	70	5.51		E Kanter	130	2.88
P George	10	12.67
				EIGHTH ROUND				FOURTEENTH ROUND
SECOND ROUND				Name	Pick	PR Score		Name	Pick	PR Score
Name	Pick	PR Score		T Gibson	71	5.5		R Allen	131	2.85
J Wall	11	12.56		P Gasol	80	5.23		M Morris	140	2.53
C Bosh	20	9.69
				NINTH ROUND				FIFTEENTH ROUND
THIRD ROUND				Name	Pick	PR Score		Name	Pick	PR Score
Name	Pick	PR Score		A Afflalo	81	5.19		J Crawford	141	2.5
P Millsap	21	9.41		B Bass	90	4.92		R Jefferson	150	2.27
T Ariza	30	8.47
				TENTH ROUND				SIXTEENTH ROUND
FOURTH ROUND				Name	Pick	PR Score		Name	Pick	PR Score
Name	Pick	PR Score		Westbrook	91	4.92		GHenderson	151	2.25
L Aldridge	31	8.42		V Carter	100	4.46		MWilliams	160	1.79
K Thompso	40	7.74
				ELEVENTH ROUND
FIFTH ROUND				Name	Pick	PR Score
Name	Pick	PR Score		MBelinelli	101	4.43
G Green	41	7.66		P Mills	110	3.9
T Lawson	50	6.59

That's a lot of data to digest. For those who are more visually inclined, here's the same data in graph form:

Lessons from the Data
Let's see what we can glean from the data.

• The Super Studs are amazing. Look at the numbers. Curry is equivalent to a late first rounder (Paul George or John Wall) PLUS a ninth rounder (Aaron Afflalo). Durant is like a late first plus a late third rounder (Trevor Ariza or LaMarcus Aldridge). Put another way, Curry is worth two Aldridges and Durant is worth two Nic Batums (again, on totals, not averages). It's like being able to start 11 guys instead of 10 every day, with that 11th guy being very valuable.
• 
  
• As the rounds go on, value flattens. Studs are very valuable, first and second rounders are next, and then picks start getting more equivalent (or "linear" in the value progression). Another way to realize this:
• 
  
• The score difference between the top pick and the end of the first round is around 10 points.
• The difference between the beginning of the second and end of the second: 3 points.
• Difference between start of third and end of third: 1 point.
• Every other round: less than 1 point difference from start to finish.
• 
  
• Based on the scores, the average "first rounder" is equivalent in value to two late fifth rounders or three eleventh rounders. Again, it's like being able to play more people in your lineup every day than your opponents.

How to Use This for Keeper Analysis

It's pretty simple. Let's say you've got Paul Millsap (9.41 value) as a fifth round keeper. Is that better than Roy Hibbert (6.00 value) as a fourteenth round keeper? Let's do the math:

• Millsap's value (9.41) - expected value of a 5th rounder (7.1) = 2.31
• Hibbert's value (6.00) - expected value of 14th rounder (2.68) = 3.32

So by this analysis, Hibbert would actually be a better keeper, even if he's a worse player than Millsap and even though both have keeper value.

The essence of this is replacement value and opportunity cost. If you keep Millsap in the fifth, sure, you're getting value there. But if you didn't keep him, you'd still draft a fifth-round-level player -- which replaces a larger proportion of Millsap's value than a 14th-round-level player would replace Hibbert's value.

Another way to look at it -- what's better, A or B?

• (A)  Millsap (9.41) in the 5th  +  a 14th rounder (2.68)  =  12.09
• (B)  A 5th rounder (7.1)  +  Hibbert in the 14th (6.00)  =  13.01

Option B gives you more overall value on your team.

Caveats and Cavills
Of course, this is pretty quick-and-dirty analysis. We play in a head-to-head league, not rotisserie, so averages are probably more indicative of value than totals. And how well you draft makes a big difference -- the guy you draft in the 5th round might end up returning 3rd-round value or he might return 10th-round value. We just don't know until the games start. (Of course, this is why we even have good late-round keepers: some guys outperform their draft slots.) And past performance doesn't guarantee future returns -- we each will evaluate players differently, and so will disagree whether a guy is, say, a 3rd-round level guy or a 5th-round level guy.

But the overall principle still applies: you should be thinking about replacement value and opportunity cost when assessing keeper value. If any of you would have taken me up on my trade offers involving Klay Thompson a couple years ago, you'd now have a current 5th-rounder (by Yahoo pre-ranks; probably a 4th rounder in our league with all the keepers) in like the 10th round -- or, in other words, someone returning 3.5 to 4 surplus player-rater points (the equivalent of having an extra Danny Green, Jimmy Butler, or Josh Smith on your team).