For years, I have hyped up my deep sleeper picks, called them “narcos” and pleaded for you to target them in your drafts. Those sorts of claims have made for good blog posts but they’re not worth a damn unless I have something to back up the assumption that these sleepers actually produce more value than your average draft pick. So, before unleashing the 2013 fantasy baseball narcos, I started to research narco selections from the past five years to see if they performed better than a standard player. In order to do this, I researched the “expected” WERTH value a player should produce at a draft slot versus their actual value produced (which led to a draft pick value chart I posted previously). From there, the project turned into a much bigger study that resulted in a NEW AND IMPROVED narco system for 2013.
There were 94 players who met my old narco qualifications over those five years that were researched. The 94 players were being taken at points all over the draft with many even falling within the Top 100 picks. But, do the players taken that early even have a chance to be “sleepers”? If not, when is the “right” time to find a sleeper? Heck, what other factors should I be considering when looking into these narcos? These are the type of questions that started circling my head as I started to research these narcos. The answers to those questions were intriguing.
To come up with those answers, I isolated the 94 narco candidates then created a few variables for each player:
- WERTH value produced in the year they were labeled a sleeper
- ADP value for them that year (hat tip to Fantasy Gameday for the ADP data)
- Expected WERTH value for that ADP (based on that previous research)
- The difference between actual and expected WERTH value
- Whether they were their team’s starter on Opening Day
- Number of Plate Appearances in previous season
- Number of Plate Appearances in career prior to that season
I dove in and took a deep look at those 94 narcos and then also compared them to the total fantasy baseball draft pool over those five years as well (1969 players total). Here is a summary of the lessons that I learned when selecting deep sleepers:
Early Round “Sleepers” Aren’t Worth The Risk
Sleepers who have an ADP below 100 are being taken with a high risk. You’re passing over good proven players at that point to go with a high potential sleeper (think Desmond Jennings or Brett Lawrie last year). The sad thing is that these players generally don’t deliver on the hype.
I looked at five different ADP tiers (Under 100, 100-200, 200-300, 300-400, Undrafted) and the Under 100 tier of sleepers is the only one that significantly performs under expected WERTH value on average.
Only 29% of those drafted in this tier actually even performed at or above their expected WERTH value that year (compared to 52% within the other four narco tiers). In addition only 14% performed +2 roto points above their expected value (compared to 40% among those other tiers). The only example of a success to come out the Under 100 tier was Giancarlo Stanton in 2010.
For the total draft pool of 1969 players, there were 315 players that had an ADP in the Top 100 and they produced close to their expected WERTH value on average. Meanwhile, the narcos in the Top 100 performed 2.7 roto points below expected WERTH value on average. Losing nearly three points in the standings by drafting one of these players is a huge loss to your team early on.
It Matters If A Player Is An Opening Day Starter Or Not
Of those 94 narcos, 60 of them were starters for their team on Opening Day. From that group of Opening Day starters, 37% of them performed at or above the level of the average fantasy player while only 15% of the non-starters could claim that. This difference becomes even greater in the 100-300 ADP range where 45% of OD starters performed above average while 17% of non-starters are.
Number of Plate Appearances Matter
I generally look at players with around 100-400 PA’s from the previous season. The theory is that they are undervalued because they didn’t get a full season of stats the previous year so they don’t jump out to the average fantasy player who looks at season totals. I get a little lenient here and there with those that had slightly over 400 PA’s but I’ve found that I shouldn’t be. 18 of the 94 narcos had over 400 PA’s the previous season and on average they performed 1.7 roto points below expected WERTH value. That’s not very good. Those with less than 400 PA’s from this pool (but over 100 PA) were all performing above their expected WERTH value on average.
Number of Career Plate Appearances Matter
The other part of the narco theory is that I am looking for guys that just haven’t gotten their chance to play full-time yet which would be why they are being slept on. Despite that, there are potential narcos who slip in there despite accumulating a lot of previous PA’s over their career. Of the 94 narcos, 7 of them had over 500 PA prior to their narco-qualifying season. On average, that group of 7 players performed 2.0 roto points below expected WERTH value.
My Threshold Could Be Improved
Since I’m looking at players without a full season of stats, one of the main things I’m looking at is their production per plate appearance. In doing so, I found that I’ve been too lenient with my threshold of their production per PA when determining who qualifies as a narco. If I make the threshold a bit more stringent, the success rate of the narcos improves dramatically.
The End Result
By taking all of those lessons into account, I can tweak the system to be ridiculously more successful. I do end up with a much smaller narco pool over those past five years as we drop from 94 to 34 candidates. But, those 34 new narcos produce +3 roto points above expected value 56% of the time versus 26% from the total draft pool (and 34% from the old system). Over half of this group is performing way above average which is great to see.
That value even increases more if we only include Opening Day starters for their team (but that’s sometimes hard to predict prior to the season starting). When only including the Opening Day starters, we go down to a pool of 25 candidates but they perform +3 roto points above expected value 64% of the time. Nearly two-thirds of the players from those qualifications are vastly outperforming expectations, which nearly doubles the success rate of the old sleeper system. Me likey.
So, my modified selection process now includes a few new caveats:
- Must be below 400 PA in previous year (and above 100 PA)
- Must be below 500 PA in years prior to that
- Points/PA at higher threshold
- ADP greater than 100
- Ideally, his team’s Opening Day starter
These new caveats will give a much more limited pool but a much more successful pool at the same time. Stay tuned for the results of the 2013 narco selection process. Get your drumroll ready.