Mea culpa. I recently posted a piece on handedness in daily fantasy baseball that was presented poorly with a couple of major errors. I apologize if you read it and thank those that took the time to privately contact me and not publicly throw me under the bus. As penance, if you read it please challenge me to a head-to-head matchup on your favorite DFS site and I promise to use nothing but cold players with bad better versus pitcher history.

Here’s the deal. The real point I wanted to make was how some are misusing reverse splits to gain a perceived edge when identifying value hitters in their daily lineups. So today, instead of taking a confusing circuitous route, I’ll try to be more direct.

Using hitting splits is one of the ways to gain an edge in daily fantasy. Examples are home versus away, day versus night, facing left-handers versus right-handers, batting average with runners in scoring position and first half versus second half. Some splits are random while others have predictive capabilities. Our concentration will be on those that are non-random, specifically home versus away and the topic at hand, pitcher handedness.

The measuring stick with splits is plate appearances. What we’re looking at is plate appearances under multiple conditions. Inherent is the player’s likely performance is better under a particular scenario. The extent of this edge can be determined by treating the stat in question as a weighted average, with plate appearances as the denominator.

The reason splits are helpful is most sites base their salaries on a player’s stats to date. Think of the salary as a measure of expected performance. The salary can also be treated as a weighted average in proportion to the number of plate appearances. We just stated that a player is likely to perform better under a specific scenario thus their salary commensurate with that scenario should be higher – but the site doesn’t “charge” you that greater amount. This is where you gain an advantage using splits. The salary corresponding to the entire body of stats is what goes towards the cap but the player’s potential exceeds that level.

The metric we’ll look at to study a player’s handedness splits is weighted on base average or wOBA. This is the brainchild of noted statistical guru Tom Tango and takes on base percentage and slugging percentage to the next level by assigning coefficients to the various components of both OBP and SLP and combining them into one calculation. These coefficients weigh the individual components relative to each other with respect to run expectancy.

Here’s why handedness is important. Below is a table showing the four possible batter versus pitcher scenarios using full season data from 2011-2013 plus 2014 to date:

 2014201320122011
RHB vs LHP.325.323.326.329
LHB vs RHP.319.325.325.325
RHB vs RHP.306.305.307.306
LHB vs LHP.294.287.286.292
league average.314.314.315.316

Big-picture wise, matchups of the opposite handedness tend to generate better results. Obviously, this is by no means earth-shattering and you don’t need me to tell you in most instances, matchups of opposite handedness can lead to an edge.

So what do you need me for? I’m not saying this is universal but I have read and heard some applications of split data that are misleading and may not lead to the expected edge. The rest of this discussion will do what the previous posting attempted and that is to elucidate the errors so you don’t make the same mistake when setting your daily hitters.

The key to it all is regression. In general, all performance to date needs to be regressed towards what’s expected with the extent of the regression proportional to the sample size of the performance to date. The earlier in the season, the more we push expectations to historical and not current. We’re still early enough that historical exceeds present.

With respect to hitter wOBA and handedness, you’re no doubt aware this data is available versus lefties and righties. Each player obviously has unique splits – both season to date and for his career. The question is how many career plate appearances are necessary before a player owns his splits? That is, how many plate appearances are necessary before we regress current performance towards career norms as opposed to league norms?

The answer can be found in The Book: Playing the Percentages in Baseball by Tom M. Tango, Mitchel G. Lupman and Andrew E Dolphin. Their number-crunching reveals it takes 2200 plate appearances against righties and 1000 versus southpaws for the handedness splits to be real. Assuming around 600 plate appearances per season for a full-time player, that’s over five seasons.

Put into daily fantasy context, unless the player has five years’ worth of regular at bats, their splits need to be regressed towards league average expectations. The extent of the split is in proportion to how close to the 2200/1000 threshold they sit but the main point is an adjustment needs to be made.

The most frequent misuse of this notion is gaining a perceived edge using reverse splits. Reverse splits are when a right-handed hitter fares better against right-handed tossers and a lefty swinger hits better against a lefty thrower. Unless the hitter possesses five seasons worth of experience, these splits needs to be regressed to league expectations so the advantage isn’t as steep as thought. That is, choosing a player based on his reverse splits isn’t going to get you the elevated bang-for-the-buck that you assume if the player is still new to the league.

On the other hand, if the player has sufficient experience and owns his reverse splits, you can indeed get an edge. This is especially true if others aren’t aware of the reverse splits and categorically dismiss using the player since he’s scheduled to face a pitcher of like handedness.

Note that to this point, the focus has been on players receiving regular playing time and even though it hasn’t been explicitly stated, aren’t switch hitters. The treatment of platoon players and switch hitters is slightly different.

A regular faces right-handers about 70 percent of the time. It’s no coincidence that 2200/3200 is .69 as that jives very closely with the percentage of plate appearances versus righties and lefties. A left-handed platoon hitter usually faces righty hurlers for 90 percent of their trips to the dish. A right-handed platoon batter sees closer to 85 percent lefties. A switch-hitter is basically a full-time platoon player so they see the opposite handedness virtually every time with the rare exception of not wanting to mess up their lefty swing against a knuckleballer or having an injury like Shane Victorino last season.

The reason regular versus platoon (and switch-hitter) is important has to do with the weighted average discussed above and the relative amount of same-handedness dragging the wOBA down. For the purpose of this example, it’s necessary that the players have less than the minimum number of plate appearances necessary for their splits to be real.

Let’s take three hitters, each with the same season-to-date wOBA.

  1. Lefty-hitting regular facing a righty
  2. Lefty-hitting platoon player facing a righty
  3. Switch-hitter facing a righty

The order these are listed is the order of their adjusted wOBA. The lefty regular has about 30 percent of his plate appearances versus southpaws which drags his wOBA down more than the 10 percent incurred by the platoon guy. The switch hitter’s wOBA doesn’t need adjusting – it is what it is.

This is a good spot to stop since we’re getting more involved than we need to be with respect to setting your lineups. Besides, Fantasy Alarm has a tool that does all the dirty work for you with respect to making a projection based on the pitching match-up that day. All this is factored into the little black box that sits in a folder on my hard drive and is backed up in several places.

In the coming days and weeks, a lot of bandwidth will be dedicated to how to use the projection and pricing tool to your advantage. But for now, if you’re eyeballing a lineup

  1. Reverse splits can be useful so long as you are cognizant of how many plate appearances the player has with respect to the number needed for split stabilization
  2. Be careful about “double-dipping” when eye-balling a platoon player and especially a switch-hitter’s value based on their matchup. The weighted average adjustment is basically built-in.

Again, please accept my apology for the previous posting. If you’re reading this it has been pulled.

And yes, I was serious about challenging me to a head to head contest.