Setting a weekly lineup in a points league

In many (I would say manly) H2H leagues, you have to set a weekly lineup. This is always frustrating and, even in the best cases, a bit of a guessing game. You have to decide who is going to be most productive that week, and if there are any reasons you may need to pre-emptively sit a player.

As a case in point, I had to decide yesterday before there were any updates whether to start or sit Ryan Zimmerman. He was listed as day-to-day with an abdominal muscle strain, but it was an open question as to whether or not he'd play at all or hit the DL. I sat him, and that turned out to be the right move.

In general, I preach caution with injuries when weekly lineups are involved. There are three possibilities when there isn't much information. Either a) the player is fine and plays all games, with or without some cost to production due to the injury, b) he plays a few games but less than the full compliment, either because he's held out for a few games or he aggravates the injury, or c) he's held out the entire week and possibly placed on the DL. In all of these scenarios, with the exception of a best-case a-scenario, there is some lost production. So generally, I prefer to sit the player.

But even without such frustrations, the decision to start/sit a player for a given week can be a tough one outside of the obvious studs. Fortunately, points leagues lend themselves to some easy analysis, as long as you make a fe

w assumptions. Here's how I do it.

I make a list of all my players in excel, one for each matchup (e.g. one line for player x against the first team he plays, another for the second, etc.). I list the number of games against each opponent as well, and then create a space for every outcome assigned a point (e.g. hits, walks, strikeouts, etc). I then head over to Baseball Reference and search each player, and get his results versus a current team's pitching staff. I then plug in all the numbers given at the aggregate line at the bottom, as well as a few extras (plate appearances, at-bats, and OPS mainly).

After all that is done, I set up a formula that calculates the number of points from the aggregate line. There is some bias- you can't add in runs or steals this way- but you can either mentally adjust for that, or ignore it. I ignore it, but that's because in my H2H points league I largely ignored players whose value is greatly influenced by steals. In any event, this bias is the same for all players.

After that, there are two more steps. I divide total points by plate appearances against the pitching staff to get points per PA. I then multiply th

at by the expected number of plate appearances per week, which is number of games times average number of plate appearances per game. This number depends on where they hit in the lineup, so I also make a note of where in the order a player tends to bat. Plate appearances per game varies a little year-to-year, so digging for current data is a chore. I use the data from 2008, which is a little outdated, but again you're applying the same bias to all players (so all your projections are off by the same amount, give or take). I use this site from fantasy baseball sherpa, but if you have more recent data I recommend that instead (the numbers are PA per season, so divide by 162). So you have (xPts/PA)*(Games)*(PA/Game), which gives you an estimated number of points for the matchup. You then add together all the lines for the same player, and you're done. You pick the best expected point totals to fill out your lineup.

When doing this, I also make a note as to whether or not a certain player does particularly well/poorly against certain pitchers, and check to see if those pitchers are expected to start during the matchup as well. I don't take those numbers out necessarily, since there's often some on-the-fly adjustments to a rotation (see the uncertainty about whether or not the White Sox will use fifth starter Humber this week or skip him because of the off day), but I do make a note of it. Personally, I use Tristan Cockcroft's Fantasy Forecaster over at ESPN. If a player looks like he's going to get a favorable/unfavorable matchup, I make a mental adjustment.

This is a lot, so it would probably help if I gave an example. This week, I was deciding between starting Logan Morrison and Denard Span in one of my OF spots. First up, I look at Morrison's numbers. He has three games each against the Braves and Phillies. So I head over to baseball reference and look up Morrison's numbers versus Atlanta:

It's not a big sample, but he has 25 PA, 4 hits, 1 double, 1 HR, 1 RBI, 2 BB, and 6K for a .588 OPS. Plugging that into my league's points formula yields ten points, which divided by 25 plate appearances gives 0.4 points per plate appearance. Since he's batting around the middle of the order, that's 4.25 PA per game. The final line is (0.4 points/PA)*(3 games)*(4.25 PA/game) is an estimated 5.1 points for the matchup. I also make a note that he'll likely face Hanson, against whom he has good numbers, and Hudson, whom he doesn't. I then repeat this process for his games against the Phillies, noting he faces Blanton/Hamels/Oswalt, all pitchers he fares well against. That works out to an estimated 11.41 points, with four favorable matchups and one unfavorable.

I then move on to Denard Span. He has two games against the Royals and four against the Rays. His numbers against those teams work out to an estimated 12.1 points for the week using the same basic math, adjusting for a) number of games, and b) the fact that Span hits leadoff. There's no notes on the Royals matchup, since he doesn't have extreme numbers against either Hochevar or Davies. Against the Rays, he's horrible versus Price and OK versus Niemann, with nothing notable about Davis or Hellickson. That's 1 moderately favorable matchup and one bad matchup.

The points work out pretty similar, with Span expected to score roughly 6% more against the teams they face generically. This makes me lean a little towards Span. However, since they have an equal number of unfavorable matchups but Morrison has more favorable ones, I end up going with Morrison since he's more likely to outperform the prediction.

It's a time-consuming process, but well worth doing once per week. Obviously, you're going to start your studs every week. But around the margins- deciding who will be the 4th OF or how to cover for someone you need to take out- it helps immensely. This week, after deciding to sit Zimmerman (necessitating Beltre at 3B), it gave me a quick and easy reference to decide who should be my UTIL player- Roberts (19.34), Span (12.1), or Drew (16.23). Roberts won out pretty easily. The same calculus also made it easier to decide whether to be conservative with Drew (16.23) and sit him for Tejada (19.37), since Drew is a candidate to re-aggrivate his strain. Since Tejada has the better matchup anyway, Tejada won out.