I should probably do this more often to illustrate use cases for the GLSP apps but the problem is that between writing and working on the site’s apps, the week usually gets away from me. In any case I wanted to illustrate how I go about using the GLSP apps to make start/sit decisions for my own fantasy teams. I’ve written similar pieces in the past but we always have new readers so I should do it more often.
Let’s take the example of a PPR league that I’m in where you can start up to four WRs. Being able to start that many WRs means that I usually have a good number of start/sit decisions to make on a weekly basis. This week most of the decisions are pretty well made for me, but there is one decision that isn’t as straightforward as it might seem. I have Doug Baldwin going against the Eagles defense, and Vincent Jackson going against the Detroit defense. This is exactly the kind of match-up that GLSP is made for. On the season Jackson obviously outscores Baldwin, so the “keep it simple stupid” method of setting lineups would say to just start Jackson. But how do we know the effect of the matchups in question. Oh, right, GLSP.
The first choice we have when using GLSP is whether to adjust any of the app’s settings. GLSP allows you to consider a player’s most recent 16 games as the relevant sample in determining what kind of players they’ll be compared with, so let’s look at Vincent Jackson’s most recent 16 games.
When I look at a player’s game log I’m primarily looking to see if the games represent a reasonable range of outcomes for the player’s existing usage and ability. For instance, if a player has seen a recent bump in usage, like Odell Beckham did after Victor Cruz went out, then the pre-Cruz-injury games aren’t entirely relevant and I might look to filter them out. But in Vincent Jackson’s case even going back 16 full games leaves a range of games that I think all represent the type of player Jackson is. Over the past 16 games he’s been on about a 1,000 yard and four touchdown pace. So there’s no real need to adjust his game log.
Then if we look at Doug Baldwin’s game log we see this:
Baldwin’s case is a little tougher because at some point Percy Harvin was traded and presumably Baldwin would have had to become a bigger relative part of the offense. I don’t want to have to filter out games out of the game log if I can avoid it, because a larger sample is always better. But a quick slide of the recency slider to the seven most recent games shows that Baldwin has averaged a full target per game and about eight receiving yards per game more over the most recent seven games, than he has over the most recent 16 games. So in this case sliding that recency slider to throw out older games seems like it’s warranted.
When I do that I get the following projection for Baldwin:
I can compare that to the projection that I get when I used Vincent Jackson’s full game log:
My league is a PPR league so I can see that Vincent Jackson has the higher floor and median projection. But the interesting thing is that Baldwin has the higher ceiling. There are some conclusions that I could draw just from having done this part of the exercise, but there’s actually another step I can do before I’m really done. I can check both player’s sensitivity to their game log. So basically, if I move the sliders around so that various games are filtered out, and that changes the projections, that’s also valuable information to have. No matter how many games I include Vincent Jackson’s projection tends to hover in that 9-11 points range. But if I move the slider back and forth so that Doug Baldwin has fewer than seven games in his game log (essentially asking the question of how sensitive is Baldwin’s projection to one big outing versus STL) then his projection drops down quite a bit.
In this specific instance I think GLSP tells me something valuable about this start/sit decision. I know that Vincent Jackson probably has the safest floor. If I felt like I was a favorite in the matchup I would probably just plug in Jackson and expect a solid and unspectacular day. But if I felt like I was an underdog then I might think about playing Baldwin against the PHI defense. I would know that the chance to get a real crappy game out of that roster spot would increase, but I would also be opening up the upside as well. I know this both because of the range of projections that I show above, AND because I know that the projection is sensitive to my view of Baldwin’s usage. Depending on whether Baldwin is the 11 target receiver we saw against STL, or the two target receiver we saw against ARI, the range of expectations could be really wide. But also, of the two receivers Baldwin is more likely to have the higher upside.
GLSP is there to help you do some of the math in your start/sit decisions. When you adjust filters and game logs GLSP makes some quick calculations and then finds other similar players that have roughly the same average stats and are facing roughly the same type of defenses in games where the offense is projected to score roughly the same (because the team total from the point spread is one of the similarity criteria). GLSP takes care of the science part as the calculations are almost instant, while you still have some room to apply the art of decision making on your end.