Photo: Imagn Images

We're four weeks into the season, and my system for predicting weekly quarterback points is doing pretty well. If you want the full explanation, click here. Just as a quick refresher, though, it's based on three variables:

1. The QB's average going into the week, with David Dodds' preseason projection counted as a "game"
2. The point spread of the QB's game
3. The over-under of the QB's game

So, the numbers show that 2013 has been the most accurate year to date, both in terms of points error and plus-minus prediction, Week 4 had the least points error of 2013, and Week 3 had the highest plus-minus percentage.

In terms of outliers, the system was off by 10 or more points for only one of 24 quarterbacks, Drew Brees against Miami on Monday night. With a 25.7-point average going into a game where New Orleans was -4 and the over-under was 47, Brees' projection was 23.4 points. He ended up scoring 36.9, which was actually higher than his upper bound projection of 30.8. In a system like this, we can expect one out of every 10 QB games to produce a score outside his expected range, so Brees just happened to be part of that 10 percent last week.

On the other end of the spectrum, Week 4 saw four QBs score within one point of their projection. Ironically, the most accurate of the four was Brees' counterpart on Monday Night, Ryan Tannehill: The system projected 18.1 points, and he scored 18.3. The other three bullseyes were Matthew Stafford versus Chicago (projected 22.6, scored 21.9), Tony Romo at San Diego (projected 20.9, scored 21.7), and Geno Smith at Tennessee (projected 16.3, scored 17.2).

In addition to having the most accurate projections in Week 4, Tannehill, Romo, and Stafford have received the most accurate projections so far this season. On average, Tannehill's have been off by 1.0 points, Romo's have been off by 1.2 points, and Stafford's have been off by 2.3 points. Meanwhile, Aaron Rodgers (10.5), Eli Manning (10.3), and Peyton Manning (10.1) have been the most unpredictable QBs among the top 24. Odds are good, however, that the error averages for the Manning brothers will come down over time because both have had at least one bizarro game that's significantly clouding the picture. Eli's dud at Carolina in Week 3 and Peyton's record-setting performance versus Baltimore in Week 1 are not likely to repeat themselves. Also, let's not forget that the system will have a better idea about their "true" averages as the season progresses.

week 5 qb projections

Here are the projections for Week 5:

As an example of how to read the table, take a look at Peyton Manning's row. The "Points" column says he's projected to score 29.1 points against Dallas, which is lower than his current average of 31.8 points per game. The system is 90 percent confident that he will score between 21.0 points and 37.2 points. As a proportion of his projection for the week, matchup-related factors (i.e., the spread and over-under) boost him 19.6%, with a range of 16.3% to 21.8%. Viewing Manning's row from 30,000 feet, he has a distintinctly positive matchup this week, but his current average is so high that regression to the mean says he'll probably score below it.

Now for a few highlighted matchups...

eli manning

The Giants-Eagles game has an astronomical over-under of 53 points, currently the third-highest on this week's slate. Unlike the two games with higher over-unders, however, New York is only favored by one point, which, if you remember from the introductory piece (linked to earlier), is right in the wheelhouse of where QBs benefit most from the spread -- not too big in either direction. In contrast, Peyton Manning (56.5) and Aaron Rodgers (53.5) benefit from high over-unders, but both of their teams are favored by more than a touchdown. Furthermore, we've seen Denver pump the brakes with a big lead over the past few weeks, and we've seen Green Bay do so almost religiously under Mike McCarthy.

From an X's and O's perspective, Eli should also get somewhat of a break from the pass-rushing onslaught he's faced the past two weeks. According to Football Outsiders, the front sevens for Kansas City and Carolina have two of the top four adjusted sack rates in the league; Philadelphia's front seven ranks 16th. To boot, the Eagles rank 30th in pass defense efficiency and 26th against opposing No. 1 wide receivers. Not to mention that with nearly 70 percent of the Giants' running back carries in Week 4, and the subsequent release of Da'Rel Scott, David Wilson finally seems to be out of the Coughlin-Gilbride doghouse. Having at least some semblance of a running threat is better for a pass offense than the alternative.

alex smith

Speaking of the Giants, Smith is coming off a 28.1-point game against New York last week. And with an average that's now 2.5 points above his preseason projection, he's seemingly everyone's favorite, unheralded QB2 right now. That said, he's not a recommended play in Week 5 according to my system. Mainly, at 38.5 points, the Chiefs-Titans game not only has the lowest over-under of the week; it's the lowest over-under of any game so far this season. In short, Vegas (and the betting consensus) sees this one shaping up to be a low-scoring affair -- and they're probably going to be right. As we'll find out tomorrow, when I introduce my RB projection system, although the over-under doesn't matter much, fantasy points allowed does, and both of these defenses rank in the top half against opposing backs. Although this will probably affect Tennessee's game plan with Ryan Fitzpatrick more than it affects Kansas City's game plan with Smith, it still doesn't bode well for the latter having another big fantasy output.

matt schaub

Conventional wisdom after Week 4 says that Schaub's season has literally turned into a dumpster fire -- or a sardonic burger special if you're less pyromaniacally inclined. Of course, it's worth pointing out that his year-to-date average puts him at QB12 heading into Week 5, and he's scored over 24 points in three of Houston's four games, which, notable, includes last week. That said, although he's a better fantasy quarterback than popular opinion thinks he is, Schaub's Week 5 matchup drops him from a borderline QB1 to a borderline QB2. From the system's perspective, being a six-point underdog is a problem, but I think it's overestimating how much that will matter. Over the course of his fantasy football career, Schaub has been almost entirely unaffected by the spread. (I'd show you the graph, but the trend is a boring flat line.)

What's more concerning is the game's 41.5-point over-under, which is one of only five below 43 points on this week's schedule. No doubt, this is due in part to the quality of the two defenses, but also, both offenses figure to focus on running the football. For the Texans, San Francisco has become uncharacteristically porous against offenses not helmed by a Schottenheimer. For the 49ers, their last three games have made it obvious that their points are going to have to come from the run given the situation that's befallen their receiving corps. Furthermore, this is exacerbated by the on-field fact that, in 2012, Wade Phillips kept his starting cornerbacks matched up on specific receivers more than any other defensive coordinator in the NFL. That means perennial Pro-Bowler Johnathan Joseph will be locking horns with Anquan Boldin for most of the game, and anyone who's watched a San Francisco game this year knows none of the other wideouts can beat man-to-man coverage on a consistent basis.

The Texans will host St. Louis in Week 6, so it's best to wait a week before exploiting Schaub's underrated status for personal gain.

reply hazy player of the week -- sam bradford

Schaub (17.1%) and Tony Romo (12.9%) have larger uncertainty ranges in their matchup percentages, but Bradford's (12.1%) is the third-largest of the week, as well as the largest one that covers both positive and negative territory. You might remember that Russell Wilson won this award two weeks ago when his team was hosting the Jaguars, so the same logic applies here. St. Louis could be up big early, and shut down the passing game. To wit, offensive coordinator Brian Schottenheimer never met a running play he didn't like. On the other hand, the Rams could easily build an early lead via the pass before shutting things down, which would obviously inflate Bradford's stats. Or, the Rams' moribund running game might not be good enough to exploit Jacksonville's 29th-ranked run defense (24th in fantasy points allowed to opposing backs), which means more passing from Bradford in a closer game. I mean, who knows; and when you aren't confident in a projection, the best advice is to simply stay away.

early week 5 rb projections

As I said earlier, the RB system in all its glory will come out tomorrow, but I do want to post the projections for Thursday night's Buffalo-Cleveland game so you can use them to your benefit. Here they are: