Thursday, August 5, 2010

Year One Strategy Review

At long last, the NFL season approaches, and with it, BadNFL makes its much anticipated return to the blogosphere. Before the season gets underway, I've long been planning to do a 2009 Season Recap that takes an introspective look at BadNFL's inaugural year, breaking down some of the things that went right and some of the things that didn't. In doing so, I hoped to articulate some of the meta-lessons learned from that inaugural season.

An apology though: this summer, studying for the bar exam while working full time inhibited my ability to do the intense sort of data analysis I wanted to do this offseason. Hopefully it can happen next summer! But that said, I don't want to deprive BadNFL Nation of the benefits of a few reflections on the prediction methodology of Year 1.

Coming in the next few days or so, before the season gets underway, will be a counterpart post explaining my takes on a couple of teams that the public and most prognosticators have misvalued heading into the season, in the hopes that those observations can aid in early season game predictions.

Without further ado, here are 6 observations and thoughts:

1. Explosive passing offenses against porous secondaries form a potent combination. This point seems patently obvious when stated like that, but for those of you who followed my analysis closely last year, a recurring theme was that explosive (top-5) passing attacks vs. pathetic (bottom 5) secondaries routinely covered. In fact, there were 3 games where my pick was based in part on the observation of this phenomenon: Week 16 Chargers +3 @ Titans (SDG won 42-17), Week 7 Colts -13.5 @ Rams (IND won 42-6), and Week 5 Colts -3.5 @ Titans (IND won 31-9). Admittedly, the sample size is rather small, but it is remarkable how lopsided these victories all were. The reasoning is fairly simple (first explained here): aggressive, passing-heavy teams are able to jump out to huge leads rather quickly in these games, which renders the opposing offense one-dimensional (leading to turnovers) and fairly unmotivated--both of which help facilitate blowouts. Moreover, teams like the Colts and Bolts last year generally did not know how to lay off the gas pedal; it's not like they had dominant running games to milk the clock late. As such, the score kept growing.

By way of contrast, this might also explain my spectacularly wrong prediction in Week 6, where I took the heavily favored Eagles against the woeful Raiders; in that game, although the Eagles had an explosive offense and the Raiders a pitiful defense, the Raiders actually sported a decent secondary--headlined, obviously, by the superb Nnamdi Asomugha--and sat back and dared Andy Reid to run the ball. He didn't, and because of that (or perhaps because it was a trap game), the Eagles suffered one of the most stunning losses of the year.

Thus, this year, I will keep an eye out for the foregoing phenomenon; if the lines are lower than -14, they might prove appetizing.

2. The Trap Game phenomenon. This phrase is bandied about a lot among the pundits, and no doubt the worry about the vaunted "trap game" will surface during one of my predictions this year. As I explained in my recap of the Eagles shocking loss to the Raiders last year, I remain reticent to make predictions that are heavily founded on psychology, since the ability of NFL fans to estimate teams' mental states seems limited at best.

But then again, there seem to be games where psychology plays an important role; the Bengals' gutsy win over the Ravens following Coach Zimmer's wife's tragic death, the Saints' first game in the Superdome post-Katrina, and Brett Favre's MNF game the night of his father's death all spring to mind. Also, in late season games, where teams out of the playoffs have little to play for, marginal spreads can be beaten easily (witness the 49ers covering against the pathetic Rams in week 17 last year). But for the most part, I want to stay away from psychology this year.

As such, I'm not going to obsess about the "trap game" phenomenon, just because I think it's difficult to predict. However, I'll endeavor to keep in mind my estimation of the constituent elements of a trap game, derived in the wake of the Eagles/Raiders debacle last year:
The lesson is this: you should beware betting on (1) a heavy favorite, (2) playing against a team that has already hit rock bottom, where there has (3) been some catalytic event (here the Pierce criticism) that might serve to inspire them, and (4) where that favorite has not played a good team in several weeks.
3. The relevance of the 1st head-to-head matchup in predicting the outcome of a 2nd game: I don't think I had a good handle on the vagaries of intra-divisional games in Year 1. Specifically, when picking late-season divisional games, I cited as evidence -- without thorough reflection -- my pick's performance against the opponent in their prior game. Conversely, with the Bengals and Ravens, I misjudged the second game, which the Bengals won, because I mistakenly thought the Ravens would refuse to lose to the Bengals twice in one year.

Instead of just unreflexively citing a gut feeling about the previous game, this year I plan to more methodically examine what some sharps call the "revenge factor" when evaluating late season divisional games. That said, I think that term is a little misleading, since it evokes a focus on psychology, something that I've already largely forsworn this year, and as such, many betting guides counsel against its use. Instead, I think it's largely about the team's and the coaching staff's ability to make adjustments in repeat matchups (the same theory explains why playoff basketball is so much different from regular season basketball, where coaches have much more time and film with which to make adjustments). This will be a factor worth watching, but as Miami and Buffalo proved last year, a huge win in an early season game does not mean that the winner will cover the spread in the late season rematch.

4. The "coming off of a big win" phenomenon. There is a tendency to overreact to one week's good performance by thinking that one good game will carry over to the following week. To be sure, a big win can inspire players and confer momentum upon a team. And there is no doubt that teams play better or worse depending on what part of the season it is.

But there is an equally compelling (at least in theory) opposite effect: a big win can emotionally exhaust and depress motivation, leading to a subpar performance the following game. In fact, this basic theory as at the heart of the venerable and popular "zig zag" theory of basketball betting. Moreover, this basic theory was instrumental to the reasoning behind one of BadNFL's best Year 1 picks, when the Redskins easily covered (and should have won) against the Saints. A similar dynamic may have been at work in Week 2, in which the Jets beat the Pats in what was at the time considered an upset. Some analysts, particularly faithful BadNFL readers will remember, had predicted a Pats victory in part because they had demonstrated guts by eeking out a close win against Buffalo in Week 1. But BadNFL knew better; the Pats close win over the Bills revealed structural deficiencies with the Patriots, deficiencies that the Jets were able to exploit.

So, this year, I will be careful in how I invoke a team's performance from the prior week. When a team's good play in recent weeks can be traced to some structural change -- a coaching change, a QB change, or getting a key player back from injury -- then it probably have a greater chance of carrying over. But when it was merely a flash in the pan type performance coming off a bye (as apparently was the case in my ill fated Ravens-Bengals Week 9 pick), then maybe the Saints-Skins effect will dominate.

5. The psychology of the betting marketplace. As all smart bettors know, the Vegas lines do not actually reflect what the bookmakers think the score will be; rather, they reflect the bookmakers' prediction of what the public thinks the score will be. Thus, as I pointed out on occasion in Year 1, finding teams that the public significantly over or undervalues can be helpful in locating lines that are deliberately set inaccurately.

One way to do that can be to rely on advanced metrics like Football Outsiders' DVOA, which are presumably more sophisticated than basic stats like yards and points scored, which are the stats traditionally relied on by squares. Of course, as Chad Millman noted this week, the dividing line between sharps and squares has gradually dissipated, as mainstream sportsfans become increasingly savvy. Like my point #2 above, I'm reticent to probe the public's psychology too much. But I remain convinced that looking for structurally over or undervalued teams can be an important component of a game analysis, even apart from the more traditional scouting analysis that BadNFL usually employs. Thus, even though my use of FO-type stats backfired last season, I remain committed to the concept this year (plus, that Ravens/Bengals game was unduly impacted by the absence of Heloti Ngata, given that he is arguably their best player).

6. Does the preseason matter? Overall, the data show that while there's a correlation between preseason record and regular season wins, that correlation is small, although perhaps most significant in cases where the team was mediocre the year before (between 7 and 9 wins). Granted, some people who parse the data in a more simplistic way can show that in certain cases, a dominant preseason can translate into season success, but in general, preseason record seems only trivially important. That being said, specific stories and experiences that emerge from the preseason can have a big impact on the season. Which preseason storylines will matter this year?

This typical list isolates what you might expect -- big injuries (Ty Warren etc.), contract disputes (Vincent Jackson, Revis), etc. But that's not what I'm chiefly interested in. I'm wondering what performance-related issues from the preseason will carry over to the season. There are several, like Aaron Rodgers' dominance, that are tantalizing. But the one in which I'm quite interested, and one that might serve as a springboard for further insight into what types of preseason issues carry over, is the Cowboys' offensive malaise. Simply put, their offensive line and running game have looked horrendous in the preseason, and the coaches and players keep insisting they'll turn it on in the regular season.

I don't know. This was a topic that I didn't analyze much early in Year 1, but one I want to be vigilant about early in the year. Because it should certainly be helpful to early season betting in Year 3.

No comments:

Post a Comment