Monday, April 14, 2014

Gridiron Solitaire #101: Reflections

One of the qualities in football that I like very much is what I call reflection. Football is composed of very discrete entities (plays), and your team is either on offense, defense, or special teams.

In theory, there's no interaction between a team's offense and its defense. In reality, though, there's a ton of reflection. Everything the offense does, really, has a reflective effect on the defense.

I know--I'm butchering "reflection". I just like how it sounds.

Here's how it works. When a team's offense has a high time of possession, it means that its defense doesn't have to be on the field very often. That's a huge, huge difference in terms of how a game progresses. When a team runs 60 plays on offense, it's an entirely different game than when it runs 40.

That's one consequence of the high-speed offense, and you particularly see this in college because so many teams go no-huddle. It's exciting to see your team score in ninety seconds on an 80-yard drive, but it also means that the other team is going to get many more chances to start drives.

If a team has a run-first, possession dominant offense, there are other reflections. These kinds of offenses are much lower-scoring, generally, so while your defense might be on the field less often, there's much more pressure on them to stop the other team from scoring.

Special teams have their own set of reflections. A great punter can pin the other team deep in their own territory on a regular basis, and teams are much less likely to pass when they're starting inside their own 10-yard line. That gives the defense an advantage. On kickoffs, good coverage can force the opposing team's offense to start from a difficult position.

In a game sense, that's an excellent set of interactions and dependencies from seemingly discrete entities.

In the original version, Gridiron Solitaire nicely modeled some of these reflections, particularly time of possession. Since Big Play presses on defense were a limited commodity, controlling time of possession was critical. Because of that, and because run matches were worth 4 yards each, running the ball became the dominant strategy for 75% of players, at least. It certainly was for me.

This wasn't realistic, though.

With the new definition of 3 yards per running match instead of 4, running becomes situational instead of dominant, which correctly models real football strategy. That's good, and it's much more accurate, but I still wanted to more correctly model the influence of special teams.

That's being done in two ways now. First, on kickoffs, both the kick distance and return distances are modeled on real NFL data and are influenced appropriately by special teams ratings. Second, with the new punt code (which I discussed last week), good punters are now a real weapon.

Plus, there's something more.

I was frustrated, trying to model the difficulties an offense faces when pinned deep in their own territory. Garret suggested something (reducing the CPU's offensive efficiency and max gain), and as often happens, his suggestion made me think of something that I would not have previously considered.

In this case, what I thought of was adding Big Play presses for the defense.

Here's how it works. If the punt is downed inside the 10, two BP Presses are added. If it's inside the 15, it's worth one. Those BP presses are very valuable, and in combination with more predictable CPU playcalling in that area of the field, models real life in a fairly satisfactory way.

I've played about a season with this new setup, and with a good special teams rating, I find myself punting from midfield much more often. Plus, there's an added element of interest, because where the ball is downed becomes interesting in itself.

Now I feel like, for the first time, I'm modeling all the different interactions between separate units of a team and the effect they can have on each other.

Generally 1.3 testing is going pretty well. I could see it being released in early May. Right now, I'm working on bug fixes, plus Fredrick is adding as many big images as possible. It's going to be different, but good different.

Site Meter