Defensive Formations - Mix it up or keep it consistant?

[Mike D]

When defenses see a particular formation do the same thing too many times they learn to key in on it. For instance, running a 2TE I-Formation Run many times could get the the Defense Looked Very Familiar With That Play response in the gamelog.

First of all, is that fair to assume that or not?

And the next question is the use of Man (Loose), Man (Bump & Run), and Zone defenses. If you run the same defense over and over again, does the offense key in on that for it's own advantage?

Or better yet, has anyone seen any advantage of mixing up their defensive scheming as much as the AI recommend button does? Honestly, I can destroy AI recommended gameplan's week in and week out in Single Player and they keep you guessing with a good mixture of Loose, BnR, and Zone defenses.

It might be hard to get responses on this with the MP lockout on Q&A sessions, but I figured I'd throw it out there for thought anyway.

[Icy]

Sadly defensive strategies are mistery to me, i use this usually:

If I use 3-4 defense (my favorite) I set blitzing to 95% and blitzing more than one player to 90%. Then i set run agressive to 60% and pass agressive to 65%. For the defensive formations, i usually use 90% M2M or B&R (depending on my personal). With this defense and a M2M especialized personal i managed to have one of the best run defenses and got a lot of pressure to the QB's in the NAFL but gave away some deep passes (i guess when the QB could read well the blitzes and the WR's were better than my dbacks).

If i use a 4-3 defense i set blitzing to 60% and blitzing more than one player to 55%. Run agressive to 80% and pass agressive to 55%. For the defensive formations i use a mix (recomended) or Zone if my players are good enought for it.

I would really like to know for which is good/bad each kind of defense, also i would like to see a speed ratting for the WR's and CB's to decide how to match them instead of just the abstract M2M, B&R and Zone players ratings.

[Ben E Lou]

It's easy to get responses on things that are conjecture and speculation. It's just the hard numbers that are harder to come by, so get ready for some conjecture and speculation. ;)

I've wondered quite a bit about this one. NPL played almost all BnR (if not all) when they had the lights-out defensive season in IHOF, but they had obvious talent on defense, so the argument could be made that it was talent, not scheme, that did it. We've played 100% BnR this year, and are 3rd in scoring defense, and in the best 10 in both yards per carry against and yards per pass attempt against, and we don't have a lot of obvious talent on defense, but every one of our LB's and DB's is rated in the upper 50's or better in BnR coverage.

My *GUESS* on the matter is this: knowing Jim, there's probably an advantage that the offense gets from knowing that you're in the same defense all the time. I don't think he would have left that out, since disguising coverages is such a big deal in the NFL. However, if your team's talent is heavily skewed toward a certain coverage (like mine is), then my observation is that you're probably better off to heavily skew your percentages toward that coverage.

The best specific evidence I have for this is LCB Kenneth Nason. Since we got Strawn in 2010, Nason has been able to cover the second-best receiver on offenses, and pretty much always in bump and run coverage, and has performed quite well, despite having an unimpressive level of all-around talent. Basically, Nason is only good at run defense and bump and run coverage, but his PDPct has been at 20 or above since we got Strawn. He'd struggle more against #1 receivers, no doubt, but he's been quite a bit above average against the #2's.

I've never set up a league with a CSV file, but this might be worth doing it. It is possible to specify players' ratings to the point where you could set up all your LB's and DB's to have 50 ratings in all coverages? If so, a fair comparison could be made by running a season with default game plans, then save and rerun the season again, with all BnR, all M2M, etc.

[Mike D]

If FOF terms, it's all how you define conjecture and speculation. From some folks, it's really educated guesses, and from others, there truly WAG's (wild ass guesses).

So, to add a bit of WAG to your educated guess...

I have found very little evidence to back-up anybody's claim that a good mix of Loose, BnR, and Zone is a good thing. It seems like the worst decision you can make for your defense. I've run it a lot and have been absolutely destroyed by opposing QB's.

However (and here comes the WAG part), what if your defense isn't built around a certain style of play. What if your guys have mixed ratings? Like you (Skydog) says, you have players so that you can play a certain style of defense. If you have committed to a certain style, it probably takes away a bit of the thought process on what style you want to run your DB's. Now, that's probably a good thing to keep defensive gameplanning simple, since we know so little about it.

But what if have built a roster full of guys that don't steer towards a particular style or worse yet, aren't even any good at any of the styles? How do you prepare your defense against a MP opponent? To me (the WAG continues), I think the key is to still focus on one style and build your strategy around that. Mixing it all up just never seems to pay off. (Sample Size=Me.)

[albionmoonlight]

I tend to run one style of defense in both my MP and SP games.

You can look at how Fairbanks' defense has done and make your own conclusions about whether that is a good or a bad thing.

[Ben E Lou]

Here's a concern that I have for my team, though, even if there is *no* direct penalty for not mixing coverages. For argument's sake, let's say that BnR defense has some difficulty against outside runs, because the CB's are right up on the receivers and therefore easily set up to be blocked out of the play. Let's say, then, that I play a team that is good at outside running, and does it a lot. If most of my DB's are only good at BnR. then either I'm going to have a tough time against the outside run, or I'm going to have to put my DB's in a coverage that they don't execute well.

[stevew]

Slight imperical evidence, but I've always seemed to create lockdown defenses in solo play with M2M experts, and have seen great MP defenses that play man exclusively. If you can get 4-5 secondary guys with 80+ in m2m, and especially if they can pick off a lot of passes, chances are your team will be nearly impossible to throw against. I suppose you could beat this defense quite easily, but it would take having the right kind of receivers and running a pretty specialized gameplan.

[Warhammer]

I prefer zone defenses because I have seen too many BnR and M2M coverages get beat for the big play.

I have noticed with zones that a 4 Deep Zone gives up a lot of short passes.

I have noticed that a 3 Deep Zone is better against the run than a 2 Deep is.

The problem I have with mixing defenses is that it is too hard to predict what you are going to be in when. If I go 50/50 zone/M2M, what am I going to be in when they decide to go with middle routes, etc.?

[Hoosierbuckeye]

Wanted to give you some feedback on your defensive strategies. In the DFL where I play as the Indy Invaders, I struggled for the first half of the season but still managed a .500 record. We have not been a defensive juggernaut, despite the fact that the d squad is much more talented than the o. We were the second leading team defensively in pass protection w/o the gaudy sack numbers or the gaudy interception numbers. I did away with the four deep zone, pulled up my CB into the two deep that you recommended, played an equal part bump and run (my db's are very good at that) with loose man to man, 3 deep zone having the least percentage points assigned.

We sacked the QB five times and intercepted the opposing qb 4 times and forced a fumble. Absolutely a world of difference defensively.

Offensively we used recommendations found here to utilize a short yardage passing game with a heavy emphasis on the run, despite having an o-line that specializes in pass protection. Our QB threw for only 171 yards mostly to the te's and possession receivers but rushed for a nice 66 yards on his own. Couple that with the utilitarian 103 yards that our power back picked up, I've seen a dramatic improvement in that area.

Invaders hung close with a better team and won in the last 24 seconds on a 45 yard td pass to our speed receiver.

What I'm saying is that the recommendations I have found here in gameplanning has helped me one hell of a lot.

Now one more question: I used the same gameplan in The USFL despite having a far less talented backfield defensively and offensively despite having a qb that specializes in the long ball (my two top speed receivers were out with injuries). Looking at the game logs, my dbs kept getting burned in a close game (we won by three over the reigning Division rival/champs) in the four deep zone. I assigned zero percentage points to the four deep zone, but when I downloaded the game files, all the percentage points were changed. Did I do something wrong when I saved and uploaded the game files or did my staff change it? Can your staff change the game plan.

P.S. Defensively the USFL stats with that same game plan you provided
look like this: 3 sacks, 3 interceptions, and 1 forced/recovered fumble.

Offensively the USFL stats looked like this: 226 passing yards with 128 rushing yards.

Just thought you deserved some feedback. I've seen dramatic improvement in my teams when I utilize the gameplanning strategies I've found here. Thanks guys.

I love this game.

[OldGiants]

When defenses see a particular formation do the same thing too many times they learn to key in on it. For instance, running a 2TE I-Formation Run many times could get the the Defense Looked Very Familiar With That Play response in the gamelog.

First of all, is that fair to assume that or not?



Its completely wrong to assume that. I'm going back to discussions of FOF2K here (and much has been added, I grant) but there was general agreement after much discussion that the "That play looked familiar" response was simply a result of a counter being triggered and posting that message. It had no impact on the defensive play that would be called. If you think about it, that has to be the way FOF works, because the specific defensive formation on each play is chosen randomly.

The defensive game is a set of situational matrices that have the probability of a specific defense being called in each situation. Those probabilities never change during a game. You can change them between games, but never during the game.

As Skydog has pointed out, each coverage scheme has certain weaknesses, so if you go exclusive you will be vulnerable to those weaknesses on every play instead of having them disguised on occaison.

[Mike D]

Its completely wrong to assume that. I'm going back to discussions of FOF2K here (and much has been added, I grant) but there was general agreement after much discussion that the "That play looked familiar" response was simply a result of a counter being triggered and posting that message. It had no impact on the defensive play that would be called. If you think about it, that has to be the way FOF works, because the specific defensive formation on each play is chosen randomly.

The defensive game is a set of situational matrices that have the probability of a specific defense being called in each situation. Those probabilities never change during a game. You can change them between games, but never during the game.

As Skydog has pointed out, each coverage scheme has certain weaknesses, so if you go exclusive you will be vulnerable to those weaknesses on every play instead of having them disguised on occaison.

I had a Voodoo5 card and missed the entire FOF2k1 days...and sadly so. So as a result I missed a lot of FOF information.

I am really not sure what you mean when you say, "there was general agreement after much discussion that the "That play looked familiar" response was simply a result of a counter being triggered and posting that message."

What triggers it, I wonder, if not the gameplan results accumulated during that current game? Is it a pre-determined decision looking at lopsided offensive gameplan numbers? Coaching? (Asking just to speculate.)

[QuikSand]

Its completely wrong to assume that. I'm going back to discussions of FOF2K here (and much has been added, I grant) but there was general agreement after much discussion that the "That play looked familiar" response was simply a result of a counter being triggered and posting that message. It had no impact on the defensive play that would be called. If you think about it, that has to be the way FOF works, because the specific defensive formation on each play is chosen randomly.

Isn't it still possible that the FOF game, either then or certainly by now, has a separate adjustment made to the expected outcome of a play that is triggered by the "familiar" situation? That whenever the message in the game log gets triggered, so does a downward adjustment in the expected result of the play being called, even if the formations and such are the same?

I don't see how there's any conclusive evidence that the "familiar" message means nothing, nor do I recall any general agreement to that effect at any point.

[QuikSand]

...each coverage scheme has certain weaknesses...

I'm not sure that this statement is really accepted either. The cautious posters above in this thread have acknowledged that their thoughts are just speculation (and the like) or are based on their own impressions (potentially flawed, like anyone's). I don't think there's any strong evidence that any one kind of coverage does a particular thing better than another, all things being equal.

We know how we think real football works, but attempting to translate that to firm statement about how FOF works is precarious. I remain very open to the general notion that man and bump coverage, for instance, are equally effecive against screens, short passes, and long passes -- and that the only effective difference ends up being the selected skill of the DBs involved in the play. I don't claim that I know that the game works this way, but I certainly don't claim to know that it doesn't work this way, either. But there are an awful lot of places where uncertainty about the workings of this game was eventuallly revealed to be a fairly simple system, rather than a deeply complex one.

[MalcPow]

This might not be the right place for this post, and it certainly doesn't solve the "which is better" debate, but I did run some testing on this a while ago and found some interesting, although probably not extraordinarily useful data. I do think that QS is right to raise the point that some things are more straightforward in the game's mechanics than our "football" knowledge would lead us to believe, and that was basically where I was coming from when I did this testing. My general theory was that there was nothing inherently positive or negative about any of the coverages (I think sometimes we try to think about ways the game might be coded, and we end up thinking things like "coverage x gets a +10 against short passes" or some equivalent boost on an outcome matrix), but I basically rejected that in favor of a more straightforward hypothesis that the different coverages simply assigned different coverage responsibilities to different positions, and then used the corresponding coverage rating of whichever player is responsible for that coverage position on a particular pass attempt. Which sounds extraordinarily obvious, but in the discussions I've seen it doesn't seem as though it's really being digested. I'm going to just dump a lot of information and try to tease out some theories afterward, or hopefully spark some more useful ideas from you guys.

The first place to start is to have a solid idea of what happens in an FOF pass attempt. There is a set of possible outcomes on every passing play (passing plays meaning sacks + attempts here, there's really no easy way to add in plays where the QB scrambles for positive yardage during a called pass play). About 6% of passing plays end up as sacks, the ball is never thrown, and the offense is left to pick up the pieces after the resulting loss. The other 94% of the time a pass is thrown, and it's subject to all the possible outcomes that exist for an actual attempt. Almost every single attempt is assigned an outcome by the game as either a catch, drop, interception, bad throw, hurry, pass defensed, or a blocked pass (there are a handful of incomplete pass attempts that are never classified by the logs, it's a very small number and you can go through most of a season without seeing one of these "mystery passes" but they exist as some sort of strange anomaly).

To steer things closer to the original discussion here's a breakdown of outcomes, with a focus on the coverage responsibility of pass attempts. (These numbers are essentially an approximation of the percentage of pass plays in which a particular position player was the "defender" on a pass attempt. I can talk more about the collection of data, but suffice it to say it's the result of over a hundred runs of a single season with no injuries, a locked depth chart, a 4-3 gameplan with no double teaming and corners oriented by left/right, where nothing was altered from run to run other than the isolation of a particular coverage at 100%. There was a mixed control gameplan that was also run, although in many ways the control was less useful as any kind benchmark, conceivably because it was more vulnerable to the random play call rolls of the game, so there ended up being a lot of variance.) To clarify, a player is the "defender" if he is credited with a pass defensed, intercepts the pass, or if the receiver he is covering catches the pass. So here's generally what you might see over the course of a season of pass plays with mixed coverage...

LCB coverage ~ 12.75%
RCB coverage ~ 9.75%
FS coverage ~ 8.5%
SS coverage ~ 8.5%
SLB coverage ~ 9.5%
MLB coverage ~ 9.5%
WLB coverage ~ 9.0% (includes nickelback, although this data less clear)
DL coverage ~ 4.0%
Pass Blocked ~ 3.0%
Pass Hurried ~ 11.5%
QB Sack ~ 6.0%

This adds up to 92% of pass plays, with the other 8% coming from bad throws, dropped passes, and mystery throws.

So that's all well and good, maybe interesting, maybe not, but there it is. Now I'm going to post some of the things about the different coverage schemes that jumped out at me in the results, and as anyone that's tried to run broad tests on the game knows, it's sometimes difficult to draw conclusions, with anything you're seeing there is often only a very subtle effect on stats even with the most extreme adjustment of variables.

2-Deep Zone

- The LCB can see some very high coverage percentages here (there's a blip or two at just less than 17%), but most notable is the gap between LCB and RCB, with the LCB averaging 13% and the RCB 9%

- The SS is involved in more coverage, ~9.5% of plays, while the FS remains relatively neutral at its "base" 8.5%

- SLB and MLB are both slightly less active at ~9.2%

3-Deep Zone

- Lowest coverage for LCB at 12.2%, RCB creeps back up to about 9.5%

- FS is higher at about 9.2%, and SS is also a little over 9%

- MLB has increased responsibilities at ~10.3%, SLB a low 9.1%

4-Deep Zone

- LCB back up to around 13%, but RCB falls to a very low 8.5%

- FS higher at 9.4%, SS around 9%

- MLB back down to 2-Deep 9.2% levels, SLB at average 9.5%

Bump

- LCB at an average 12.6%, and RCB comes back up to around 9.7%

- Both safeties fall to just below 8%

- SLB goes way up to 12.5%, and MLB is about 9.5%

Man

- LCB and RCB are most balanced at 12.2% and 11.5% respectively

- FS is a little low at 8.3%, but SS is a very low 7.4%

- SLB and MLB are both high at 10.6% and 10.3% respectively


So... some conclusions.

- Your LCB is by far the most active DB, and, other than in loose man coverage, he's going to see significantly more action than your RCB.

- If you play bump and run exclusively, your SLB is almost as busy as your LCB.

- 3-Deep stresses your MLB, as does loose man.

- Outside of zone coverage your safeties, especially the SS, can be significantly less active in coverage.

- The differences are not really that huge between different pass coverage defensive positions. Your shutdown LCB is only seeing a few more passes per game than your second RCB guy and linebackers and safeties are relatively active as well.

- Your pass rush, sacks and hurries, on average will account for 17.5% of defensive pass play outcomes (and if it's really awesome, it can top 25%).

- You could try to surmise that different coverages simply distribute pass defense responsibilities differently and that in bump and run a certain route is rolled against the SLB's bump and run rating, when it might've been rolled against the RCB's zone rating if you were in 2-deep, with no core penalty or advantage, just a different rating from a different player. This could be a stretch, but at the very least there is a difference in coverage distribution in the different schemes and that may really be all there is to it.

- I think that could explain away most things, but at the same time, it does seem like there must be some advantage between bump and loose man, or a positioning effect that has some impact. It appears that in bump the SLB has responsibilities that shift to the FS or elsewhere in loose man, presumably to a player further from the line of scrimmage. You would think this would have some kind of effect regarding short passes, but I can't say I saw anything conclusive. (But admittedly, that's not exactly what I was looking for, and I know, I'm just taking things back where they started by going on about this.)

Anyway, that was a lot of somewhat relevant, mostly jumbled ramblings on some testing I've done on the different coverages. Personnel almost certainly make the biggest difference, the rest is just about a few percentage points here or there. Like others have mentioned, if you can get a team that is great at one coverage and keep them together under the cap, that's probably the way to go. I saw no advantages between using specialized or mixed coverages, but again, I was more focused on the distribution to different defenders.

I think the only way to really test what everyone wants to know is to set up a fake mp league with a controlled roster set (maybe 50's for everything), and set up specialized defenses against offenses that throw nothing but short passes, and then nothing but medium passes, then long, etc., that's probably going to be the only way to figure this out. I'm sure there isn't a "better" scheme, but some sort of rock-paper-scissors trade off model. Or like we said, maybe it's just as simple as, bump the SLB's bump gets rolled, in 2-deep the corners zone gets rolled. It's fun either way. :)

[flere-imsaho]

Post of the Year. I salute you, sir. :) Excellent work.

[Icy]

Great informative post MalcPow, now i need to re-read it and look at my own team gamelogs to see if i can give any use to it.

[QuikSand]

That is a great contribution to the knowledge base about this game.

[Subby]

The BnR stuff is really interesting.

Thanks!

[Daimyo]

Awesome! Best post here in a long time.

[Daimyo]

Anyone have any theories why would LCB get so much more action than RCB? Does the game naturally target more whichever of FL or SE lines up on that side? Is the TE lined up on the L defensive side?

[Warhammer]

I would imagine that the LCB also qualifies as the #1 CB depending on how you are running your coverage. That could account for him seeing more passes (more going to the #1 WR).

Otherwise I don't know why you would see that.

[Warhammer]

Standard positioning has the strong side of the offense going to the right, which would be the left side of the defense. But that can be changed in game, so I doubt that would account for the discrepancy.

[Samdari]

MalcPow - from:

"a player is the "defender" if he is credited with a pass defensed, intercepts the pass, or if the receiver he is covering catches the pass."

Where do you get the information about who was covering a receiver when he catches the pass?

[MalcPow]

MalcPow - from:

"a player is the "defender" if he is credited with a pass defensed, intercepts the pass, or if the receiver he is covering catches the pass."

Where do you get the information about who was covering a receiver when he catches the pass?

The "caught" pass coverage stat, means the player that defender was covering caught the pass.

And just to clarify, the gameplan oriented the cornerbacks by formation side, not top wide receiver. The whole LCB-RCB difference is one of the things I don't have an explanation for (but I'll speculate in a minute), it shrinks a little when you go to bump or a lot when you go loose man, but it's still there and it's significant when you're in zone coverage (again, as much as 4% on average in 2-Deep). The funny thing is that the tight end lines up next to the LT in the personnel formation screens (in "WR moves to slot" formations, the TE lines up next to the RT, but in that case he's the only receiver on that side of the formation), meaning the RCB would usually be on the strong side, or the side with more receivers on any play where there are more receivers on one side. This basically leads me to believe that the LCB coverage responsibility is simply larger (i.e. there are more routes the game assigns to his coverage rolls). The LCB appears to be typically matched up against the SE and they're both on something of an island out wide, whereas the FL and TE share the strong side, and the RCB, SLB, and SS probably split more routes among their more crowded coverage responsibilities. I could try to explain why the same doesn't apply as much on the other side with the WLB, and use some marginal "football" ideas, but that's probably not that productive. I think the basic theory I have is that the LCB has more routes to cover out there (presumably anything the SE does that isn't deep or over the middle in a zone) while the RCB is in a more crowded space, with more defenders nearby, and so some routes are picked up by somebody else. I don't think it necessarily has to do with more throws on one side or the other. But this is all really just speculation, the only thing I really know is that the difference exists.

[Samdari]

The "caught" pass coverage stat, means the player that defender was covering caught the pass.

And just to clarify, the gameplan oriented the cornerbacks by formation side, not top wide receiver.

Ah, ok, I thought you were somehow getting the data from the log file, and thought I was missing something.

[Vinatieri for Prez]

Well, I think often in the log file whoever makes the tackle on the pass is the guy who had the pass caught against him. So that helps with looking at individual plays (e.g. who exactly was in coverage on the deep ball that was caught -- provided it didn't go for a TD and thus no tackle). Of course, in just judging gross numbers, the easier way is to simply check the caught passes stat.

[Celeval]

It could also be a balance of offensive formations - for 2 TE formations when there is only one wideout on the field, for Slot formations when both wideouts are on the same side of the field, etc.