A. Building plays- "the gimmicks"

Have you ever wondered how the other coach's play totally bewilders your defense? For example: it's late in the fourth quarter with your team ahead by 4 points, 2nd and 5 to go, the opposing quarterback fakes a hand-off, then drops back to deliver a bomb to a man wide open in the field. Oh my... TOUCHDOWN!!! You lose! Sound familiar? How did it happen? Well, don't feel too bad (it's happened to us all!), the one thing to keep in mind is that every successful offensive play has what I call "the gimmick".

Every good running or passing play has one. Really! In fact, in passing plays "the gimmick" is the essential feature of the play that makes it work.

Oh sure, you say, running plays use gimmicks? The other coach just lets his running back (RB) go from the line of scrimmage and gets 5 yards every time. Where's the gimmick in that? Well, if you look more carefully at the entire play you will find "the gimmick". For example, does the RB line up on the strong side of the field but dive to the weak side? Does he pause before making his run, or dash out from behind a fast guard? Do the wide receivers (WRs) run backwards or sideways off the field drawing their man to man coverage with them? (Personally, I wish this were not allowed since it's most unlike real Football- unless of course your team has a bunch of wimps for receivers!) [ed. note: Most leagues now do consider this illegal]

At the line of scrimmage watch the offensive linemen. Do they draw the defensive line into the offensive backfield before the RB starts, or do they push the defensive line back giving the RB room to run? All these ploys are examples of "gimmicks" that are used to make plays more effective. Below I give some examples of gimmicks and their uses in offensive plays. It should be noted that the really effective plays often use several gimmicks in the same play.

Weak/strong side attack.

The advantage to paying attention to strong/weak side is that defenses that stack their men to one side can be exploited by attacking the weak side of the field. It's not always immediately apparent which side is strong, but the sure way to find out is to set up a defensive play with a linebackers (LB) in zone coverage on the strong side and look for that player on the practice field.

As mentioned above, a RB on the strong side who runs to the weak side is an effective play since the coverage is more vulnerable on the weak side, and the offensive linemen can block LBs pursuing the RB. Usually the RB makes a dive play to the weak side hoping to break free from the line of scrimmage and out-distance his pursuers.

Another play that exploits the weakside is a quick pass to a receiver sprinting up the sideline. Here a RB or WR is usually matched-up in man coverage with a slower defensive back (DB), who the receiver can usually out-run and make the catch.

In order to exploit the weak side gimmick fully, you should also make complementary plays that attack the strong side of the field e.g. using sweep or screen plays. Assuming your weak side offensive plays are successful, the stacked strong-side defensive plays will be called less often, which will then permit an overbalanced rushing attack on the strong side of the field.

Movement away from the action.

As described in the above running play example, receivers running away from the site of action can be very effective in drawing man and zone coverage out of position. Overall, this type of gimmick works best against man coverage and LBs reading RBs. In order for this gimmick to be really effective the run/pass play should be geared towards a talented player who can get the most yardage on his own, because his team-mates (along with the defense, hopefully!) are out of position.

For example there's no point in passing the ball to a slow tight end (TE), unless he's totally uncovered deep, since once the pass is caught the defense will catch him quickly. On passing plays, receivers can run deep in order to fill the zones allowing a trailing receiver to become open. Alternately, they can run across the field to take man coverage with them, and open the side of the field they came from for a running or passing play.

Movement behind the line of scrimmage is important if you want to take LBs reading RBs out of the target area, either on a pass or on a running sweep. Set the backs in motion in the opposite direction to where you are targeting the pass or run and the LBs (either reading RB or in man coverage) will follow briefly in that direction, again opening the field up for your offense. This is particularly effective for quick slant passes.

Pauses.

In the first issue of the newsletter there was a very good article on the use of pauses in offensive/defensive plays. A receiver can be given a pause at the beginning, during, or at the end his route to cause confusion in both man to man, and zone defense.

Pauses at the beginning of the route allows other receivers to "occupy" zones allowing the trailing primary receiver to become open. Pauses during his route allows a fast receiver to momentarily break from man coverage to make the catch. Pauses at the end of the route serve to occupy zones and to create downfield blocking as described below. Pauses can also be used with men not the target of the pass.

For example, two receivers can converge in one area with one man pausing at the site of convergence. The second receiver runs through the paused man who then acts as a "blocker" in man coverage. This is an effective and legal way of creating a "pick" on a passing play.

RBs can be assigned pauses before beginning their running route. Pausing RBs causes man coverage to close in toward their man. Since the RB is usually positioned in the backfield, this creates an opportunity for the offensive linemen to trap the pursuing man coverage, giving the RB a head start once he gets the ball.

Pauses can also be assigned to linemen either for screen passes or draw plays. Pausing linemen keep the defensive line at bay without actively blocking. After the receiver/runner gets set during the pause period, the offensive line can then be activated using the continuos blocking symbol to create lanes for the receiver/runner to work through.

Crossing receivers.

This is a similar notion to the idea of men converging with pause signals, except that the receivers cross paths without pausing in an effort to "brush off" or confuse the coverage. This gimmick works best with man coverage of receivers. The timing on these patterns is crucial and many attempts are usually needed to coordinate the receivers. Receivers may follow similar routes before breaking apart. Alternately, they may come from opposite sides of the field. Using a deep crossing pattern you can get a man open behind short zone coverage.

Overloading zones.

While on the subject of zones, you can use several receivers lined up on one side to overload zone coverage. Receivers can either pause in their running routes, or run in divergent directions to confuse the coverage. Set passes up to cover 10, 20, 30, and 40 yard gains using this gimmick and test them against your opponent's defensive plays to find out where the zone coverage is weakest.

Remember, you can pass effectively up the middle of the field too, especially if your target receiver weaves his way through different zones.

Linemen blocking.

You can momentarily confuse defensive linemen and LBs by using a variety of offensive blocking at the line of scrimmage. For example, a run block can be assigned to linemen on the side of a passing play to confuse LBs who are reading the linemen before dropping back to cover their zones. This may buy you a little extra time for the receiver to get open. This type of play seems to work best with receivers streaking down the weakside sideline (weak side-another gimmick, right!).

Running plays can use pass blocking away from the RB. (Pass blocking near the RB is not a good way to gain yardage!) This type of blocking tends to draw pass rushing defensive linemen into the offensive backfield, thus allowing the RB more breathing room from pursuing linemen.

Pass blocking can also be used (as currently allowed) to block pursuing man coverage. A receiver can start in the backfield (or run across from one side) through the pass blocking linemen. Men covering the receiver will get trapped by the offensive linemen behind the line of scrimmage, thereby freeing the receiver. This type of play is very similar to a screen pass, except the linemen use pass blocking exclusively. Receivers can also spurt from the backfield through the pass blocking linemen in an effort to "tangle" LBs with DBs.

Play fakes.

Both pass fakes and play action fakes work to momentarily draw defenders away from the target zone while the receiver/runner gets into position. Play action fakes work best with the "hot" RB, so you may want to wait until the second quarter or half before exploiting this type of play.

For best results have your RB move behind the line of scrimmage away from your target passing area. LBs reading RBs will follow this movement which leaves an open space to fire in your pass.

Pass fakes seem to work well almost all the time and tend to freeze DBs covering the target receiver. You can direct your pass fake to another secondary receiver (it helps in man coverage if that receiver is the "hot" receiver), before delivering the ball to the intended receiver.

Alternately, you can "pump fake" to the primary receiver then drop the quarterback (QB) a few paces before passing the ball to the primary receiver. This pump fake has the same affect as a pausing receiver since the DBs usually freeze allowing the intended receiver to break from his coverage.

Flea-Flickers/Option passes.

These types of plays invariably incorporate several "gimmicks" in the one play. At their best these types of play are wonderful to watch; the interplay of the whole team is finely coordinated in order for the play to work. Expect to spend a lot of time mouse clicking to perfect these type of plays.

Fleaflickers can be used to good effect especially if your RB is "hot". The RB takes the snap and runs with the ball with defenders pursuing. The RB then laterals the ball to the waiting QB who then passes to a receiver. The receiver's route can incorporate pauses/zone overloading etc. to maximize the chance of getting open. The RB's run should be made to look like just that- a running play. Linemen are in continuous blocking routes (except for those needed to protect the QB).

On the option pass, the QB dumps the ball to the RB, and right away all the offensive linemen chasing the QB in pass rush are taken from out of the defensive play. The RB then runs across the field with run blocking (another gimmick- right!) before throwing the ball to the intended receiver. Having the linemen in run blocking mode gives the RB some protection (this is the option part!) if the play is busted by a defensive player breaking into the backfield. For best accuracy on the pass the RB should pause or pass fake before throwing the ball.

Zone coverage can be disrupted by either of these plays if the passing player approaches the line of scrimmage before releasing the ball (see below).

Passer running towards LOS.

This tactic is effective for passing plays against zone coverage. As the passer (QB or RB) approaches the line of scrimmage, defenders in zone coverage move toward the player. The closer the passer gets to the line of scrimmage the more likely the defensive players will leave their zones to try to tackle that player. This leaves the deep coverage weak and allows a receiver to get open. This type of play is best used with deep passes, so get good blocking set up for your QB or RB before the pass is made. When engineered well, this play can be a "make or break play", large gains can be made against mediocre defenses, or severe sacks taken if the defense is well prepared.

Final thoughts.

These are just some of the gimmicks used in PMFB, and this is not a detailed list of what goes on "behind the scenes", but more an overview of what to try in your own plays. Next time your team is handed a spanking by another team, watch carefully as their plays unfold. In your mind, isolate the players and see what each does in the play. Then, and only then, will you appreciate the gimmicks used by the other coach.

B. Evaluating plays-reward frequently and well.

Having designed your new plays, how do you set up the Artificial Intelligence so that successful plays are called most often? This may sound an obvious question, but considering that the average playbook may contain 15-20 offensive plays, and that your team will only get 60-80 offensive plays per game, the computer coach needs to find the winning combinations in a hurry!

In order to find the most successful plays as quickly as possible I use a two-fold approach in the Artificial Intelligence: reward the play frequently, and reward it well. The overall result of this reward strategy is that as the game progresses, the Artificial Intelligence will quickly work the playbook into a hierarchy of plays where the most successful plays are called for each situation. Scoring may be slow in the first half as different plays are tested, but the yardage should come in the second half after the Artificial Intelligence has sampled enough plays.

Evaluate the play every time.

When you create a new playbook the plays are set to be evaluated every two times that they are run (the default sample number is set to 2). This means that if you have a terrific running play which gains 10 yards every time (don't we all wish!) PMFB will not increase the play's priority until the next time the play is called. This could be a long time later if you use a large playbook. In addition, after the first increase in priority the play has to be run twice more before the next increment is added to the Artificial Intelligence.

Since PMFB is a dynamic simulation where situations change constantly, I believe instant evaluation of plays is necessary to weed out your ineffective plays, thereby giving your team a better chance of finding the winning game plan.

So, to speed up the computer's analysis of my plays, I set the sample number in the Artificial Intelligence box to 1. This means that the simulation will evaluate my play every time it is tried on the game. This instant evaluation of my plays rewards my effective plays immediately, which results in their more frequent usage. It also means that the simulation will be able to try a variety of plays under different situations, and to sort my playbook more quickly into a hierarchy of good plays. The overall effect is that the most effective plays are called against the defense, while the unsuccessful plays are less likely to be tried repeatedly.

It's possible that a normally terrific play can be overlooked using this method of instant evaluation, but this usually occurs only when your other plays are successful against the defense- an enviable position to be in where your "best" play is not required to beat the opposition!

Changing priority Artificial Intelligence.

In addition to instantly evaluating the plays, I also make sure to significantly change the priority Artificial Intelligence when a play meets or fails expectations. This is important in order to "positively reinforce" the simulation to call the good plays again, and for lowering the priority of a busted play.

For example, I have a running play with a priority of 4. A gain of 4 yards is good for this play, three yards is acceptable, but 2 yards is not getting the job done! So with the Artificial Intelligence priority is set to 4, I will increase the priority by 2 if the play gains 4 yards but decrease it by 2 if it only gets 2 yards. This way the play is rewarded well if it meets expectations and will be called again in a similar situation.

Conversely, assuming I really needed those 4 yards but got stuffed by the defense, I want the Artificial Intelligence to try another play next time. Overall, if this play is successful 2 out of 3 times then I will have gained 8 yards or more, plus the Artificial Intelligence will be favorable (with a net increase of 2 in priority) to call the play again.

Changing the Artificial Intelligence more dramatically (e.g. by plus or minus 4, in the above example) will alter the play hierarchy more quickly, but can also greatly decrease the chance that a play will be sampled a second time against a susceptible defense.

I think it's more important to challenge the defense with a small number of good plays than it is to sample a large number of plays (see below). Every play will usually gain some yards against a given defense; it's just a matter of luck in the timing of the play calling.

One thing to bear in mind is that the play calling by the simulation is never infallible. Be prepared for some "off the wall plays" to be called in certain situations. Sometimes they work and other times...

C. Building a Playbook

OK, so now you've experimented with "gimmicks", made some plays that are effective on the field, and have set up the Artificial Intelligence to evaluate the success of your plays. How do you go about putting a playbook together? (For purposes of this discussion I'm going to ignore 4th down trick plays and 2 min. drills- the latter was covered by Bill Hall in the first newsletter.)

When to call a play.

This aspect of the game depends entirely on your own individual style of coaching. If you are a conservative coach you may want to call a lot of running plays on first down. This style of play would be designed to guarantee your team of facing no worse than 2nd and 8 yard in every series- a solid position to be sure!

However, by calling running plays on most first downs your team will gradually get fewer yards since the defense will learn to negate these plays through it's Artificial Intelligence.

Alternatively, you may be a free wheeling kind of coach who likes to shatter the opposition with big yardage bombs. Again, however, if you continually call these plays, the defense will finally wise up and shut down your offense.

A versatile/flexible playbook.

If you fall somewhere between the two extreme cases described above, you may be interested in my strategic approach. I'm an advocate of a "versatile" or "flexible" team playbook; that is, a playbook that contains several plays that can gain the required yardage using different gimmicks on both running and passing plays. This is especially effective if different gimmicks can be run from the same play formation since a particular defense will set up exactly the same way for the different plays, and should be susceptible to at least one of the gimmick plays.

Also, I try to work the Artificial Intelligence so that there is little chance of the same gimmick play being called consecutively. That may sound difficult, but it's not! This type of strategy is well suited towards a static playbook since it does not rely on scouting the opposition (provided you have a sound defense- more on that another time!). This is because the simulation will have several choices of plays that use different gimmicks at any given situation. Since no defense is infallible (there's always a weakness- even if it's to a QB bootleg/sneak), your offense will eventually find the plays that move the ball up the field.

So how do you go about this type of strategy? First, design 3-4 good plays that take advantage of different gimmicks to get the following yardage: 2-4, 5-10, 10-20, and greater than 20 yards. That makes between 12-16 total offensive plays (excluding 4th down situations etc.). This may not sound like many plays, but it is more than enough for the simulation to sample at any given time. (You can make more plays, but the likelihood of them being sampled multiple times decreases- remember one successful play per down is all you really need).

You can construct a very competent playbook from only a handful of plays e.g. 3 running and 3 passing plays, and these type of "minimalist" playbooks often do very well because the plays are tested often and rewarded frequently.

After you have made your plays set them all to equal priority (e.g. 4) for each down (except 4th down). For 1st and 2nd down yardage situations fill in all of the yardage boxes so that any play can be called (we'll change that later). For the 3rd down yardage row, mark in the yardage that you expect to gain for each individual play. (For all the plays be sure to set the position of the field according to the logistics of the diagrammed play, i.e. don't call a 30 yard pass at the goalline!)

On first down, let the simulation try about half of your total plays against the defense. To do this decrease the priority (e.g. to 2) for those plays you do not want to call. Since the plays are already set to cover any yardage situation, the effect of this Artificial Intelligence will give variety and versatility to your attack, because any yardage play (short, medium, or long) will be called on the first down. This will keep the defense out of any "rhythm" against your offense.

Depending on your coaching style you can include more short or long yardage plays in this group. If you are more conservative, you can incorporate more short yardage running plays on first down; if you are more adventurous add more deep passes.

Now, on second down, increase the Artificial Intelligence for plays more likely to succeed for the remaining yardage. Again, the best way to accomplish this, (and it's written in the manual) is to not call the plays under certain situations. Keep the plays set at equal priority (e.g. 4 as above) but clear the yardage boxes to avoid certain extreme plays being from being called.

For example, clear the 15-25 yard boxes for most of your short yardage plays, and clear the 0-5 yard boxes for most of your long yardage plays. There is little sense in calling call a 2 yard play if there's 20 yards or more to go! This way, if you are left with 2nd and 20, the Artificial Intelligence will call plays designed to get at least five yards. But to keep the defense honest you should include a short yardage situation for 2nd and 10, and conversely, call a few long yardage situations for 2nd and short!

This type of variation is essential to keep the defense off balance: an inside run can be very effective against a defense that's prepared for 2nd and long, while a long bomb may unglue a defense looking for 2nd and 2 yards to go.

Finally, for third down increase the Artificial Intelligence for plays designed to get the remaining yardage or better. On third down, you must become more focused on getting the remaining yards; so call your best plays for the given situation. If you remember from above, the yardage Artificial Intelligence of each group of plays was previously set for third down situations, so if you leave the priority Artificial Intelligence set at equally at 4, then the appropriate group of plays will be called at third down. If you have 3-4 plays that can get the required yards, then at least one of these should work during an offensive series. Don't be afraid to include a 5-10 yard play when you need 10-15 yards, it's surprising how a play designed to get 5 yards sometimes gets 10 yards or more. Even if the play gets only 7 out of the 10 yards needed, that leaves you with 4th and 3 to go- a better situation than 4th and 10 when you are trailing late in the 4th quarter!

Final tweaking.

Now you have the basic playbook which will give your gameplan spontaneity, and hopefully success, on offense. If you have an absolute favorite play e.g. for 3rd and less than 1 yard, or 3rd and more than 20, then include them in the playbook and give them a much higher priority (say 8, if the others are set to 4). There's no guarantee they will be called every time, but unless your play is a real killer you probably don't want it called all the time.

You can also make a special "comeback offense" by setting up some good plays to be called only when you trail by 21 points or more. Just clear the Artificial Intelligence boxes corresponding to minus 21 points or greater from your regular plays, and leave them filled in on your "comeback" plays. This way only your long yardage or special teams plays will be tried when you trail by 21 points.

Conversely, you can make "stay ahead" plays which will grind out the clock when you are ahead in the game. Just clear the plus 21 point boxes in the Artificial Intelligence of your regular plays, but leave them filled for your solid running/passing plays. This will result in your team changing offensive strategy as the game situation progresses. Of course, all of these are just suggestions; you may want to pour it on once you are ahead.

And there you have it. A versatile playbook, which will sample your best plays often and with "intelligent" randomness. Actually, I have found that PMFB calls my plays more successfully than me, even though I created the Artificial Intelligence. Humans tend to be biased, hot-headed creatures who act impulsively at times; the computer acts coolly under pressure, and remembers the success/failure of the plays. I hope this has been helpful, and good luck to you all, (but please, not too good when your team meets mine!).

* For more articles like this one, check out the Coaches' Handbook on PMFB.net.