All Forums > What do you expect?
8/9/2013 8:31 AM
I am curious about something, and I mostly am pointing this question at the "no randomness at all" point of view, but anyone should feel free to chime in.

The scenario
Evenly matched offense and defense -- all player attributes are the same, let's say 50.
Offense is set for Pro Set, balanced.
Defense is set for 4-3, balanced.  Since we're talking the 3.0 engine here, pass coverage is set to medium, and no blitz.  All three LB are set to "LINE"

The sim selects an inside run on offense, and "run" on defense.   What is the expected result?

The sim selects an inside run on offense, and "pass" on defense.  What is the expected result?

The sim selects an outside run on offense, and "run" on defense.   What is the expected result?

The sim selects an outside run on offense, and "pass" on defense.  What is the expected result?

Now let's add some "variable" -- same as above with ILB set to "COVER"
And now let's set one OLB to "COVER" and the ILB to "LINE"
And now let's set one OLB to "COVER and the ILB to "COVER"
And now let's set both OLB to "COVER and the ILB to "LINE"
And finally for this scenario, let's set all the LB to "COVER"

Would the results be different if all of the player attributes were 25?   Or 75?

8/9/2013 11:24 AM
Bob, in my opinion (for what it's worth) in each of the scenarios there should not be A result but a range of results.

For example, equal teams + appropriate defensive formation  + no variables = play gains between A and B yards (I don't think it fair the sim's run/pass call on defense significantly affect the result of the play unless pass coverage is set to short or long)

Equal teams + inappropriate defensive formation (ex.:nickel vs. wishbone or box) + play run to strength of offensive formation (run) + no variables = play gains between C and D yards

Changing LB assignments should affect the OL/DL equation, as the 'cover' and 'line' settings add or subtract personnel primarily stopping the run or defending the pass, in each case modifying the appropriate range of results. If an OLB is set to 'line', it would only make sense that it should be harder to break an outside run, again modifying the range of results. If all LB are set to line, the range of running results should reflect additional run stoppers.  If all LB are set to cover, a running play's range of results should be modified to reflect all LB playing off the line of scrimmage.

The main driver of results, the initial setting of the ranges, should depend on talent. A three point differential between OL and DL should yield one range of possible results, while a ten point differential should substantially limit the potential for a lower end result and extend the possible higher end of the range (something like one percent chance of 0 yards gained versus twenty percent chance of gain of 10 yards or more). The talent level of LB, DB, QB, RB and WR would also have to factor into the equation - determining the appropriate range of results, weighting the calculation, and settling upon a result within the range.

I don't know if that's helpful or not.
8/9/2013 2:23 PM
Interesting and appropriate question Bob. I also like much of bjaygee's response. I like the current idea of steps or sequence of a play that Norbert designed, but his formation of the bucket distribution was way off (which I feel is the main reason we are having such trouble in this game). I also feel that if we can't control more specifically a run vs pass defensive call it should mean little to the outcome. LB settings (since we can control them) should be a factor in the outcome. DB behavior should be more dependent on a run vs pass DEF call.

The examples below would all fall into the STEP 1 category -

The sim selects an inside run on offense, and "run" on defense.   What is the expected result?
OL (all 5) + TE + FB = 7 blockers vs DL (+ LB set to "line")  = 7 tacklers - since even in this scenario expected Ave = 0 with range for this step -2 to +2 - Remember this is for step 1. DB depth and DEF "run" call would influence next step as DB playing run.
The sim selects an inside run on offense, and "pass" on defense.  What is the expected result?
Same as above for step 1. Step two outcome depends again on DB depth and "pass" call as DB playing pass.
The sim selects an outside run on offense, and "run" on defense.   What is the expected result?
Same
The sim selects an outside run on offense, and "pass" on defense.  What is the expected result?
Same

Now let's add some "variable" -- same as above with ILB set to "COVER"
And now let's set one OLB to "COVER" and the ILB to "LINE"
And now let's set one OLB to "COVER and the ILB to "COVER"
And now let's set both OLB to "COVER and the ILB to "LINE"
And finally for this scenario, let's set all the LB to "COVER"
Same formation - still step 1 - DEF variables as above + OFF variables of TE or FB  "block or into pass pattern"  changes #'s of blockers vs tacklers. With even ratings the OFF or DEF with greater needed ratings would gain an advantage in the play, skewing the average towards the OFF or DEF based on appropriate positioning settings by the players involved. Ratings advatages for OL or DL for even #'s of players would skew the average up or down and perhaps a 10% OL advantage now makes the average 1 with a range of -1 to +3 for STEP 1, with higher differences further skewing it to the OL favor - same for DL only the opposite direction.

Formation advantages or disadvantages would show up as player # differences for that play - and could be reflected in the PBP for us to follow. "Jones takes the handoff and rushes inside. FB and TE forward to block. LB fall off in coverage. Offensive push forward and Jones gets past the line." or "Jones takes the hand off and rushes inside. FB forward to block and TE rolls outside in pass pattern. LB forward to the line and pushes through the line to the backfield."

STEP 2 would play off STEP 1 and would include extra blockers or tacklers not used in STEP 1 to continue on. Confrontation then would be RB skills vs LB/DB skills and field position for further positive or negative outcome working on STEP 3. Step three would be more of a one on one confrontation for the tackler and ballcarrier. If the ball carrier loses STEP 3 a tackle is made. If the ball carrier wins STEP 3, depending on the field position determined by STEP 1 and 2 the play continues downfield until a tackle is made or points scored.

8/11/2013 7:53 AM
I'm surprised there hasn't been more response to this thread.  It's not a trick question or a trap.  :)

I would point out, however, that any answer that includes "a range of results" (which, by the way, is what *I* would expect to be the outcome) must necessarily include SOME kind of random generation to determine where in that range the result of the play falls.
8/11/2013 10:14 AM
Posted by bhazlewood on 8/11/2013 7:53:00 AM (view original):
I'm surprised there hasn't been more response to this thread.  It's not a trick question or a trap.  :)

I would point out, however, that any answer that includes "a range of results" (which, by the way, is what *I* would expect to be the outcome) must necessarily include SOME kind of random generation to determine where in that range the result of the play falls.
If talent is the primary determinant to the range and weighting of possible outcomes, I don't think anyone should have an issue.  But when the effect of talent disparity is artificially muted (hello, normalization) to give lesser teams a better opportunity for success AND unexpected results are justified by suggesting that a computer generated team inexplicably "played over its head", the strategic side of the game is undermined and coaches leave.
8/12/2013 11:23 AM (edited)
People often confuse randomness as used in a simulation to generate results, with randomness as in "no apparent cause and effect".  The latter is something that, IMO, should be avoided - I don't want the chance of a "Bad Day" or "Playing Over Their Head" programmatically included in a simulation.

Now, statistically it's possible that I "have a bad day" when it comes to results, in that I "hit" on the lower portion of the results significantly more than I do the higher portion, or that I just "get lucky" and hit on the high side much more, but those really should be few and very far between.  And even then, the difference in attributes should be the primary factor of where the range starts and begins, modified in a smaller way by formation matchup and play settings.
8/12/2013 6:54 PM
Isn't that what we have with bot versions right now? Did they ever fix the TE issue? I'm seeing the same results in v3 that I do in v2.
All Forums > What do you expect?

Popular on WhatIfSports site: Baseball Simulation | College Basketball Game | College Football Game | Online Baseball Game | Hockey Simulation | NFL Picks | College Football Picks | Sports Games