It started with me thinking about the holding calls against the offense. How could I incorporate those instances where the holding occurred downfield and the 10 yd penalty applied from there, instead of from the line? But the more I looked into it, the more I realised that all of the penalties needed overhauling. So I looked at the 2013 season (it’s one where I have every playbook), and over a few months recorded every single penalty … 3419 all up.
on pass plays, the most frequent penalties were (cf. APBA’s) …
false start o 21% ( 3%) FAIL
pass interference d 12% (14%)
holding o 9% (14%)
holding d 9% ( 8%)
unnecessary roughness d 5% ( 3%)
neutral zone infraction d 5% (nil) FAIL
this compared with APBA’s (cf. ‘actuals’) …
holding o 14% ( 9%)
pass interference d 14% (12%)
holding d 8% ( 9%)
illegal contact d 8% ( 2%) FAIL
illegal formation o 6% ( 1%)
intention grounding o 6% ( 2%)
offside d 6% ( 1%)
pass interference o 6% ( 3%)
roughing passer d 6% ( 4%)
on running plays, the most frequent were (cf. APBA’s):
false start o 34% ( 3%) FAIL
holding o 20% (25%)
neutral zone infraction d 8% (nil) FAIL
delay o 8% ( 3%)
unnecessary roughness d 4% ( 6%)
encroachment d 4% ( 8%)
this compared with APBA’s (cf. ‘actuals’)
holding o 25% (20%)
offside d 14% ( 2%) FAIL
encroachment d 8% ( 4%)
holding d 6% ( 1%)
illegal block o 6% ( 1%)
illegal motion o 6% ( 0%) FAIL
facemask d 6% ( 2%)
unnecessary roughness d 6% ( 4%)
unsportsmanlike d 6% ( 0%) FAIL
so you can see that the distribution of penalties is way off. In particular, the failure to recognise that the most common penalty – False Start – is not represented appropriately bothered me. Of course a lot of it is a result of trying to squeeze in all the offense and defense fouls into 36 possible outcomes – it’s a hell of a lot better then the previous method which only had 11 outcomes – but it’s still nowhere near close enough (for me).
So how can it be done better? Simple – introduce 1 more die into the penalty roll. Same roll, just with 1 more die. That gives you 216 possible outcomes and you can model the distribution much more accurately.
Another thing I thought a lot about was the pass interference penalty. That 12% included all types of passes – long, medium and short passes – and we all know that there are a LOT more flags thrown on the long ones than the shorter ones. So I introduced different distributions depending on what type of pass you threw:
short & screen (S)
so now we have the following distributions for defensive pass interference:
Using these with conventional pass distribution usage will come pretty close to the actual figure.
I’ve looked at all of the penalties and redone them (mostly) with the extra die.
I changed some of the existing penalties to new ones as well:
1.Clipping – Offense is now 1.Facemask – Offense (clipping hardly ever was called, and there was no Facemask o)
5.Grasping Facemask – Defense is now 5.Neutral Zone Infraction – Defense (Facemasks are now 15 yarders when called, and there was no NZ Infraction d)
8.Illegal Block – Offense is now 8.Illegal Block Above the Waist – Offense (this one is 10 whereas all the other illegal blocks are 15 yarders)
15.Ineligible receiver downfield – Offense is now 15.Illegal Use of Hands – Defense (Ineligible Receiver d/f is now a 5 yarder and can be covered by Ineligible Player d/f, and there was no Illegal Use of Hands d)
30. Twisting Facemask – Defense is now just 30. Facemask – Defense (‘cos they don’t differentiate between grasping, twisting, turning, pushing or pulling; since 2008 it’s all just a 15 yard Facemask)
new penalties are:
36.Horse Collar Tackle – Defense
37.Illegal Touch of Forward Pass – Offense
38.Twelve Players on the Field – Offense
39.Illegal Chop Block – Offense
40.Invalid Fair Catch signal – Offense
41.Illegal Substitution – Defense
42.Delay of Game (after play) – Defense
43.Illegal Cut Block – Offense
44.Illegal Peelback Block – Defense
45.Delay of Game (after play) – Offense
46.Tripping – Defense
48.Taunting – Defense
49.Taunting – Offense
50.Offside – Offense
51.Delay of Game – Defense
52.Illegal Peelback Block – Offense
53.Illegal Touch of Kick – Offense
54.Player out of bounds on Kick – Offense
55.Illegal Formation – Defense
56.Illegal Blindside Block – Offense
57.Illegal Touch of Kick – Defense
58.Illegal Low Block – Offense
59.Illegal Low Block – Defense
All of these penalties will now be called more accurately according to their actual distribution.
Also I’ve done a ‘Special Penalty’ table for offensive holding, illegal-use-of-hands, and illegal block penalties, as well as live ball Personal Fouls, Unnecessary Roughness and Unsportsmanlike penalties by the offense. This table can be used when these penalties are called on running plays, passing plays for gain, and punt & KO returns, where the foul occurred either over the line or during the return. In these situations, the penalty is assessed from the point of the foul, not the line of scrimmage or point of possession. These are all based on the actual statistical spread of penalty enforcement locations.
And, I’ve introduced a No-Penalty result on kickoffs, because there just aren’t that many flags on the kickoff. Including kickoffs-out-of-bounds, I counted just 46 penalties all 2013 season on KO’s (I’m not talking about returns here, just the KO part). With 2684 kickoffs and (typically) 3 penalty numbers on KO cards, the cards would otherwise produce over 220 penalties for the season!
Similarly, on punt returns, the opposite occurs – there are actually a lot more penalties than the cards can produce. Typically there are 3 penalty numbers on PR cards, but there should be more like 7 to reproduce the actual number of penalties. I fix this by rolling an extra die on the punt return roll. If it is a 1, there is an automatic flag on the play. This would result in a 24% penalty percentage – that’s too high (it should be about 20%), so I introduce some No-Penalty TQ rolls to correct the percentage to reproduce the actual figure of just under 20%
I also separated KO return penalties (as TKO) from TK because they were completely different in their type & distribution. And with typically 3 penalty numbers on KO return cards, they also generated too many penalties.
Lastly, you may notice that I do pass interference penalties different to APBA and others. First, I think you should be able to decline the penalty and accept the completion. I don’t believe it skews the stats that much and it’s much more intuitive to me. Second, on incompletions I use the interception identifier columns to determine where the pass interference occurred. It seems more logical to me to do it this way.
Hope you find these useful. Feedback (and corrections) welcomed.