The Analytics Debate

[jonboy79]

On the second 2 pt try, I am hearing the concern that we needed to go from 30-19 to 30-21, in case KC scored another 7 to make it 37-21. In which case we could tie with 2 Tds and 2 more 2-pt'ers.

There was 12:22 left at the time. That's going to likely be a grand total of 4-6 possessions (2-3 each) left in the game.

The 37-21 scenario would have required exactly that: KC score 1 TD and 1 XP. The Ravens score a TD and get 2. THEN the Ravens stop them cold. THEN the Ravens score ANOTHER TD and ANOTHER 2-pt. Hold again, go to OT.

If they kick the XP and make it 30-20, the Ravens are banking on 2 holds, then a TD, XP, and a FG to tie and go to OT. Or they could opt for a 2 pt on the TD and go for the win.

That said, 30-21 ALSO sets them up for a win if they get the 2 holds, TD, XP and FG.

If they fail (which they did) they're down 11, now they still need those 2 holds, a FG, TD and 2-pt. A KC FG puts them down 14, but they can get two TDS and 2 EXPs to tie (or go for 2 and try to win). The only scenario that screws them by failing on that 2-pt is the KC TD (37-19) putting them down 18. They weren't likely to outscore them 16-0 from that point on anyway, I suppose, so very little opportunity, if any really, was lost in not making that 2.

So maybe I just talked myself into agreeing with that decision?

Basically... the real key is this. Assume you can stop Kc and they don’t score again.
30-19. Two scores. Two tD’s to win, TD 2 pt and FG to tie
30-20 two scores, two tds wins, TD xp FG ties, TD 2 pt is for win/lose no chance of tie if going for 2.
30-21 two scores, TD, XP, FG wins. So in this case, the XP is tie vs win, no loss scenario

8in scenario 2, on the road, we would go for 2. If not, we would have taken the point. So since we will go for two at least once, we might as well go for it early, because it leaves more options to escape the game without a loss. It’s the aggressive win call, but it’s the right one.



Sent from my iPhone using Tapatalk

[jonboy79]

An actuary will tell your smoking reduces someone’s life expectancy. Just because George Burns surpasses 100 doesn’t mean insurance companies should take away the penalties for health and life insurance based on decades of statistics.


Sent from my iPhone using Tapatalk

[PTORaven]

so this analysis from 538 backs up going for 2 down 11 in the fourth quarter: https://fivethirtyeight.com/features...or-2-for-real/

[Matchstick]

From PFF:

RAVENS GOING FOR TWO DOWN 11 WITH 12:27 LEFT IN THE GAME
The Baltimore Ravens, down by 17 points on the road to the 2-0 Kansas City Chiefs, scored a touchdown on a one-yard run by Mark Ingram after an extremely unlikely conversion of a fourth down by Lamar Jackson and Seth Roberts.

Down 11 points, John Harbaugh opted to go for two points and was unsuccessful. Unlike their attempt after a first-quarter touchdown, this try was not elicited by a pre-snap penalty by the Chiefs. They went for it because of the math. Qualitatively, going for two points on the first in a sequence of potential scores means that you know what you need to win or tie a game on subsequent scores and can plan accordingly. For example, had the Ravens converted the two-point conversion, they know a priori that scoring a field goal and a touchdown with a conventional PAT with the game outright, instead of leaving similar decisions to the end of the game. Mathematically:

After missing the two-point conversion attempt: KC: 94.3% to win, BAL 5.7%

If they made the two-point conversion attempt: KC 89.1% to win, BAL 10.9%

If they would have kicked the PAT: KC 92.8% to win, BAL 7.2%

Hence, the benefit of making the two-point conversion over kicking the PAT was 3.7% win probability points, while the loss via missing the two-point conversion over kicking the PAT was just 1.5%. Assuming the Ravens are a modest 50% on two-point conversions, this is easily the preferable decision. In fact, it would only cease being the preferable decision if the Ravens were under 41% making two-point conversions, a futility that would not be expected with a quarterback like Jackson and the Ravens’ offense we would give a 57% chance of converting.

[Kyle Cactus]

Didn't Billick do that analytic stuff with his computers etc and now we're at it again. Football is a game of heart with the mind following not the other way around.Use creative thinking and will not some formulaic system.Damn they're taking the soul out of the game.Those old time QB's Graham,Unitas,Starr etc would hate this stuff.

Billick always thought he was the smartest person in the room. If he was using "analytics" it was probably something he made up himself.

[JimZipCode]

We don't know that for sure. What is fact is playing the game based on analytics didn't get them the win. Imagine the boost of energy the Chiefs get when their defense stuffs Lamar on 4th in his own territory. In any scenario I'd rather Mahomes have to go 80 than hand the ball to him on our 47.

That's true.

So your argument is that we would have won, if we'd taken the XPs and not gone for it on the 4th downs?

Since you didn't deign to answer, let's look at it on paper.

Here's the real vs hypothetical scoring summary. What happened in real life is on the left. The right side assumes Harbs doesn't play the game on analytics; no weird 4th-down attempts and 2-pt tries :

Real-life vs if Harbs didn't outsmart himself with analytics

Team	Real	Overall	–	Team	“Should’ve”	Overall
Balt	TD (miss 2)	6-0	–	Balt	FG	3-0
KC	TD	6-7	–	KC	TD	3-7
KC	TD	6-14	–	KC	FG	3-10
KC	TD (miss XP)	6-20	–	KC	TD (miss XP)	3-16
KC	FG	6-23	–	KC	FG	3-19
Balt	TD	13-23	–	Balt	TD	10-19
KC	TD	13-30	–	KC	TD	10-26
Balt	TD (miss 2)	19-30	–	Balt	FG	13-26
Balt	FG	22-30	–	Balt	FG	16-26
KC	FG	22-33	–	KC	FG	16-29
Balt	TD (miss 2)	28-33	–	Balt	TD	23-29

Notes:

• Our first TD came after Ingram converted 4th-&-3 at the KC 9. In the right-side scenario, Harbs "takes the points": we open the scoring with a FG.
• KC's 2nd TD came after we gave them the ball on downs near midfield (our 47). I think the assumption is we might have held them to a FG if we had punted? So I went with that. (The field position didn't help us so much on Mecole Hardman's long score, but whatever.)
• Ingram's Q4 TD (that pulled us to within 30-19) came after Lamar converted that miracle throw to Seth Roberts on 4th-&-5. In this scenario Harbs "takes the points" from the KC 27 with about 12 mins to play.

So instead of losing by 5 – we would have lost by 6?

What are we critizing Harbs for, again?

[jonboy79]

Thank you JZC for including the 4th down decisions as well, as that is the only fair manner.


Sent from my iPhone using Tapatalk

[lobachevsky]

Football is a game of heart with the mind following not the other way around.... Damn they're taking the soul out of the game.

Posted with the same mindless certainty as the pre-Great War French general staff asserting that "the spirit of the attack" would carry the day in any future war. (The German machine guns & artillery ripped that delusion to shreds, & destroyed an entire generation of young Frenchmen in the process.)

[HotInHere]

Didn't Billick do that analytic stuff with his computers etc and now we're at it again. Football is a game of heart with the mind following not the other way around.Use creative thinking and will not some formulaic system.Damn they're taking the soul out of the game.Those old time QB's Graham,Unitas,Starr etc would hate this stuff.

Would they? If it helped them win? Unitas spent a lot of time studying film and opponents' tendencies. That's analytics, just without the hard quantities.

You ever discuss a QB's completion percentage? Passer rating? YPA? That's all analytics. Do you read Filmstudy's offensive line evaluations? Really good analytics.

Why would you NOT want additional information? That makes no sense. If I tell you your opponent runs the ball to the right 73% of the time that there are two TEs in the game, wouldn't that be helpful to you? What if I tell you the CB you are facing next week allows a passer rating of 140.2 in man coverage against double-moves? Does it take the soul out? Or does it just help you win a football game?

[lobachevsky]

I'm an analytical person - but remember that our analysis is only as good as the data we feed it, & that data may be lacking information that's crucial - information we might not even be able to fit into the analysis.

Remember 2015, when the Crows were scheduled to play their first two games in the west (Denver & Oakland)? Harbaugh decided to keep the team out there in order to minimize the disruption caused by time zone changes.

Now my last employer was a researcher who modeled the effects on performance of things like irregular sleep patterns and time-zone displacement. (Very important for regulating work schedules of airline pilots, long haul truckers, etc.) And we would have said that Harbs' decision was absolutely supported by the analytics.

In fact it was a disaster - they lost both games & never recovered from it, finishing 5-11. In particular the team looked more unfocused & listless the second week than they did the first.

What happened? My guess is that the disruption of staying away from home for that extra week - living in a hotel, practicing on a strange field - more than cancelled out the advantage in not crossing multiple time zones. but that's not something that you can quantify - or even have much of a sense for, since the team had never done it before.

Just sayin'.

[JimZipCode]

What happened? My guess...

We lost Suggs in Q1 of the first game, and came within a trice of winning both tames.

Inferring causality from two results that close, is a sucker's game. Way too easy to reason beyond the evidence.

[AlbionHero]

btw... if the analytics argument is sound, wouldn't every team go for 2 if they're down by 11? Wouldn't every team go for it on 4th and 2 on their side of the field? Yet we don't see this a lot.

I mean, the same reason it took forever for teams to adjust to shifts in baseball. Its a radical concept that takes one team having the balls to be willing to try it to start the trend.