I watched the fourth quarter of a game played Saturday afternoon in Joliet between the Eagles from Carl Sandburg High School and the Tigers from Joliet West. Our computer algorithms had both predicted a win for Sandburg, which is how the game ended, but it almost wasn’t so.
Late in the fourth quarter, a Sandburg touchdown put the Eagles up, 36-20. When Joliet West got the ball after the kickoff, they put together an impressive, no-huddle drive, straight to the end zone. A successful 2-point conversion after the touchdown left them down by only eight, 36-28.
Sandburg’s next drive was lackluster, resulting in a failed 4th-down field goal attempt from the Joliet West 43-yard line. The Tigers took over on the 20, according to the high school rules, and quickly moved the ball down the field. It would have been even quicker if the drive’s second play, a pass completion that ended near the end zone, hadn’t been called back on an illegal block penalty. Still, quarterback Anthony DiNardo quickly picked the team back up and brought them down field.
A 4th-down conversion, which was so close that officials had to measure it, had them standing within 8 yards of pay dirt, but that’s when things turned sour for the fourth-quarter comeback. A quarterback sack considerably deflated their chances of winning: although DiNardo’s fumble was recovered by the Tigers, they were standing all the way back at the 37, 2nd and goal. DiNardo threw an interception close to the final buzzer, and the game ended 36-28.
Score one for our computer prediction algorithm for Illinois varsity high school football. It wasn’t so good at a few other games in week 4, and I’ve created an ad hoc report, which will run throughout the season, showing the biggest upsets in the state, based on how our computer thought the game would turn out. In order to keep the report fresh, games will only stay on it for seven days. As the season progresses and more data comes in about the teams, there should be fewer games on this report.
You can view the upset-victory report, any time, here
The biggest upset victory this week appears to be the LaSalle-Peru victory over Yorkville, 16-9.
Our computer algorithm was so sure Yorkville would win that it gave the match-up an alpha (α) value of 52.82. That gave Yorkville odds of winning higher than 99 percent: LaSalle-Peru’s 0-3 record, all big-point-spread losses, made the Cavaliers look bad to the computer, despite the fact that their opponents were all undefeated. And Yorkville came into the match-up 3-0, but all their wins came against weaker teams.
The computer takes team strength into account, but with only three strong teams for LaSalle-Peru and three weak teams for Yorkville, a successful prediction was unlikely. Furthermore, playing strong teams in the first three weeks tends to strengthen a team and render algorithmic estimates of their strength highly inaccurate, while facing only weaker opponents in the early season, as happened with Yorkville, tends to make a team feel secure in each victory and perhaps a little complacent.
“I don’t want to say the morale was down, but we were hanging our heads,” the LaSalle NewsTribune quoted LaSalle-Peru quarterback Colby Sassano as saying. “We knew we had to pick it up.”
That’s the kind of motivation only failure can provide.
“We needed this,” he continued. “This is really going to help us going on to the rest of the season. Those first three teams were tough, but that’s in the past and now we’re looking into the future.”
Keep in mind that until about the seventh week of the regular season, our algorithms are greatly affected by a small sample size bias. Once about six scores are available for all the Illinois teams, the reliability of the algorithm increases substantially.