How does the new playoff formula really work?

With the NJSIAA now officially replacing the controversial Born power Index with the Opponent Strength Index - a similar formula used in a different way that doesn't encourage larger margins of victory to improve playoff standing - just how will it work?

Everything in 2019 will be very similar to 2018, with one key difference:  in 2018, a team's BPI was used to determine it's playoff standing, but in 2019, the values of its opponents (hence the Opponent Strength Index) will be used to determine a team's playoff standing.  The value given to an opponent (its OSI) will be determined in a way similar to the BPI.

Sure, it sounds confusing.  And to run the numbers on the entire 2018 season would mean running tons of teams' numbers for every game.   Why?  Because your OSI (the strength of your opponents) is based on all their games.  So, if we wanted to run the GMC Red Division from Week 0 to Week 8, we'd have to find the strength of all South Brunswick's opponents.  They played Lenape from South Jersey, so we would need their number as well.  Which means we'd have to run the numbers for their entire schedule.  To do that, we'd need the values of their opponents, which means running eight or nine more South Jersey teams, and if they play someone out of the area.... you get the picture.  It gets exponentially difficult.

But what if we took a small sample size of all teams we know?  For example, what if we just calculated the GMC Red Division based on divisional games??

2018 GMC Red Divisional Season

Let's use this year's rules for last season.  The Born Power Index used the final numbers from 2017 to begin 2018, then used varying formulas throughout the year.  The OSI resets, and uses one formula throughout the year.  So, let's set it up.  Here are the GMC Red teams' 2017 final BPI, converted to SI.  (Strength Index - remember, it's not OSI until you take the SI of your Opponents!)  Each team is reduced by 1/3 of the amount they are over 60, or increased by 1/3 the amount they are under 60.  (Also note the official rules call for two decimal places, but we'll use one here for simplification.)

East Brunswick:  53.5 BPI, 55.7 SI (6.5 below 60 - 1/3 of 6.5 is 2.2 - add 2.2 to 53.5 and get 55.7)
JP Stevens:  46.7 BPI, 51.1 SI
Old Bridge:  86.2 BPI, 77.5 SI
Perth Amboy:  40.7 BPI, 47.1 SI
Piscataway:  82.2 BPI, 74.8 SI
South Brunswick:  89.4 BPI, 79.6 SI

The OSI will be calculated at the end of the year using numbers locked in after Week 7.  So, we need to go through our "mini" season here (a round-robin schedule of 5 games) and let's see what we come up with.  Here's the formula:

The Expected Result is the difference between two team's SI.  Find the difference between the Expected Result and the Actual Result for the winning team and divide by 5.  That's how much a team goes up or down.

Week 1 Examples

East Brunswick (55.7) vs. Piscataway (74.8)
Expected Result:  Piscataway +19.1
Actual Result:  Piscataway +41
Difference:  Piscataway +21.9
Change in SI:  +/-4.4
East Brunswick is now a 51.3
Piscataway is now a 79.2

JP Stevens (51.1) vs. Perth Amboy (47.1)
Expected Result:  JP Stevens +4
Actual Result:  Perth Amboy +14
Difference:  Perth Amboy +18
Change in SI: +/-3.6
Perth Amboy is now a 50.7
JP Stevens is now a 47.5

We'll skip the tedious task of showing the math for the other game that week, and the 13 other games the GMC Red Division played each other.  Moving on...

2018 GMC Red Divisional Strength Index Numbers

These numbers - like the BPI - should represent the relative strength of all these teams.  Note that in the preseason, South Brunswick was 4.8 points higher in SI than Piscataway.  The Chiefs won their head-to-head matchup with the Vikings early in the season.

Piscataway:  92.6
South Brunswick:  73.7
Old Bridge:  66.9
Perth Amboy:  54.3
JP Stevens:  51.3
East Brunswick:  47

Not surprisingly, here were the Divisional Standings - and remember, every team played every other team, so it's an apples to apples comparison here, and it yielded the six teams in the same order:

Piscataway:  6-0
South Brunswick:  5-1
Old Bridge:  3-2
Perth Amboy:  2-3
JP Stevens:  1-4
East Brunswick:  0-5

So, it appears the SI is a good evaluation of team strength.  In a 6-team conference, where everyone plays each other, the rise and fall of teams' SI coincided with their rise and fall in the standings, and fell in line with head-to-head play:  the undefeated team had the highest OSI, the winless team had the lowest.  Everything fell in line alongside teams' winning percentages.

But is it similar to the Born Power Index??

Actually, it is.

While Born's adjustment period came in the first 3 or 4 weeks of the season, the OSI's happens in the preseason in order to use a consistent formula across all games.   But a number of teams closed the gap by the end of the season.  That is to say, their final SI and BPI were generally closer than their preseason BPI and SI.  Here's a look at the difference between systems from pre-season to the cutoff:

East Brunswick:  2.2 to 2.5 (0.3 further apart)
JP Stevens:  4.4 to 3.4 (1.0 closer)
Old Bridge:  8.7 to 3.3 (5.4 closer)
Perth Amboy:  6.4 to 0.5 (5.9 closer)
Piscataway:  7.4 to 2.9 (4.5 closer)
South Brunswick:  9.8 to 8.6 (1.2 closer)

And with four more games played in a regular season outside of the division, one could surmise most of these numbers would only be even closer.  This would seem to indicate that the formula for determining a team's strength index gets essentially the same results as the Born Power Index, which happens to be a very good indicator of strength.  Inotherwords, the NJISAA is now using a very public formula to achieve the same results.  That was one of the NJFCA's key criteria for making changes in 2019.

However, it was a bad way to seed the playoffs, because teams that won by larger margins were rewarded with higher seeds, and that could potentially result in more blowouts.  So, now we're ranking teams on the value of their opponents - the OSI - which means beating your opponent by more points will lower their rating, this lowering your OSI, the index - or average - of your opponents' strength.

How do we calculate the OSI?

We won't show you all the teams' math, but let's look at an example of a team that won and lost a few games:  Old Bridge.  Here's the Knights' schedule, results (score doesn't matter here, just wins and losses), the opponent SI, and the value they get (full for a win, half for a loss)

at South Brunswick - Loss - 36.85 (half of their value of 73.7, because Old Bridge lost)
Perth Amboy - Win - 54.3 (their full value of 54.3, because Old Bridge won)
Piscataway - Loss - 46.3
at JP Stevens - Win - 51.3
East Brunswick - Win - 47

Total:  235.75
Averaged Over 5 Games:  47.15

Old Bridge's OSI is 47.15

Here's the GMC Red final OSI - the average of all their opponent values, with full values given for a win and half values given for a loss:

Piscataway:  58.64
South Brunswick:  53.16
Old Bridge:  47.15
Perth Amboy:  42.98
JP Stevens:  38.15
East Brunswick:  33.88

Once again, the six GMC Red Division teams ended up in the same order as their Strength Index numbers.

Finding the UPR:

Now, like last year with the BPI rank, we would assign an OSI rank to the teams, 1 through 6.  That rank would be 60% of a team's UPR, and traditional power points would be 40%.  We can't really compare this to the UPR from last season, because Piscataway, Perth Amboy at JP Stevens were in the North Group 5 supersection, while South Brunswick, Old Bridge and East Brunswick were in the Central Group 5 supersection.  But we can use last year's results to make a "GMC Red Section" to compare.

Here are the teams ranked in the GMC Red by Born Power Index rating:

  1. Piscataway - 95.5
  2. South Brunswick - 77.8
  3. Old Bridge - 60.2
  4. Perth Amboy - 51.8
  5. East Brunswick - 48.8
  6. JP Stevens - 48.5

Here are the same teams ranked in the GMC Red by Power Points average:

  1. Piscataway - 19.25
  2. South Brunswick - 15.33
  3. Perth Amboy - 9.63
  4. JP Stevens - 9.38
  5. Old Bridge - 8.5
  6. East Brunswick - 5.63

Here would be each team's GMC Red UPR:

  1. Piscataway - 1
  2. South Brunswick - 2
  3. Perth Amboy - 3.6
  4. Old Bridge - 3.8
  5. JP Stevens - 5.2
  6. East Brunswick - 5.4

What's interesting here is that Perth Amboy now finishes ahead of Old Bridge.  Both the Panthers and Knights missed the playoffs last year, but Old Bridge overall was 2-6 at the cutoff, while Perth Amboy was 4-4 and missed, while Bridgewater-Raritan or Westfield - each with just two wins at the cutoff - qualified.

So how would it look based on the OSI?

  1. Piscataway - 1
  2. South Brunswick - 2
  3. Perth Amboy - 3.6
  4. Old Bridge - 3.8
  5. JP Stevens - 4.6
  6. East Brunswick - 6

As we can see, using the OSI and using the BPI didn't really change much.  JP Stevens ended up with a better UPR, while East Brunswick was worse off, but the top 4 teams didn't move, because even though their OSI and BPI differed, they were still ranked in the same order, leading to identical UPR numbers.

What does it all mean?  Well, it's admittedly a small sample size.  (For the record, we ran the numbers on the GMC White and found similar results:  the OSI and SI standings were similar to the BPI, and the numbers by the end of the season were closer than they were in the preseason when the adjustment was made.)

The bottom line appears to be this:  the Born Power Index was a good indicator of strength, but a poor choice to use for seeding teams because of the impact of scoring more points.  With a similar system used in a different way and the "blowout" factor no longer part of the equation, we've got similar results without the risk of bad sportsmanship to get a leg up in the postseason chase.

Now, what to do about those multipliers...