Over the past decade, the NFL betting marketplace has evolved into a year-round process.
My process for handicapping the upcoming NFL season begins immediately after the draft. This is the time of year to focus on those aforementioned seasons win totals. My NFL prep work begins each spring with a thorough look back at last year’s results.
Why look back, you ask? The answer, of course, is simple – because that’s where I find my very first edges when approaching the upcoming season. If my base power rating heading into the offseason wasn’t wholly accurate, it results in inaccurate adjustments moving forward. Part 1 of my process begins with a thorough examination of the schedules all 32 NFL teams played last year, looking for outliers.
I’m looking to identify teams that were better than their final records would indicate because they faced an extremely tough slate. And I’m looking for teams that were weaker than their final records would indicate because they faced an extremely easy slate. But unlike most schedule based analysis, I’m taking my time to review every game that was played through the first 16 weeks of the previous campaign (discounting Week 17 results due to extreme randomness).
The widely available information that helps set the markets start with last year’s results as their base point. But they don’t do it thoroughly. The conventional models simply add up the combined records of every opponent a team faced to come up with their final strength of schedule for the previous season and to create the SOS for the upcoming campaign. Creating numbers that are more accurate than the broader market for BOTH seasons are instrumental in what I’m trying to accomplish.
The aggregate numbers for last year’s strength of schedule -- the starting point for any analysis of the upcoming campaign -- can be extremely misleading. A team’s final win-loss record tells us nothing about how good they actually were at the time the game was played. There were some enormous in-season power rating shifts last year, just as there is every year.
When you played a team it is every bit as important as who you played! Yet the markets devalue that concept entirely for at least two reasons – the analytics are time-consuming and they require human judgment. There’s no algorithm that will spit out accurate numbers based on judgment calls, and my process is all about making judgment calls (with some hard numbers thrown in to ensure some semblance of accuracy).
So here’s what I do. I go back to my spreadsheet from last season that details my power rating numbers on a weekly basis from last year. Then I go through the schedule, using my numbers for every squad on the week the game was played. Using that formula, you’ll get some significantly different results.
I also discount Week 17 results. Some teams had quit on their coach and their season, other teams were resting starters and other teams were playing their guts out to try to reach the postseason. Power rating models for the final week of the regular season are inherently flawed – that’s why we see enormous line moves that week every single year.
Next week I’ll finish the thought process, writing about the next step -- identifying accurate strength of schedule numbers for the upcoming campaign.