This article is part of the SaskPoly story. Read how it all started.
Why We're Building an NFL Survivor Pool Engine
The CFL Scorecast worked because it treated games as a market problem, not a fan problem. Here's why we're bringing that same engine to NFL survivor pools.
Why We're Building an NFL Survivor Pool Engine
A few weeks ago I looked at our content calendar and felt empty.
We had spent two seasons building the CFL Scorecast into something sharp. Market gaps. Situational reads. A model that actually moved the needle for people betting Canadian football. Then the Grey Cup ended, and the page went quiet. I was sitting on a framework that worked — win probability, ownership context, rest-travel edges — and applying it to a league with nine teams and a four-month window.
That felt small.
NFL survivor pools are the same problem at a bigger scale. One pick per week. No spread. No totals. Just win probability, resource management, and a field of thousands trying not to step on the same landmine. The structure forces every player to make the same hard trade-offs. The person who thinks three weeks ahead beats the person who thinks one week ahead. Every time.
I had been ignoring them because they felt "too mainstream." That was stupid. The edge isn't in finding a hidden sport. The edge is in applying discipline where others apply narrative.
So we're building it.
What We're Bringing Over
The CFL model worked because it treated every game as a market problem, not a fan problem. We're bringing that same engine to the NFL — but the inputs change, and the stakes get heavier.
1. Win probability, not popularity.
The public picks the team that feels safest. We start with market-implied and model-derived win probabilities, then compare them to public ownership. A 72% favourite that 55% of the pool is taking can be a worse bet than a 65% favourite that only 12% of the pool is taking. The second pick preserves your equity if the chalk goes down. This was the core of the CFL Scorecast. It translates directly.
2. Future-week planning as first-class math.
You have 32 teams and up to 20 weeks. Burn the Chiefs in Week 1 and you have nothing left for Week 12 when the board is thin. We're building a team-usage calendar that maps every club to their best remaining spots — not just this week's probability, but the probability that you still have a viable team in Week 14, 16, and 18. The model optimises for survival, not weekly safety.
3. Situational edges the public misses.
Home favourites off a bye. Teams travelling cross-country for early games. Defensive fronts against rookie quarterbacks. Weather and late-season motivation gaps. These factors move win probability more than the "who is favoured" heuristic, and they rarely show up in casual pick logic. The CFL taught us that rest and travel matter more than power ratings. The NFL is bigger, but the physics are the same.
4. No emotional bets.
The single biggest leak in survivor pools is narrative. "I can't pick the Cowboys again after they burned me." "The Lions are due." We don't do due. We do numbers. Detachment is a feature, not a bug. This was the hardest lesson to learn in the CFL model, and it's the one that will separate our picks from the field.
What the Output Looks Like
Every week you'll get the same format we use for the CFL: a win-probability table, public-ownership context where available, and a survivor-pick note when the data points to a clear edge. We'll flag the chalk train when it's running off a cliff. We'll highlight the quiet favourite with low ownership and a soft schedule ahead. We'll tell you when the board is so thin that saving a bullet is more important than firing one.
For the high-stakes players, we'll add Circa-specific notes: the 20-leg structure, the Thanksgiving and Christmas slates, the in-person registration reality. For the casual players, we'll keep the same one-paragraph discipline: here's the number, here's why, here's the risk.
The Honest Truth
We are not NFL insiders. We don't have a source in the Chiefs locker room. What we have is a framework that separates probability from popularity, and the discipline to follow it when the crowd is screaming for the obvious pick.
The real test starts in Week 1. We'll run the engine, publish the picks, and track the results. If the model leaks — if we save teams too long, or chase chalk into a trap, or miss a situational edge — we'll tear it down and rebuild it. That's the job. Build, measure, burn, rebuild.
If you're entering a pool this season, start with the numbers. Save your bullets. Avoid the crowd. And if you want a second opinion before you click submit, check back here each week — we'll have the data side ready.
If you find value in the weekly briefs, consider leaving a tip. It covers data feeds, hosting, and the coffee required to stare at NFL schedules until 3 AM.