Release Notes: 23 April 2026

Three big ones this week. Each of them unlocks a whole class of searches we couldn't do before. The kind of angles a sharp punter actually thinks about before placing a bet.

Days rest is now a stats filter

Every player's database row now carries how many days they'd had off since their previous game. We've always had this number quietly sitting there. What's new is that it's now a first-class filter alongside opponent, venue, home/away, and the recency windows.

When you load an upcoming fixture on a single search or an SGM, the app figures out the player's projected rest for that game (fixture date minus last-game date) and automatically surfaces a "N days rest" card with their history at that exact turnaround. Seven finer-grained buckets: 5, 6, 7, 8, 9, 10–14, and 15+ days. Long-rest is capped at 30 so cross-season gaps don't pollute the sample.

Why it matters. AFL rounds aren't evenly spaced. A Saturday-to-Thursday gap is a short five-day turnaround, a bye week gives ten to fourteen days, and returning from a minor injury or mid-season break can stretch to 20-plus. Players often look like completely different units at different rest levels. You might find someone is a 65% disposals hitter on seven days but drops to 45% on five. That's the kind of edge that doesn't show up in a season average.

The rest card gets its own weight slider too, so you can nudge up or down how much it contributes to the final blended percentage on the algorithm card at the top of the page.

Loose Mode: combine opposing players in a single SGM

This is the big one. Up until now every SGM leg had to be from the same team. That's how most bookies model it too. The thinking being that teammates' stats correlate (if Collingwood dominates possession, every Collingwood midfielder's numbers tick up). Valid for some bets. Not so valid for others.

Loose Mode v1 lets you bundle players from both teams playing each other. Toggle Loose at the top of the SGM page and you can build things like Nick Daicos 30+ disposals AND Zach Merrett 30+ disposals in one bet, Collingwood vs Essendon, one player per side.

Concrete example. Say you build Nick Daicos 30+ disposals AND Josh Daicos 25+ disposals AND Zach Merrett 30+ disposals as a single SGM across Collingwood and Essendon. The app first runs a precise same-team SGM on the Collingwood side: how often have Nick and Josh both cleared those lines in games they actually played together as teammates. That gives you one number. Then it runs the Essendon side on its own: Merrett's over rate at 30+. That gives you a second number. Finally the two are multiplied to produce the cross-team combined rating. You get each side's full weighted breakdown in its own column (opponent, venue, home-away, days-rest, season windows, the lot), and the combined percentage shown front and centre at the top.

One caveat we're deliberately transparent about: the combined number assumes the two sides are independent. In reality they're not perfectly independent (a blowout game affects both teams' numbers, just in opposite directions). So treat the Combined percentage as approximate rather than exact. The per-side breakdowns and the game-log drill-down are still the best way to check whether an angle actually has legs.

And because v1 accepts a single player per team, you don't have to build a heavy four-leg multi to use it. Most bookies will happily take a cross-team SGM off you, but the model they price it with is a black box. With Loose Mode you can see the full weighted breakdown of each side, the game logs behind each player, and the combined probability before you place the bet.

Career toggle: flip between season range and whole-career view

At the top of every search result there's now an emerald-ringed segmented button labelled Context view. It flips the Home/Away, Venue, Opponent, Days Rest, and All Games cards between two states: the season range you picked (default) and the player's full career.

You'd use this when you want a recent-form read on the algorithm (say, 2025–2026 only for the weighted prediction) but you also want to eyeball long-term patterns. Things like: does this player just play well at the MCG? Always? Across 150 games and 8 seasons? Or is it a 2025 run of form only?

Critically, the weighted prediction at the top of the page never changes when you toggle. Career view is a read-only display flip. The algo still uses whatever season range you actually set. So a user chasing recent-form data for their bet can still glance at career-wide venue or opponent history for extra context, without messing up their prediction.

Time-based cards stay anchored. This Season, Last Season, and Last 30 Days don't move when you flip the toggle, because those are intrinsically time-bounded by definition.

What's next

A way to manually override the auto-derived days-rest bucket for a team (right now it's driven off the fixture; some users want to run what-ifs). And we're still working through expanding the stats to include kicks, handballs, clearances, and hit-outs across all search surfaces.

Full list on the Feature Roadmap.