For the sake of those using these to help shape their tips, here is how the FFSS predictor sees round 22.
As I wrote last week, my attention has shifted away from my prediction model in recent weeks as I start to play with other interesting models and tools. The FFSS predictor has performed admirably this year, however the list of potential improvements is long and I'm sure I will be kept busy over the off-season tweaking the model and adding new features.
The most obvious of these potential changes is accounting for injuries and player changes, and also factoring in the importance of each match relative to each team's season. If you are using these predictions to shape your tips this week, remember that FFSS is essentially dumb to both of these things. It's important to keep monitoring the model's tip confidence and performance relative to the market in order to find any other areas of the game that might be over or undervalued by FFSS.
As for this week, we can see potential value in both of the "top-8 clashes". FFSS sees the Friday night clash as a virtual coin-flip whereas some bookies are offering up to $2.15 on the Hawks. Disregarding vig, this equates to an implied win chance of about 46%. Not great value, but certainly potential if you have trust in the Hawks (which I don't necessarily have).
There is similarly slim value on the North v Sydney game. If you can get on North at $3.40, you'll get them with an implied win chance of about 29% when the model thinks they actually about 31%.