1) Nobody has ever argued against the idea that some players are simple
better shooters than other players and therefore have higher shooting percentages. That's patently obvious.
The issue is when a player's shooting percentage
deviates significantly from his own baseline or career average. That's when we know they're either unsustainably lucky/hot, or unsustainable unlucky/cold.
Different shots from different locations (like scoring chances or high danger scoring chances) do come with higher sh%, absolutely. But a player can still shoot 5-10% higher or 5-10% lower than what they usually shoot from those locations, just as they can from low danger areas. They can't control that variability.
2) You're over simplifying the 'a shot is a shot is a shot' argument.
The 'a shot is a shot is a shot' argument comes from the repeatably observed fact that over the course of large samples, a team will score roughly as many goals off of 'low danger' or 'perimeter' shots as they will 'high danger' shots or 'scoring chances'. The reason for this is rather simple.... high danger shots and scoring chances happen significantly less frequently than 'low danger' shots. They're hard to create. So it may take many more shots to score off low danger chances, but the raw volume advantage evens out the gains that high danger shots or scoring chances provide as far as success rate.
From a historical and practical perspective, the teams that pour rubber onto the opposing net at a high rate, indiscriminate of 'quality', generally have greater success than those that pass around looking for high quality and getting relatively few chances off. That's obviously not 100% the case across the board, every season, but it's a pretty clear, repeated trend.
That doesn't mean you shouldn't try to create high-danger chances. You absolutely should when you can. You just can't
always be looking for high danger scoring chances, especially when a shot on net is the best option and you're an at least competent possession team that can probably get the puck back and generate more shots.
Of course, this obviously requires that you also prevent the other team from putting even more rubber towards your own net. The power of volume works both ways.
It's obviously
possible that the hypothetical team that eschews volume in pursuit of quality could find sustainable, repeatable success. It just hasn't happened yet.
3) As far as how scoring chances and high-danger scoring chances are defined, it's been an evolution. Currently, it's dictated by aggregate sh% variance across factors.
Over the course of a season (and their careers), players take a ton of shots. In certain scenarios, their sh% consistently sees an increase or a decrease from their overall sh% baseline.
Factors include:
- Shot type (Wrist shot, slap shot, deflection, etc.)
- Shot distance (Distance from net)
- Shot angle (Angle in absolute degrees from the central line normal to the goal line)
- Rebounds (Whether or not the shot was a rebound)
- Rush shots (Was the shot off the rush or off the cycle)
- Strength state (5v5, PP, PK, etc)
- On/Off wing
We aggregate the change that players see as a result of all combinations of these factors, and then we get an idea of how much more dangerous any given shot is, or at least, how much higher or lower the percentage chance is that a player scores under such circumstance vs their overall career baseline sh%.
For example, here's a heatmap demonstrating the relative danger based on distance and angle (the two are tied at the hip, given the location of the net will force changes to angle based on distance away).
As stated earlier, a player's sh% under these circumstances varies. Over a large sample, a clear sh% range scoring chances and high-danger chances appears, but players can move out of this range at various times. A player that normally shoots say 15% better from high danger areas compared to the rest of the ice, might go through stretches where they only shot 2% better, and stretches where they shoot 30% better, for example. It's not controllable obviously, so we know when we see this, or we see the impact it has on their overall sh% as a whole, that it's either an unsustainably hot or unsustainably cold player that will regress toward their norms eventually.
For the record, all of this is fed into the xG model that provides the numbers for xG , xGA, xGF%, which have all proven more predictive of future goal scoring than current goal scoring.
Perfect? No.
Useful and predictive? Yas.
