Decoding True Strokes Gained: Field Strength/Cuts/Category Drift

Website Issues/Suggestions
#1

@will_courchene @matt_courchene apologies for what seems like a simple question, but how am I able to replicate the true SG transformation being performed, or access the data directly from an endpoint? Looking at Sepp Straka at the WM, my understanding is that a +.19 avg SG vs baseline would yield a total of +.19 SG across putting, arg, app, ott for the event. Totalling up the averages, its about a gap of +.34, and primarily coming from OTT and APP. Are true SG being re-determined post-cut with vs the r3/r4 remaining field (ie Sepp Straka made the cut and gained strokes vs an even stronger remaining field), and then the distribution of SG mainly coming from the two categories means the field’s strength was OTT + APP?

I find your True SG data valuable and am trying to find the best way to tap into it!

#2

Hi, I’m not sure what you are asking. What is this +0.19 avg SG versus baseline you are referencing? The field-strength adjustments are done at the round/course level, so each round can have a different field strength. The categories should always add up to total.

#3

image

I was assuming the +.19 on the tournament summaries page was the relative adjustment:

I’m not seeing the True SG figures aligning with the raw data from the API endpoint. SG App for the 4 rounds totalling to .696 vs .85 on True SG.

#4

The tournament summaries should closely correspond to round 1 skill / field strength. The post-cut rounds will have stronger field strength. The 5th Q&A in this section explains why tournament summaries skill isn’t exactly equal to R1 field strength.

#5

Thanks! Apart from that digression on my side, how would True SG be accessed via the API, or replicated, then? I see another question on the forum asking just about the same from a few hours ago. I can track your response there if that’s easier.

#6

Sorry for the late reply. True SG is not in the API. It’s something we’ve debated putting in but haven’t pulled the trigger on, as it would be another step towards making our model more easily replicable. As for replicating it, you would need to start with the raw SG data and apply a similar statistical method to what we’ve done (which we outline in our methodology blog post).

1 Like
#7

So to get going on the right path, start with SG data and adjust for field strength using tournament summaries.Recognizing this will be not identical but is at least on the right path?

#8

Yes, that’s a good start. Post-cut rounds differ from R1 (which is what the summaries page displays), but it’s not a huge difference typically.

1 Like