RPS +/- Trend Request
Perhaps someone is currently going through the data and not sure if it would ultimately be useful or no. But…would we expect to see a more pronounced/outsized RPS+ over the last few years with what was allegedly happening in terms of advanced scouting/knowing signs? Is anyone in a position to compile a chart showing the long term trend on the offensive or defensive side of the ball? Just seems like that could around the competitive advantage of going through with this whole operation.
October 25th, 2023 at 8:31 AM ^
Hunh?
October 25th, 2023 at 8:40 AM ^
RPS=rock/paper/scissors. When UFRs are compiled there is a measure that attempt to gauge which team outschemed/outsmarted the other team based on base sets/calls for the play. If we benefitted from the code breaking of which we're accused you'd expect to see our RPS results be better, on average, over the period that Conor Stalions was doing his stuff when compared to the Harbaugh era pre-time in question.
October 25th, 2023 at 8:42 AM ^
Thanks for explaining better than I could Chaco.
October 25th, 2023 at 8:43 AM ^
Michaeljordanstopit.gif
October 25th, 2023 at 8:51 AM ^
get_some_help.png. Also username checks out.
October 25th, 2023 at 9:12 AM ^
Maybe by accident. But the name is reflective of where I live (South Carolina) and Blue as part of the “Go Blue” for Michigan. It was a nice double meaning for me, so your comment is understandable.
October 25th, 2023 at 8:46 AM ^
It wouldn't mean there was causation. The goal of RPS tracking is some measure of luck, but some measure of how well the coaches gameplan. Your proposal is a variation of the "UM record against the spread" analysis before & after 2021
October 25th, 2023 at 8:50 AM ^
I think this is different and have been wondering about this just like OP.
Performance against the spread required execution (basically Deion’s argument). RPS wins could theoretically be attained just by knowing the other team’ signals.
The big confounder is that the coaches have changed. You can’t compare the RPS success rate of the team before and after because playcallers have changed.
October 25th, 2023 at 8:52 AM ^
That's the biggest confounding variable, along with OTHER teams changing their coaches too. They call it the "coaching carousel" for a reason.
October 25th, 2023 at 9:34 AM ^
Eh, it's data.
Certainly, if Michigan's RPS was comparatively terrible in 2022/2023, it would suggest either illegal sign stealing didn't happen, or it was extremely marginal in benefit. But doesn't entirely disprove it.
If Michigan's RPS was comparatively very good in 2022/2023, it would suggest either illegal sign stealing happened, or it was extremely beneficial. But doesn't entirely prove it.
You could also do in-season comparison of games we know we've been accused of sign stealing vs. those we haven't been. E.g., 2022 RPS vs. Colorado State/Hawai'i/UConn vs. 2022 RPS vs. Penn State/Ohio State. It wouldn't be entirely dispositive ("we didn't open the playbook vs. inferior opponents" or on the other side "inferior opponents have terrible playcallers, so it's easier to RPS them even without sign stealing")
But either way, data is data. I personally feel like the "illegal" aspect of the sign stealing (scouting other games) probably was happening, but it probably was of very marginal benefit. Analysis of RPS could possible help quantify the benefit, but it wouldn't be 100% dispositive.
October 25th, 2023 at 8:50 AM ^
I actually did this a few years ago when Don Brown was our DC. I wrote one part of a 3 part diary showing how poor his RPS trend was. I used multiple years of UFRs.
Unfortunately my diary didn't get much of a reception, so I didnt publish parts 2 and 3 then Don Brown was shown the door so it was a moot point.
I found it interesting because I started my analysis when Don Brown was very popular in these parts, and my findings was that his RPS was crap...and then reality played out.
October 25th, 2023 at 9:07 AM ^
Solve your problems with execution has turned out to be much more effective than solve your problems with aggression.
So much so that crybabies wanna say it was solve your problems with espionage.
But it's been Herbert and Ravens', and Jimmies and Joes, even more than Xs and Os.
October 25th, 2023 at 9:42 AM ^
I can promise you right now we had a better RPS in the 2.5 years where we had more success. Not only did we have someone skilled devoted to sign stealing (which is legal), but we had a whole new set of coordinators starting in 2021 (with Gattis demoted in 2021 before leaving).
What I want to know from this exercise is, was our RPS 100%? 50%? 10%? How often were we in the exact right play? Because to listen to anyone taking the OSU/MSU/ESPN side of this, they believe we knew every play of every game.
But no matter what the number is, we'll never know how good could Stalions have been without the extra filming. Only he could tell us that.
October 25th, 2023 at 9:48 AM ^
Go Check Harbaugh's record vs the spread since 2022 hiring vs before.... People are taking these minor bits of data points and using them as a forest fire to burn something down that could have had a million other reasons. The only question is how far up the chain did this get. Does the OC, DC, Head coach know? Were their game plans designed around this, was their reimbursement for those tickets. If there is a money paper trail we are getting the shaft if their isn't this is going to be a lot of he said she said BS.
October 25th, 2023 at 9:55 AM ^
Feels like we’ve been on downside of RPS last two years. First half a lot last year and first quarters this year.
October 25th, 2023 at 10:21 AM ^
I wrote a plotting tool that would scrape UFR data and visualize it for players and team stats. Because it was based on web-scraping, it was easy to break whenever they changed formatting of the tables in UFR. That said, I just went to it and it did scrape some 2023 data correctly. So there is a big gap from early 2021 to the beginning of 2023, which is the central region you want to see.
That said, if you aggregate by year, early 2021 defense was better than late-stage Don Brown but was distinctly lower in RPS than 2017/2018 Don Brown. Early 2023 was looking like the best RPS we'd seen.
Offensive RPS was the lowest on record in early 2021. That was before Michigan had beaten OSU though and when it looked like Michigan was hopelessly slamming its face into stacked boxes against Rutgers, MSU, Washington, etc. I think the view of the blog at the time is there was no way that would work against OSU and we were setting ourselves up for frustrating failure. Instead, the power running flipped the script against OSU and suddenly the offense looked more palatable. Offense UFR in 2023 looks great as well, but that's pretty consistent with having the most skill talent on O that we've seen at Michigan in a long time, including a generational QB. I suspect 2022 is the real data you want to see.
Maybe during the bye week I'll work on re-scraping the data from the end of 2021 and all of 2022, so that the complete data set is available. In the meantime, you can see the data here: https://mikegros.shinyapps.io/ufr_analysis/
October 25th, 2023 at 10:31 AM ^
Thanks for providing the link. Will take a look!
October 25th, 2023 at 10:35 AM ^
For what it's worth, after the comments made by James Franklin I went to take a look for the scenario he described and I couldn't find it. Our coverage shell is in the UFR and while there's plenty of two-high, there's very very little 3rd or 4th and short and none that fit the described profile.
That doesn't alone mean anything; I think it's pretty clear in context that Franklin was explaining how you might have an idea that signs are stolen and what a suspicious play might look like rather than giving an oblique hint about a specific actual event. I don't think there's likely to be an in-game smoking gun.
October 25th, 2023 at 10:35 AM ^
Statistically, there is an issue with just trending one analyst’s opinion on RPS - especially given that it is coming from a fan. Games we lose are going to inherently be viewed more negatively and games we win more positively - and all of this screwed through the pseudo-objective nature of the analysis and of being a fan.
October 25th, 2023 at 11:16 AM ^
I just happened to have compiled these stats in a comment I made a few days ago. Here's our defensive RPS scores this season (without MSU yet, obv):
ECU: +3
UNLV: +1
BG: +6
Rutgers: +4
Nebraska: -1
Minnesota: +7
Indiana: +11