Spring 2025

[KingInTheNorth]

Can someone shed some light on the Cmax rankings? I thought that I had a good handle on the methodology, but today’s release doesn’t make sense to me. Cal is ranked #1, with a 2.5s projected margin over UW —despite losing to them by 0.5s last weekend. Even if Cal beat the same teams that Washington raced by larger margins, shouldn’t the head-to-head supersede this? Please make this make sense. Thanks!

https://www.row2k.com/polls/index.cfm?c ... 20Rankings

[Mango]

Can someone shed some light on the Cmax rankings? I thought that I had a good handle on the methodology, but today’s release doesn’t make sense to me. Cal is ranked #1, with a 2.5s projected margin over UW —despite losing to them by 0.5s last weekend. Even if Cal beat the same teams that Washington raced by larger margins, shouldn’t the head-to-head supersede this? Please make this make sense. Thanks!

https://www.row2k.com/polls/index.cfm?c ... 20Rankings

There are some years where the cMax gets NCAA/IRA shockingly accurate right down to the margins. Then some years there is some shuffle. It isn't perfect but generally a good tool. But this is sometimes a flaw, especially early to mid season.

The algorithm takes into consideration every data point through every race across the country. So based on margins between all their common opponents. Even though Cal lost to Washington they have bigger margins against 1st degree and 2nd degree of separation opponents. Washington did halve their margin against Cal from the week prior. So it it isn't like it didn't take the result into major consideration.

There is still a month until the IRA. I suspect it will level itself out as more data points are collected.

[lt.wolf]

“
That was what I thought. I know that Olympians are not inherently faster just because they have the 5 rings, but it’s kind of insane that UW won with such a discrepancy. It’s not like Cal’s olympians are from countries that are developing their rowing programs.“


Well there was / is scuttle over Cal using Fat2 blades . I don’t take much stake in it unless it was a recent switch.

[lt.wolf]

This week is slow week of racing but we get

Brown vs Princeton
Stanford vs Cal
Dartmouth vs NU
Penn vs Holy Cross
Washington vs NZ??
Yale off
Harvard off
Syracuse off


Let me know if we missed anything

[LWTlightning]

Friday May 2, 2025
Oregon State v. Cornell (Camden, NJ)

Saturday May 3, 2025
Madeira Cup: Cornell @ Pennsylvania w/ Holy Cross (Camden, NJ)

Sunday May 4, 2025
Stevenson Cup: Navy @ Columbia (Princeton, NJ)

[Cowbells123]

MARC is this weekend.

[lt.wolf]

Brown > Princeton
Stanford <Cal
Dartmouth >NU
Penn >Cornell> Holy Cross
Washington over competition in Windermere
Cornell > OSU
Columbia > Navy

Yale off
Harvard off
Syracuse off

[yeahbuddy]

Brown < Princeton
Stanford <Cal
Dartmouth >NU
Penn >Cornell> Holy Cross
UW > NZ
Cornell > OSU
Columbia > Navy

[str8four]

Can someone shed some light on the Cmax rankings? I thought that I had a good handle on the methodology, but today’s release doesn’t make sense to me. Cal is ranked #1, with a 2.5s projected margin over UW —despite losing to them by 0.5s last weekend. Even if Cal beat the same teams that Washington raced by larger margins, shouldn’t the head-to-head supersede this? Please make this make sense. Thanks!

To add to Mango's point above, we all know that a boat (especially an eight) has a range of speed it can display in any given race or on any given day. Sometimes the boat steps up and has a max effort and focus race and performs at 99% of their fastest speed, and other times they have a mediocre effort and perform at only 80%. By using ordinary least squares to compare time margins between common opponents you can establish the range of speed for a given crew. CMAX then standardizes that by assuming every winner went exactly 6:00 even thought some races were done in 5:30, some 5:50 and some 6:15.

So when comparing Cal to Washington the data says that Cal has had the better relative (and standardized) performance against opponents. So if both crews had their average performance and the race was exactly 6:00, Cal would win by 2.5 seconds.

The CMAX data does factor in "improvement" by weighing recent results more highly than ones from a month ago. However, nothing is perfect as it can't control for the impacts of travel, psychological motivations, nor the random injuries, illnesses, and other life events that impact the performances of individuals, and therefore boats.

Lastly, IMO Cal is the faster crew and CMAX is the better quality poll when compared to coaches opinions. UW won on home water in moderately rough conditions by 5 feet. I recorded the race on the Big 10 network and I am surprised Cal only lost by 5 feet given how much better UW looked - longer, cleaner, more powerful. I would guess that UW executed a 95% performance and Cal had an 80% performance. I am pulling those percentages out of my butt just to demonstrate how a faster crew can lose any given race to an opponent with overlapping speed. I would have been impressed if at least one coached watched the video and then had the balls to vote Cal #1 despite the loss.

[Remomex]

I would guess that UW executed a 95% performance and Cal had an 80% performance. I am pulling those percentages out of my butt just to demonstrate how a faster crew can lose any given race to an opponent with overlapping speed.

If you are using percent of performance to mean percent of speed, then that would suggest a team like Cal can go (assuming fair conditions) ~5:30-5:35 at their best and 6:30+ at their worst. So losing to top end D1 women's 8+s.

I know you admitted those percetanges were pulled out of your *** but that's underselling it.

[lt.wolf]

Cal by about 10 seconds over Stanford today, any changes from last week?

[LoreMaster1]

Absolute barn burner between Princeton and Brown. 5:20 on Carnegie is insane.

[lt.wolf]

Princeton 5:20.8
Brown 5:21.3
New Carnegie course record.
Content Cup stays in Princeton.

[lt.wolf]

No more fat2s at Cal

[KingInTheNorth]

Thanks for the detailed explanations regarding the Cmax ranking and its methodology. I appreciate that it’s weighing a large body of data as a predictor of a crew’s overall potential, and how it will play out at the IRAs. I’d be interested to see how predictive this model has been in past years (which teams were over/undervalued etc, and the characteristics if these errors —similar to how Fivethirtyeight used to conduct post-mortems).

My original assumption was that Cmax was similar to college football rankings, where teams moved up or down in the rankings based on quality of wins, comparable opponents and margins of victory. What makes sense to me in this case is that relative wins are all speculation until two teams actually play —at which point the victor of that contest is almost certainly ranked higher than their vanquished opponent, unless the two teams are separated in rankings by a very wide margin (e.g. a massive upset). I find this to be a better model, as speculation should go out the window after two crews go head-to-head in essentially identical conditions. Cmax would still be relevant for predicting the outcome of two crews who haven’t yet faced one another, however.