[Goose]

I've always thought a good way to look at a coach is by looking at their win rate over a four-year period (trailing four years). I like this because it matches the (traditional) four-year player cycle. I know it's simple and there a million factors, but I think it normalizes for one-off over/under performance enough so that I could spot an obvious trend in either direction, but especially a decline. Once a threshold is crossed, it should be clear what has to be done.

Here are the trailing four-year win rates for Harbaugh and Shaw. I pulled these from the GoStanford results page. I did not make any adjustments for the shortened season.

Harbaugh - 2010 - 58%
Shaw - 2014 - 78%
Shaw - 2015 - 78%
Shaw - 2016 - 76%
Shaw - 2017 - 72%
Shaw - 2018 - 74%
Shaw - 2019 - 62%
Shaw - 2020 - 58%
Shaw - 2021 - 47%
Shaw - 2022 - 33%

I think it is pretty clear.  

Now for Taylor. Here is what he would have to do for the next two years to match Shaw's last four years. To make it simple, I assume the next two years have the exact same record and 12 game schedules:

To match Shaw's 33%: 5 wins per season
To match Shaw's 47%: 8 wins per season
To match Shaw's (and Harbaugh's) 58%: 11 wins per season
To match Shaw's 61%: 12 wins per season

If Taylor went 6-6 and 9-3 the next two seasons (which most of us would be excited to see), his win-rate would be 44%.  How do you feel about this?

IMO the slope of the "rectangular low pass filtered" data is the most important factor. If the slope is rising, things are getting better year over year. One can then ask how long it would take to reach an "acceptable" level at the current slope. It is obviously a lagging statistic but that may be what it takes to filter out the noise. OTOH, if the year over year values are static or dropping, in the case of TT a change will be required. I am guessing most on this board aren't even willing to wait for year four to get the first "filtered" value and year 5 to spot a trend.

[Goose]

I've always thought a good way to look at a coach is by looking at their win rate over a four-year period (trailing four years). I like this because it matches the (traditional) four-year player cycle. I know it's simple and there a million factors, but I think it normalizes for one-off over/under performance enough so that I could spot an obvious trend in either direction, but especially a decline. Once a threshold is crossed, it should be clear what has to be done.

Here are the trailing four-year win rates for Harbaugh and Shaw. I pulled these from the GoStanford results page. I did not make any adjustments for the shortened season.

Harbaugh - 2010 - 58%
Shaw - 2014 - 78%
Shaw - 2015 - 78%
Shaw - 2016 - 76%
Shaw - 2017 - 72%
Shaw - 2018 - 74%
Shaw - 2019 - 62%
Shaw - 2020 - 58%
Shaw - 2021 - 47%
Shaw - 2022 - 33%

I think it is pretty clear.  

Now for Taylor. Here is what he would have to do for the next two years to match Shaw's last four years. To make it simple, I assume the next two years have the exact same record and 12 game schedules:

To match Shaw's 33%: 5 wins per season
To match Shaw's 47%: 8 wins per season
To match Shaw's (and Harbaugh's) 58%: 11 wins per season
To match Shaw's 61%: 12 wins per season

If Taylor went 6-6 and 9-3 the next two seasons (which most of us would be excited to see), his win-rate would be 44%.  How do you feel about this?

IMO the slope of the "rectangular low pass filtered" data is the most important factor. If the slope is rising, things are getting better year over year. One can then ask how long it would take to reach an "acceptable" level at the current slope. It is obviously a lagging statistic but that may be what it takes to filter out the noise. OTOH, if the year over year values are static or dropping, in the case of TT a change will be required. I am guessing most on this board aren't even willing to wait for year four to get the first "filtered" value and year 5 to spot a trend.