Gainwell

Considering those an extension of the running game is what people do when they want to make excuses for a poor running game. The backs were incredibly valuable as receivers, but when it comes to the primary job of running the ball, they left a lot to be desired.

 

Here we go with the subjective "advanced metrics" in place of the actual measures of what the running backs did. For all its flaws, the Steelers' passing game was better in 2025 than it had been in years, but the running game got worse.

Using the anecdotal evidence of the Ravens to dismiss the running game is even worse. It ignores the other teams that ran the ball well and did have great success. The AFC Champion Patriots were sixth in the league in rushing. The Super Bowl Champion Seahawks were 11th. The worst rushing team in the playoffs, of course, was the Steelers.

 

As shown in the previous thread, our RBs were better at running the ball this year than our RBs last year. It's only by a very surface level look (total yards) that we were better in 2024.

We just didn't run as many total plays this year.

I think this is mostly because the defense was generally not good at getting off the field. Except by turnovers.

The running game still needs to be improved IMO. But the bar from the 2024 season was pretty low.

They were also better at receiving. Part of this is because Gainwell was kind of forced into being the WR2 because we only had 1 NFL WR on the team in 2025. But we only had 1 NFL WR on the team in 2024 too.

If our 2024 RBs only got the number of plays our 2025 RBs did, the running game in 2024 would have been even worse.

 

If you go from 29th in pass attempts to 17th I sure would hope it would be better. Meanwhile, they went from 4th in rush attempts in 2024 to 28th in 2025. They ran some of the fewest plays on offense and they didn't run the ball all that often. Maybe those are numbers you can understand.

 

Who knew that "carries" was an "advanced metric"?

Also, Scribe still doesn't seem to understand the definition of the word "subjective".

 

You are assuming they have the 2024 passing game, which is not an apples-to-apples measure. The much-improved passing game made it easier for the backs to find room to run in 2025. Give the 2024 backs the 2025 passing game and they likely become a lot more effective.

It is not as if the running game was consistently better in 2025. It wasn't. It was inflated by two plays against a Detroit defense that had been decimated by injuries. Don't believe me? Those two long runs were the difference between Warren averaging 4.15 yards per carry and 4.5.

 

Your hypothesis here is testable.

It wouldn't be if all we had was YPC. But we have YPC and succ%.

If you are correct, YPC would be higher, but the increased succ% from those two plays would be negligible on the total carries for the year.

So we'd see a higher YPC (from the big carries), but the same succ%. Presumably worse succ% because you seem to be saying that last year was better.

I'll let you look for yourself. But these numbers were provided to you in the old thread that showed that YPC was about the same before those two long runs (and significantly higher after), but succ% was appreciably higher.

But I'm glad to see that you've changed your mind about long runs driving YPC.

ETA: How different was the passing game this year? Rodgers was better in the red zone, so that might account for the higher scoring. I haven't looked, but I think I remember seeing many posts comparing Rodgers this year and Wilson last year saying that outside of TDs (clearly super important) they were close to a push.

 

As we have discussed, success percentage is subjective at best, not a valid statistic.

I haven't changed my mind about long runs driving yards per carry because that isn't always the case. This is one incident that one player's numbers were drastically inflated by two long runs, but steady high-level production can also raise an average per carry. The reason those two runs were so impactful for Warren is that his volume of carries was relatively low for a No. 1 running back. He only carried the ball 211 times. Harris never carried it fewer than 255 times in four years as the Steelers' top back. That is why ignoring the value of volume, as many here want to do, leads to misleading and inaccurate analysis.

Let's take the year Harris had his fewest carries as a Steeler, 2023. Adding two 45-yard runs to his total would have only boosted his average per carry from by 0.28 yards. The same two runs boosted Warren's average by 0.35 yards because he carried the ball 44 fewer times.

 

It doesn't matter what did they win. I could use that exact same argument for every team that did reach the SB, doesn't mean you don't want it. That's a lame argument.
Texans had the number 1 defense , and they didn't make the SB. So should we not want a great Defense??
If You want to be able to control a game, you need a good running game. People complained all the time this year when we would grab a lead we couldn't hardly hold on, because we could never establish a true running game. It was very sporadic.
There is a difference in 4 yards an average that people don't want to admit.
One guy can run for 3, 4,3,5,4,0, 4,6,2,3(not exact numbers), and get a 4 yard per carry average, and another can run for 0,1,0,4,0,2,-3,-2,6,0,50(again not actual numbers), and still get a 4 yard per carry average, but it's not the same , and You can't establish a running game that way.

 

It's very obvious that the number and length of the long runs that skew the mean upward will determine how much they skew the mean by. As will total number of carries (the reason why rate stats are important)

Stating that isn't the same thing as saying that they don't skew the mean.

Also, it's a pretty good bet that you won't have to worry about a 45 yard run skewing Najee's average. He's had > 1000 carries and hasn't even had 40.

Which is exactly the reason that his YPC (mean) is going to be close to his mode or median. And while I don't have the data, it seems pretty obvious that the median and mode for any competent RB is going to be pretty much the same.

So the mean is mostly a measure of how frequently they get big carries.

Just so we're clear, the word "frequency" here is a "rate stat" so it accounts for your correct observation that the number of carries matters.

And since a guy like Najee has had so many carries in the NFL, it's unlikely to expect this frequency to change. Because as "N" increases, you get closer to the "true" mean.

So if he did get a 40+ yard carry, it would affect his YPC in that season more than his career YPC.

 

Exactly. It is funny how folks were looking for excuses for Warren's average per carry when it was right around Harris's going into the Detroit game, then those same folks decided he was the greatest thing ever again after two plays.

 

I'm not talking about career averages. I'm talking about averages within one given season.

 

But you were talking about changing the number of carries in the YPC (the denominator in the mean).

I was describing to you how increasing the "N" in the denominator will always reduce the impact of any single carry.

So it was clearly going to be the case when you add Warren's 2 big runs to Najee's season, they have a different impact. Because the N is different.

When you take that to the limit, you look at career stats. Especially for a guy like Najee who's already around the top 100 RBs in carries in a career. I can't remember if he's inside or outside this number. I looked it up a while ago.

 

Then let's compare the average and the median (50%tile) run yards gained for the top 25 rushers. This will tell us how much the big runs skewed their average. So Warren's average wasn't skewed like some of the top rushers in the league. And all the top rushers benefit from big runs (avg greater than median).

2025 Top 25 rushers (yards):


(btw, Najee last year had a median rush gain of 3 yds, same was Warren this year)

 

Where did you find this?

That's great data.

It will never happen, but I wish they would plot the distribution of yards on every carry somewhere. Then we wouldn't have to explain how stats work to show people that Najee had a low YPC because (1) he didn't have a long tail that skewed up the mean and (2) his median was low (which I would have guessed from watching games, but never knew before this data). Because in theory you'd be able to just plot the distributions on top of each other.

Maybe a job for clawbot?

Failing that, I've often thought seeing the mode and median would also help. So thanks for posting this!