In-Depth Analysis of Running Back Skills

[Grid Iron]

Introduction

When evaluating running backs, I’ve always struggled with which skill category is most important to the player’s success. To help me decide, I decided to run this experiment in to test the impact of the five primary running back attributes: 1) breakaway speed; 2) power inside; 3) hole recognition; 4) elusiveness; and 5) speed to the outside.

I did not include third-down running, as that is a purely situational stat, which I may examine more closely some other time.

Methodology

To conduct this experiment, I set all running back endurance scores to “9” and the overall skill scores to “0”. I then set each of the five skills for all running backs to “1”, except the attribute being tested, which was set to “9”.

To conduct the simulation, I set all options to be CPU-controlled and injuries to “0”.

I simulated 10 seasons for each of the five attributes, as well as 10 seasons for a control group (all “1”s for RB skill scores). After the simulation, I dropped the highest score and lowest score and arrived at my results.

Results

Below are the results, expressed in average yards per carry for the entire league, listing the high, low, median and mean.

                     High   Low    Median  Mean
Control Group        3.79   3.73   3.75    3.75
Breakaway Speed      4.07   3.97   4.02    4.01
Power Inside         3.76   3.69   3.74    3.73
Hole Recognition     4.11   4.04   4.09    4.08
Elusiveness          4.07   3.89   3.99    3.98
Speed to Outside     3.71   3.64   3.66    3.67

If you would like to see the raw data, click here.

Conclusions

While I am no statistician, here are some of the things I see:

• The Control Group performed poorly as expected, as all RBs had skill scores of 1. Interestingly, it was the most consistent from season-to-season, varying just 0.06 from the high to the low.
  
  
• Hole Recognition appears to have the most impact on a running back’s success. The 4.08 average over eight simulations is 0.07 greater than the average for Breakaway Speed, and 0.10 greater than Elusiveness. Moreover, the lowest seasonal yard-per-carry average with the Hole Recognition attribute was 4.04!
  
  
• Breakaway Speed and Elusiveness appear to have an equally positive impact on a player’s running ability.
  
  
• Power to the Inside doesn’t seem to add anything to the running game, being relatively consistent with the Control Group.
  
  
• Speed to the Outside appears to be detrimental to the running game, averaging 0.08 less yards per carry than the Control Group. How can the results be less than the control group? I suspect that since the running backs had a high Speed to the Outside rating, the CPU implemented a higher number of outside runs. The results don’t look very good. It seems to me that outside running ability, absent some additional skill, hurts more than it helps. However, it may be that a high number of outside runs will, over the long haul, result in poor rushing numbers.
  
  

[DaddyTorgo]

interesting. makes sense. although i'd think that to get a true recognition of "speed to the outside" you'd very much have to make sure you had great run-blocking tackles, because poor run blocking tackles would make that not the fault of the running back. At least so I've always assumed/planned for. Interesting that Elusiveness and Breakaway Speed are roughly even. I think I personally when I have seen successfull running backs have seen moderate hole recognition combined with either elusiveness or breakaway speed being high.

[sabotai]

Nice work. I've thought for awhile that Hole Recognition was the most important attribute, but never had any hard data to support it. Interesting, that Speed To The Outside. You're right in that the AI probably set a much higher rate of outside running. I think awhile back, Jim said (or maybe even said in the help file, dunno) that running to the outside, over time, yielded lower averages but a higher chance to break a run for a huge gain. That might be what we see here. It would be interesting if we could check the data run by run (whishful thinking I guess. )

Now the question is, do anything of thes attributes work in conjuction? Maybe Speed To The Outside is completely meaningless without any bit of Elusiveness. Perhaps, with the higher rate in runs to the outside, they failed a lot more because the RB was terrible at an attribute that Speed To The Outside requires. Prehaps Outside running needs Elusiveness in order to be effective. Maybe Power Inside needs Hole Recognition to be effective....just thinking out loud...

[Loki]

It seems that everyone that runs these kinds of tests is looking for season-long, league-wide trends.

I have a theory that many ratings are only situational, and don't show up as a positive on season stats.

An example is power inside. I would imagine that more power inside would result in a better 2nd/3rd and short conversion rate. Perhaps is a back got more 2nd/3rd and short opportunities, it would lower his overall rushing average.

In fact, I'd argue that most of the ratings in FOF can't be looked at in the way most of these experiments look at them.

[sabotai]

Loki brings up an interesting point about Power Inside. We speculate that Speed to the Outside added to outside running percentages and could be the reason for the decrease in average. Poerhaps the reason we don't see an impact on averages for Power Inside is because the AI ran the ball more in short yardage situations, cause a drop in average because of the typical 1 and 2 yard runs you see in those situations.

[Young Drachma]

Great work here. I'm amazed at the detail.

[Grid Iron]

Originally Posted by sabotai

Now the question is, do anything of thes attributes work in conjuction? Maybe Speed To The Outside is completely meaningless without any bit of Elusiveness. Perhaps, with the higher rate in runs to the outside, they failed a lot more because the RB was terrible at an attribute that Speed To The Outside requires. Prehaps Outside running needs Elusiveness in order to be effective. Maybe Power Inside needs Hole Recognition to be effective....just thinking out loud...

I'm running several combinations right now and I'll post the data when I'm done (could be tonight, could be next week since I'm going out of town tomorrow).

From what I've seen so far, speed to the outside is either neutral or drags the rushing average down slightly when combined with breakaway speed, or hole recognition or elusiveness.

There is a huge jump when hole recognition is combined with breakaway speed and a big, but slightly less jump, when hole recognition is combined with elusiveness.

Right now, my hypothesis is if you need a RB, look for one with high hole recognition and breakaway speed, as well as third down running.

[condors]

very nice work! thank you

[Grid Iron]

Attribute Combinations

I decided to run each pair of running back attributes together to determine the degree of impact they have on average yards per carry. In this experiment, two of the attributes were set to 9, while the other three were set to 1. Endurance was set to 9 for all RBs, and injuries were set to 0.

Since there were so many combinations, I ran five seasons, dropped the lowest and highest and averaged the three remaining values.

Here are the results in average yards per carry over three seasons:

           Brkaway  PwrInsd  HoleRec  Elusive  Outside
------------------------------------------------------
Brkaway    ---      4.05     4.46     4.36     3.96
PwrInsd    4.05     ---      4.06     3.88     3.61
HoleRec    4.46     4.06     ---      4.39     4.06
Elusive    4.36     3.88     4.39     ---      3.86
Outside    3.96     3.61     4.06     3.86     ---

I also ran Speed to the Outside with the Hole Recognition and Breakaway Speed Combination:

                     With 1      With 9
                     Speed to    Speed to
                     Outside     Outside    Difference
------------------------------------------------------
Hole Recognition     4.46        4.35      -0.11
& Breakaway Speed
Set to 9

This was a very interesting experiment that should help in putting a successful FOF2K4 roster together.

I updated the raw data file here.

Here are some bold conclusions that should serve as interesting material for debate:

• The single most important attribute for a running back is Hole Recognition. It’s hard to argue against this, given that with high Hole Recognition and Breakaway Speed, you get a lethal 4.46 yard per carry average! No other combination comes close.
  
  
• It seems the top three attributes to look for are: Hole Recognition, then Breakaway Speed, and then Elusiveness.
  
  
• Power Inside does not seem to provide any benefit over the long haul. Rather, it probably provides a situational benefit. However, this skill is probably better reserved for the fullback. The average yards/carry dropped when Hole Recognition was combined with Power Inside. This may be the result of the CPU game planning for more inside runs, thereby dragging down the overall average.
  
  
• Unless you do all of your own gameplanning, get players with little or no Speed to the Outside ability. It seems that with players that have high Speed to Outside, the CPU game plans for a lot of runs to the outside, which drags down the average yards per carry. Look what happened when I added a 9 for Speed to Outside to the running backs with 9s for Hole Recognition and Breakaway Speed—the average yards per carry actually went down by more than .10! The bottom line, I believe, is that outside runs are severely overrated. I'll keep it between the tackles with my 4.46 average any day!
  

[gstelmack]

Power Inside is definitely not important for ypc. It is critical for attaining short yardage downs, however. I've had one RB on my team the last few seasons whose only good attribute is Power Inside. His rookie stats (only full season he's played as other RBs have been brought in since): 178 attempts, 536 yards, 3.01 ypc, Longest 18, 7 TDs. He was used as a short-yardage back and performed magnificently.

Yards-per-carry can be a misleading stat. At the very least I'd love to see TDs added to these totals.

[Grid Iron]

Originally Posted by condors

very nice work! thank you

No problem! It was very interesting to see how it turned out.

[Grid Iron]

Originally Posted by gstelmack

Yards-per-carry can be a misleading stat. At the very least I'd love to see TDs added to these totals.

I agree, but there is such a big difference between 4.46 and the 4.06, it's hard to ignore the significance.

Of course, if you can get the super stud RB, you need to go for it, but for those who are scraping by with crummy RBs (like me in the CFL), being able to narrow down the skill sets you look for is crucial.

I think I may try to track a few RBs with various skill sets over a few seasons to see how they do compared to one another. But that won't be for a week or so.

[wade moore]

I'm concerned about some pieces you did not mention here.

What was the situation with the OLine? Was it different for every RB?
What wsa the gameplan? Was it different for every RB?
How many Carries did each have each year? This could have a MAJOR impact.
Did they all have the same QB behind them?

[Warhammer]

It is hard to place a definite value on any of these skills, but let's look at what they represent:

Hole recognition: The ability to find the hole and hit it.
Breakaway speed: The ability to take it all the way after you have cleared the line
Speed to the outside: The ability to turn the corner on an outside run
Power to the inside: The ability to get those tough yards in short yardage situations
Elusive: The ability to make the defender miss

Now, intuitively, I would want a good hole recognition whenever I can get it. You want a ball carrier to find a seam. Now the rest of the figures you have are a little more misleading.

With a high breakaway speed, you have improved your chances of taking the ball 80 yards immensely! This has a huge impact on your YPC over the course of the season. Rushing TDs should go up, because they can be scored from anywhere on the field.

Inside power will result in lower YPC, but you will probably see many more first downs for the club. So if you are measuring results by YPC, this would be down, but points scored or first downs should be up over the course of the season.

Elusiveness should increase any combination slightly, because the carrier is harder to take down. So where I might be able to get 2 yards with one back, with a more elusive back I might get a few more yards, or combined with breakaway speed, I might be able to take a few more to the barn each year.

Speed to the outside is a high payoff, high risk option. Think about most big rushing losses come on sweeps and counters that try to take the ball outside. When taking the ball up the middle, you are immediately moving forward, rather than moving laterally along the line, before making your cut.

I think a study for this requires knowing much more about everything. For example, how good is the OL? Is the line stronger inside or outside? Are you running mainly from the Pro Formation, or are you throwing in some runs from the Single-Back set? Any of these answers can seriously skew the results.

My own anecdotal evidence backs up that hole recognition is the most important stat for an RB, but I would argue that Elusiveness is second, followed by speed to the outside or inside power, depending on your OL. Now, if you go with a back with inside power and poor speed to the outside, do not try running ouside of your tackles very much, keep the runs between the tackles. You might not break a huge run, but you will be moving the ball much more consistently.

Let me put this another way, do you prefer a QB with a 50% completion rate, but 9.8 YPA, or a QB with a 70% completion rate, but only 6.5 YPA? My vote is the 70% completion guy because I am always moving the ball forward, but others might prefer the 50% guy because of the increased chance to hit the big one!

[Darkiller]

great work Grid ! brilliant.

[Grid Iron]

Originally Posted by wade moore

What was the situation with the OLine? Was it different for every RB? What wsa the gameplan? Was it different for every RB? How many Carries did each have each year? This could have a MAJOR impact. Did they all have the same QB behind them?

Gameplans were set by the CPU.

Since my results are league-wide running stats, and not a particular team, differentiation in Olines, QBs, etc... would be averaged out over the long term, especially with so many seasons run.

In addition, I ran the same season over again and again (I didn't sim out several years), so each time I ran a season, the QBs were the same, the Olines were the same, etc. . . The only difference were the RB abilities.

You can check the number of carries by looking at the raw data.

[Grid Iron]

Originally Posted by Warhammer

I think a study for this requires knowing much more about everything. For example, how good is the OL? Is the line stronger inside or outside? Are you running mainly from the Pro Formation, or are you throwing in some runs from the Single-Back set? Any of these answers can seriously skew the results.

I disagree with you there.

The CPU set the plan, and each time I simmed, the QBs, Olines, etc. . . were the same. The only thing different about the rosters were the running backs.

Moreover, examining league-wide running numbers over several seasons will even out any major differentiation.

Again, I'm looking a general trends, and Hole Recognition with Breakaway Speed/Elusiveness had consistently far superior resuts than other combinations.

Toward the end, I was able to predict what the year-end numbers would be.

[Grid Iron]

Originally Posted by Darkiller

great work Grid ! brilliant.

Thanks! Hope it helps!

[wade moore]

Originally Posted by Grid Iron

Gameplans were set by the CPU.

Since my results are league-wide running stats, and not a particular team, differentiation in Olines, QBs, etc... would be averaged out over the long term, especially with so many seasons run.

In addition, I ran the same season over again and again (I didn't sim out several years), so each time I ran a season, the QBs were the same, the Olines were the same, etc. . . The only difference were the RB abilities.

You can check the number of carries by looking at the raw data.

When you said "all running backs", I did not realize you meant every running back in the league.

I stand corrected, I have no issues now.. great research! ...

[dixieflatline]

Originally Posted by Loki

I have a theory that many ratings are only situational, and don't show up as a positive on season stats.

An example is power inside. I would imagine that more power inside would result in a better 2nd/3rd and short conversion rate. Perhaps is a back got more 2nd/3rd and short opportunities, it would lower his overall rushing average.

I disagree with this. If the entire league is converting the 3rd and 1 at a better rate they are gaining more yards than a league that isn't coverting as many. Sure this yardage difference is small but remember Grid Iron's study showed that these atributes are only changing the yardage by tenths of a yard. I believe that if power inside was helping to convert 3rd and 1 it whould show up in this study.

So if power inside isn't helping with these conversions maybe it's helping by reducing the variance of the runs by reducing the number of times the back gets stuffed. Low variance is good for the offense and if power inside does reduce the variance then it could be as useful as the other atributes.

So Grid Iron could you add a variance column to your study? Very nice work BTW.

[Buzzbee]

While it is certainly interesting to review the results of GI's study, I'm not sure how much we can read into it. The CPU controlled the gameplans. This causes me to question the results somewhat (not totally mind you). If the CPU controlled the gameplans, did it change the gameplan based on what kind of running back it had? If so, then you have introduced a new variable and we can't definitively peg the results on just the running back attributes. The variances could be the results of changing/differing gameplans.

Conversely, if the CPU didn't change gameplans, then the results again become questionable. A gameplan focusing on power running when you have a "speed to the outside" guy will obviously not work as well as one that spreads it around a little more.

Don't get me wrong. There is enough data and enough control to suggest that certain attributes might be a little more valuable than others, but I don't think there is specific enough control to to claim that Hole Recognition is by far the most valuable or that Power Inside is worthless.

Just my $0.02

[flere-imsaho]

Great, I read this after drafting a RB in the first round....

[weinstein7]

I have to side with Buzzbee on this one. Although the evidence does seem to point to hole recognition as the most important attribute, it is entirely possible that the AI does not gameplan as well to a RB whose specialty is power inside.

[-Mojo Jojo-]

Originally Posted by Grid Iron

The single most important attribute for a running back is Hole Recognition. It’s hard to argue against this, given that with high Hole Recognition and Breakaway Speed, you get a lethal 4.46 yard per carry average! No other combination comes close.

Makes me feel better about my eNFL trade for Onterrio Smith... Lethal, baby!

[Southpaw330]

is that Speed to the Outside - and to a lesser degree the other two constants (Breakaway Speed and Power) - are markers of a player who will improve over time. For instance, if I draft a back with high speed to the outside and good measurables, I expect his future rating to increase by at least 10 points. This isn't an exact science but I'd say it holds true seven out of ten tries and the odds probably increase to nine of ten when all three constants are high. If you draft based on hole recognition only, You probably won't have as much luck in improving the player.

[Cap Ologist]

What's the best way to pull data from this game? Is there a utility or did you just use print to file?

[Grid Iron]

Originally Posted by Cap Ologist

What's the best way to pull data from this game? Is there a utility or did you just use print to file?

For this experiment, I just wrote the results down by hand.

[chinaski]

Originally Posted by weinstein7

I have to side with Buzzbee on this one. Although the evidence does seem to point to hole recognition as the most important attribute, it is entirely possible that the AI does not gameplan as well to a RB whose specialty is power inside.

Another thing to consider is the the known formations for the QB at the time. You can a hell of a time running the ball with a QB who doesnt have good running formation knowledge... and of course that OL thing

[OldGiants]

I was glad to see this thread. Not only have I been researching/playing with similar strategies and wanting to find the time to comment, but this thread is about FOF!! No American Idol, Survivor or other worthless crap by that putrid piece of pig poop, Kodos.

Anyway, I have come up with a way of finding late round (5th or lower) RBs that mostly matches Grid Iron's analysis. What I have been doing is sorting based on breakaway, then looking for the best combination of Hole Rec and Power Inside I could find. After I have my candidates, I go back to the main screen and see how these guys look in the skill sorts. If they are at or nearly top in 2 or more, I draft them. (If a guy is way out in front in all skills, he's not so much a gem as a 'sleeper' with the scouting mask applied. See QuickSand's thread and my comments therein for more on that if you haven't read it).

My game is based on inside running, so I ignore Outside. I look for Elusiveness, but if he has Elusiveness to go with the other three, he's likely to have a higher scouting combine rank and therefore be gone already. Further, I look for guys with no to minimal pass catching skills. Why? If he runs for 1000+ AND catches passes, he'll ask for money than I want to give him when his contract comes up.

Endurance doesn't matter. Its easy to find three of these guys who will get you 2000 yards rushing combined,with all three averaging around 4.5 per carry. And you can resign them all. this lets you devote your salary cap to the OL.

Get a third down back who can block and catch passes, a FB H-back type with low route running is easy to come by, so you can let your 3 WRs get the passes.

[OldGiants]

Elusiveness or Hole Recognition?

If you can't have them both, which is better?

That's the question I started out trying to answer. Grid Iron's study provides nice numbers, and his conclusion is thoughful, BUT...

it is based on the concept (and ratings) that all other things are equal, which they never are. Particularly in FOF.

So my questions are:

1) What do you do if you have either a crappy O-line or one that is strong in pass blocking and weak in run blocking?

My initial thought was that Hole Rec is worthless if your O-line isn't making any holes. Can your back think "Gosh, there's no hole, so I'd better be Elusive?"

2) In game terms, if your line is crap, is every RBs Hole Rec skill diminished to the point where it is so worthless you are better off getting Elusiveness?

This would negate Grid Iron's conclusion about Hole Recognition, which I otherwise agree with.

3) If you are pass-oriented on offense, is Breakaway/Outside/Elusiveness the better skill set? That is, without run blocking, is that the way to go?

4) Can we put a "price" on skill combinations? That is, what set of skills produces the cheapest RBs who can get the job done?

Hopefully these questions will trigger more great discussions.

[primelord]

FWIW. Possibly the IHOF's best running back Ronnie Kemp has ratings in the 30s in both Hole Recognition and Elusiveness. Last season he ran for 2,010 yards for a 4.5 YPC and 15 TDs.

[Chubby]

Originally Posted by primelord

FWIW. Possibly the IHOF's best running back Ronnie Kemp has ratings in the 30s in both Hole Recognition and Elusiveness. Last season he ran for 2,010 yards for a 4.5 YPC and 15 TDs.

i think the OL had a little to do with that...

[Grid Iron]

Originally Posted by primelord

FWIW. Possibly the IHOF's best running back Ronnie Kemp has ratings in the 30s in both Hole Recognition and Elusiveness. Last season he ran for 2,010 yards for a 4.5 YPC and 15 TDs.

I don't know what his oline was like, but I do know what mine was like in the
most recent CFL season. It was terrible in the run game.

I had several RBs with 0 hole recognition and high outside running. My running game couldn't do anything. That inspired me to run the experiment above.

Based on what I found, I signed a FA with a 50 hole recognition, 50 elusiveness and 50 breakaway speed and basically zeros in everything else. His overall rating was 31/36.

By the end of the year, he finished one yard short of the franchise record for rushing yards in a game and racked up 396 yards on just 75 carries (5.28 yds/carry) in five games.

[Daimyo]

Nice work... I think it might be interesting to see the initial experiment run twice, once with all OL set to 1 and again with all OL set to 9. Intuitively it seems that "hole recognition" would be the key rating for a team with a nice OL and "elusiveness" the key rating for a team with crappy OL and "breakaway speed" being a nice secondary skill to both.

[mcsestretch]

Nice work GI!

[Chubby]

Originally Posted by primelord

FWIW. Possibly the IHOF's best running back Ronnie Kemp has ratings in the 30s in both Hole Recognition and Elusiveness. Last season he ran for 2,010 yards for a 4.5 YPC and 15 TDs.

he probably could have run for 2300 yds and 5.0+ ypc (on the same number of carries) if buzz had gameplanned differently...

[mhass]

I just stumbled upon this, GI. Very nice and much appreciated. It confirms some things I think I'd 'felt' intuitively (Hole Rec. is big) and rejects some others (Breakaway and Elusiveness are equally important).

As for this:

Originally Posted by OldGiants

BUT...
it is based on the concept (and ratings) that all other things are equal, which they never are. Particularly in FOF.

I think Grid did as much as possible to hold the variable constant. The same backs ran behind the same lines in the same schemes with the same quarterbacks over the same schedule (same defenses) ten times or 160 games. Incontrollables, as I see it, would be weather and game situations themselves. I'd say that's a lot of sameness to smooth the averages.

Quite interesting. Gotta remember to check this Strategy forum more often.