arrow-circle arrow-long-stroke arrow-stroke arrow-thick arrow-thin arrow-triangle icon 2 baseballCreated with Sketch. basketball calendar category check-circle check-square check comment facebook-circle facebook-icon facebook-rounded facebook-square facebook-stroke football instagram-circle instagram-icon instagram-square long-arrow-right rss-circle rss-rounded rss-square rss-stroke rss twitter-circle twitter-icon twitter-rounded twitter-square twitter-stroke user-group user

Tempering Steele

By on August 1st, 2014 in Football 3 Comments »

                                                           A properly tempered sword will sport Auburn colors*

First off, let me say that I enjoy reading Phil Steele’s magazine and daily blog. If ever I need to know how many starting offensive linemen are returning for a given team year to year, or how each SEC team’s yards per play compares to any other at a glance, it’s just a quick search in his archived articles. His annual articles related to such data and his comprehensive magazine are anticipated by college football fans for weeks before thier publication dates. They’re a wonderful blend of hard numbers and oftentimes very astute analysis.

That being said, at other times much of Phil’s analysis is either puzzlingly vague or laughably absurd. Many of his regular articles are either overly simplistic or completely irrellevant in any rational analysis. Why this occurs in amongst his other fine articles is unknown. I can only surmise that as his entire business revolves around keeping his readers engaged, he sometimes chums the water every now and then just to keep them occupied while he works on other more relevant material.

For example, some of his more common articles involve the return of key position players for various teams. While this might seem to be an important factor in assessing a team’s potential, it is often misleading in the extreme. Take last year’s report on % of Offensive Yards Returning. I looked at what Phil reported and added the actual yards per game in 2013 for each of the SEC teams. Sorting first for Phil’s numbers from his mid-summer article shows what seems like important data on returning talent.


However, when you sort by actual yards gained during the season, you see that such analysis is completely useless in terms of predicting or analyzing what any one team will do.


Granted, he pegged a few teams correctly – Alabama and LSU started and ended in the top five and Vandy and the Gators stayed at the bottom, but few other teams are anywhere near what his numbers would suggest by the end of the season, least of all the team that led the SEC in running the ball last year. Marshall, Artis-Payne and Grant had zero yards in 2012, and although Tre Mason was a thousand-yard running back, it was under an entirely different offensive scheme that woefully under-utilized his true talent.

So my question is, if Phil Steele’s data is so often irrelevant day to day, what about his predictions for the coming year?

An aide to British military commander Field Marshal Haig wrote this in a report following a tank demonstration in 1916.

Predictions by even the most experienced and rational people are notoriously inaccurate, often spectacularly so. Mr Steele is not immune to this, and his latest predictions posted last month are no exception to this rule.


On the surface this seems an impressive list of likely candidates, especially given Phil’s extensive research which he describes in the article. His discussion is impressive, compelling and seemingly accurate. But before we take this (and the glaring omission of Auburn) to heart, take a look at what he predicted last year and more importantly how those “National Championship Mold” teams fared in 2013.


These were Phil’s most likely champions in 2013. Notice anything about them? Perhaps this tabulation might help.


While Phil had the eventual BCS champion Florida State and five other 11+ win teams in that list, he also had two LOSING teams listed, and two other mediocre teams who barely finished over .500 for the year. For all the detailed analysis he describes, that just seems to be a surprising number of low achievement teams for this to be a valid method of predicting outcomes. I mean, what if you just took the top 14 teams from the 2012 final BCS standing?


Florida State is still there, along with the horrendously bad self-blocking 2013 Gators. But what about at the rest of the field? In my selection there are eight 11+ win teams, half of the top ten teams in the final BCS standings for 2013.

Phil’s analysis must have taken long hours of detailed comparisons and study. Mine took about ten seconds. Just long enough to type ‘2012 season final BCS’ in a Google search window and choose the right link. Herein lies the problem with detailed statistical analysis, predictions, and making sense of the wealth of data generated by NCAA football. How each team performs this coming year will rely on a number of issues, of which game experience at positions and previous achievement may play a part, but by no means will that be decisive in predicting results. In fact, most of what is generated in preseason predictions is pure chaff (to use a safe-for-work term). Kernels of truth might exist in such analysis, but separating that out from the sheaf of irrelevant statistical data they are bundled with is all but impossible.

The limitations of this sort of analysis should be readily apparent. I don’t mean to imply that Phil Steele’s work isn’t worthwhile. It is, but only when you use descretion interpreting his predictions. They are by their very nature inaccurate, and his data at times misleading. Just because Slippery Rock returns 100% of their offensive line, and LSU is replacing both starting tackles doesn’t mean that the Rock will be more successful than the Bayou Bengals in the upcoming season. Many other factors involved in their overall performance will have a much greater impact than whether certain positions have multiple-year starters or not.

Factors like continuity of coaching staffs, quality of players, position coaches, offensive and defensive schemes, schedule and team morale defy precise measurement but often have a dramatic effect on team overall performance. Judging those qualities accurately requires a more profound understanding of each team and its component parts than can be found on in broad brush statistical analysis, however comprehensive it may seem. Phil Steele’s blog and magazine provide important data and good information on NCAA teams, but they do not convey complete understanding of each team’s potential for the coming season. For that you will have to dig a little deeper into each one, their coaching staff, players, offensive/defensive philosophies and quality of play as a unit.

But even then, the ultimate determination of how good or poor a team will be on the field of play during the season. For that, we’ll just have to wait and see how they do when the season starts. Until then, everyone is undefeated and for the true fan, will always have the potential to remain so for the year.

In Auburn’s case, I’ll put my money on Gus and the boys over Phil’s analysis any day. War Eagle.

It is usually a dangerous idea to test the temper of a Tiger

*Historical Note: Tempering is the process of applying heat to achieve greater toughness in steel by decreasing the hardness/brittleness while increasing ductility. Precise tempering at several temperatures across the entire cross section of a sword  enables it to be hard enough along the blade to hold a keen edge, but ductile enough at the spine to withstand severe distortion in close combat without shattering or deforming. Swordsmiths of antiquity perfected this technique visually, as the color of the metal changed in a pattern of orange, purple and blue at very precise temperature ranges. A properly tempered sword would be dark blue in the center (590 degrees F) and ranging to bright orange on the blade edge (439 degrees F) before being allowed to cool.

In other words, it had to look like an Auburn blade before it was cool(ed).


  1. wpleagle wpleagle says:

    Some thought provoking comments. As an accountant, I understand the importance of numbers, but I also know how hard it is to get accurate financial forecasts for a business enterprise. I think the conclusions are spot-on as regards the subjective factors, especially team chemistry. This last was so important in both 2010 and 2013.
    Good analysis and good read!
    War Eagle.

  2. wde1988 wde1988 says:

    Excellent analogy of the horse cavalry and the combat arm of decision: armor.

    I listen to Phil Steel on Sirius and I think I understand the jab. The guy has bama blinders on. While that bothers me… I will be the first to admit with the recruiting Saban does… he should be hard to beat. And he normally is in the top five every year.

    Last season Steel ate plenty of crow after AU wasn’t even in the mix and most said we wouldn’t even go to a bowl game. But he defended himself with the statistical information that he puts out to his subscribers. Hard to argue with a fanatic. Even one that is pretty objective and professional as Steele. I just don’t know how he can be every where at once. He certainly isn’t clairvoyant.

    On another note – I was flipping through face book yesterday and I saw the statistic that in the past ten years… there has been three teams that averaged over 10 yard a carry in the SEC. Auburn had two of them.


  3. AUcideng42 AUcideng42 says:

    I really enjoy Phil Steele’s preview magazines. He has some really good models, gives a great overview all FBS teams, and HAS MORE LETTERS ON EVERY PAGE THAN ANY OTHER PREVIEW MAGAZINE!! But you’re right; while his models/process may be one of the more accurate ones out there, he is still lucky to be right 60%. This isn’t much better than a coin toss, but it’s still more reasonable than many of the HOT SPROTS TAKES in the preseason market. College football has such a small sample size that 60% is still impressive. (NOTE: I’m always amused by how he inflates his success rate in the magazine by making BOLD CLAIMS that are only true because he covers such large bases in his predictions.)

    It reminds me of the coworker who used to sit next to and our March Madness office brackets. He watched many basketball games, would spend days working out scenarios after the brackets were announced, and probably subscribed to at least one of the “insider” ESPN or Yahoo! sites. I didn’t watch college basketball (outside of the tournament) and would spend 1 hr on Google before my bracket was due. Over 4 years he finished in the top 3 once, top half of the bracket (outside top 3) twice, and did terrible the other year. I finished in the top 3 once, top half of the bracket once, middle of the pack once, and terrible once. I’m sure over more years he would have beaten me more consistently, but in a small period we pretty much broke even.