A better way to rank tennis players all-time


Realistic rating: 1.5

Over at Deadspin last week, I wrote about a better way to rank tennis players than the current format of simply ranking the most grand slams won.

Most sports media members learned long ago how incredibly mindless the “who has the most rings” argument is—well, except Shannon Sharpe, and Dave Dameshek, and ok, well a lot of not-so-smart talking heads and writers. Still, most people are at least smart enough to understand that Robert Horry isn’t better than Michael Jordan. But for some reason, we still apply this methodology when ranking tennis players.

(Before you read any further here, you should probably go read the original post.)

There aren’t formal rules for leagues to adopt being proposed with this idea, but it still gets a “realistic rating” for the likelihood this methodology is actually adopted by analysts and the general public. The big issue to consider is simply that the bigger numbers get, the harder they are for analysts to easily spit out on TV or fans to bring up in bar debates. Smaller numbers like the number of rings (or in the case of tennis, really fancy cookie jars) are ones anyone can easily remember, and thus, they’re the ones we universally associate with certain athletes.

This isn’t to say we should just give up on ratings systems that produce metrics that have than two significant figures, but it’s simply an acknowledgement that for statisticians to push their ideas into the mainstream, they should take hurdles like this into consideration.

Is 1st-and-11 really an effective way to curtail offenses?


Drew Magary had an interesting idea on Deadspin yesterday about how to curtail record-setting NFL offenses: Make 1st downs 11 yards rather than 10.

This got me thinking about our society’s enslavement to base-10 principles (and how if we really wanted to keep with that enslavement, we should just go full metric and switch to 1st-and-10 meters, since 11 yards is roughly 10 meters.)

But is an increased frequency of first downs really the reason offenses are becoming so unstoppable? The data suggests otherwise. Via SportingCharts.com, here’s the data since 1992 on the percentage of drives that go three-and-out plotted against the average drives per season per team. (Admittedly, the three-and-out rate probably isn’t the best statistic to measure this, but it’s the best I could dig up and should pretty well mirror 1st down conversion rates.)

Drives Chart

As you can see, the three-and-out percentage has remained pretty steady the past few years but has dropped about half a percentage point from the early 2000s.

What has steadily risen over the last five years, though, is the total number of drives per team. This leads me to believe that offenses aren’t improving because teams continually pick up first downs but rather because they are scoring more quickly. This increased pace could be due to a few things: more yards per play, more passing plays (so the clock stops), or simply wasting less of the play clock.

I doubt the third reason (clock wasting) has much weight, but it certainly seems the first two (increased pace, more passing plays) play a big role in the increased number of drives.

So with this knowledge, how would I propose fixing this issue?

Probably just lengthen the field to 110 yards rather than change the distance necessary for a first down. Or conversely, I might suggest narrowing the field from 53 yards wide to 50 to combat the effects of spread offenses. (I remember seeing an argument somewhere recently that the spread offense has led to fewer injuries. I’d probably disagree… Sure, more players can find space and avoid contact, but the spread also allows both offensive and defensive players to use that space to gain more momentum and allow for bigger hits.)

So… Is this even an issue that needs fixing? Probably not. As long as there’s a solid balance between scoring drives and non-scoring drives, and it’s not skewed too far in either direction (i.e. making punting irrelevant), then the (slight) rise of offenses is fine by me. Sure, it might skew the data and make simple aggregate yard historical comparisons difficult, but that’s what z-scores are for.

Eliminating NFL Overtime


Realistic rating: 4.5

After the Week 12 Packers-Vikings game ended in a tie, making 2012-13 the first back-to-back seasons with ties since 1988-89, there was again much consternation about games ending in ties and how THIS IS AMERICA. TIES ARE FOR EUROPEAN PUSSIES.

Emerging from all the noise condemning the NFL’s OT rules was a very reasoned response by AdvancedNFLStats’ Brian Burke, who argued that the NFL’s current rules were poorly structured because they are contradictory to the extra period’s goal of avoiding ties while also prolonging the additional period.

This seems like a bit of a tautology, though. “______ is bad because it doesn’t deter the likelihood of what we’ve defined as bad.” But why are ties or prolonged OT periods something to avoid? (For a much less nuanced take on this debate, the increasingly irrelevant Sporting News offered a #HotSportsTake that basically says, “The current OT rules are bad because ties are bad,” and then goes on for many more paragraphs to basically say, “Ties are bad because teams don’t want to tie.”)

While everyone seems to focus on how to simplify the NFL overtime rules, here’s my proposal for the most simple solution: Not have overtime at all. Just end the game in a tie if it’s deadlocked after 60 minutes.

Because the NFL season is only 16 games (i.e. each game represents 6.25% of the season), it is the only sport that can accommodate (and in my opinion, should welcome) ties without significantly diminishing game-to-game excitement. In the NFL, a large swath of teams routinely miss the playoffs, earn first round playoff byes, are assigned different playoff seeding, or see major shifts in draft position by a difference of a single game, and not all games are created equal. The Cowboys’ Week 8 final possession nail-biter loss to the Lions counts the exact same in the standings as their 49-17 blowout loss to the Saints two weeks later.

The fact is that wins and losses are binary outcomes that are not nearly granular enough to truly determine the better team in a season that only produces 16 such outcomes. (In 82- and 162-game seasons, this binary outcome is not only tolerable, but necessary, as any more granularity would further sap the already-small level of game-to-game excitement.) There’s a reason statisticians don’t really look at wins and losses when analyzing teams. (Sadly, many fans and “experts” don’t even use wins and losses, but something far less granular to evaluate teams: championships.)

Now for those concerned about seeing a standings filled with records like 10-3-3 and 7-5-4, that hysteria is very overblown. Only about 7% of all NFL games go to overtime, or roughly one per season for every team. While relatively rare, ties would have the benefit presenting the standings as a slightly better depiction of a team’s actual body of work for each season.

The other big thing to consider here is injury risk and fatigue. Is it fair for a two teams to have to play an extra period, risking injuries to all their players and also further wearing them out for next weeks’ game? Overtime in a sport as violent as the NFL is almost a prisoner’s dilemma situation in which it would be mutually beneficial for both deadlocked teams to simply agree to a coin flip to determine the winner rather than for each of them to risk incurring negative effects on their respective futures.

So rather than trying to fix the OT rules many people seem to despise so much, let’s just leave them for this weekend’s games and the rest of the postseason.

Bottom line: This likely wouldn’t happen because of the heavily ingrained cultural aversion to dreaded ties. The only foreseeable way the NFL would eliminate OT is if the sport became so dangerous that any additional time playing the game was thought to be reason to avoid it.