≡ Menu

Is TD/INT Ratio Now Meaningless?

A pair of Crimson Tide/Jets quarterbacks in 1976

A pair of Crimson Tide/Jets quarterbacks in 1976

You remember Jets quarterback Richard Todd, don’t you? Before there was Favre/Rodgers and Montana/Young, Jets fans envisioned a Namath/Todd passing of the torch. Eleven years after New York drafted Joe Namath, the Jets spent the 6th pick in the 1976 draft on Todd: another quarterback from Alabama. For a year, the duo overlapped: a passed-his-prime Namath threw 4 touchdowns against 16 interceptions in 8 starts, while an inexperienced Todd had 3 touchdowns and 12 interceptions in 6 starts.

The duo even looked similar, shaggy hair and all, and you can forgive Jets fans for hoping that another Hall of Fame quarterback had come to them out of Tuscaloosa. Todd failed to meet those lofty expectations, of course, but he did lead the NFL in yards per pass attempt in 1979. Three years later, in the strike-shortened 1982 season, a 29-year-old Todd started every game for the Jets and posted the following stat line:

Year QBrec Cmp Att Cmp% Yds TD TD% Int Int% Y/A Y/C Y/G Rate Sk Yds NY/A ANY/A Sk% 4QC GWD
1982 6-3-0 153 261 58.6 1961 14 5.4 8 3.1 7.5 12.8 217.9 87.3 23 206 6.18 5.90 8.1 1 1

Does that stand out to you as particularly impressive? It may not against our modern eyes, but what about that 14:8 touchdown to interception rate?  Okay, maybe a 1.75 TD/INT ratio doesn’t scream superstar to you. After all, of the 35 quarterbacks last year to qualify for the league passing crown, 22 of them had a better TD/INT Ratio than 1.75. Heck, Josh McCown (3.00) and Brian Hoyer (2.71) eclipsed those numbers with ease.

But here’s the thing: in 1982, Todd led the NFL in touchdown/interception ratio.  Yes, a 1.75 TD/INT Ratio doesn’t sound good, but it was outstanding for his era.  But things change quickly, and Todd is the last — and will surely remain the last — quarterback to lead the NFL in TD/INT Ratio with a sub-2.00 rate.  When Ryan Fitzpatrick (31/15) or Ryan Tannehill (24/12) throw twice as many touchdowns as interceptions, it sounds good, but it really isn’t.  It’s just average now.  And given the randomness of interceptions, a few bad bounces can really torpedo a quarterback’s ratio given the rising tides at play.

The graph below shows the league-average TD/INT Ratio for every season since 1932.  This one speaks for itself:

TD INT ratio

TD/INT Ratio likely emerged as a “key” quarterback statistic because passer rating is incredibly complicated and silly, while passing yards are simple to understand but have almost no correlation to success.  TD/INT Ratio suffers from some of the same flaws as passer rating in that the correlation to winning is tricky — trailing quarterbacks tend to throw interceptions, so the causation arrow runs both ways — but that makes nice antidotes to pure passing yards (if you want a high correlation to winning) and passer rating (if you value simplicity).

Consider this: just ten quarterbacks in NFL history have a career TD/INT ratio better than the 2015 league average!  Those quarterbacks are Aaron Rodgers, Russell Wilson, Tom Brady, Steve Young, Peyton Manning, Tony Romo, Drew Brees, Philip Rivers, Donovan McNabb, and Joe Montana.   You may notice that other than Young, the top 8 players in this category were active last season.   McNabb was at exactly 2.00 when he retired.

But then think about Montana.  He retired with the best TD/INT Ratio in NFL history, courtesy of 273 touchdowns and 139 interceptions.  That’s a 1.96 ratio. Yet, given the recent trend lines, that will likely slow in as below-average in 2016. If that’s not a good enough reason to look at all TD/INT Ratios with a skeptical eye, then this is the wrong site for you.

  • sacramento gold miners

    Agree about the diminished importance of the TD/INT ratio, some QBs are just too risk-adverse with their pass selection. The lesser QBs practice “checkdown city”, choosing the safer short pass. It drives me crazy to see the eight yard completions on third down and 11. Looks good on the individual stat sheet, but a drive killer. Also, the passes well out of the end zone either at the end of the first half, or late in the game. Sometimes this occurs on fourth down, when a field goal would have been useless.

    • Adam

      “some QBs are just too risk-adverse with their pass selection. The lesser QBs practice “checkdown city”, choosing the safer short pass.”

      Alex Smith is the mayor of Checkdown City, and Jason Campbell is his chief of staff.

  • I guess we need TD/INT+ now.

    On another note, you’ve said many times (as have I) that QBs throw more picks when trailing and throw higher percentage passes when leading. It makes sense, intuitively, and it seems to match what I see when watching. However, I’ve never seen an empirical study supporting this. Have you conducted or read something along those lines?

    • Adam

      If you use to PFR’s play finder to look up INT% for leading teams and compare it to INT% for trailing teams, the leading teams throw a lower percentage of interceptions. This is true every year in which PFR has available data. Of course, there’s probably a small element of selection bias – throwing a lot of picks early will cause a team to fall behind, but I believe that effect is relatively minimal given the overall rarity of interceptions.

  • Adam

    TD/INT ratio was always an overblown statistic, but in today’s game it is essentially meaningless. TD% is highly influenced by goal line play calling, while INT% is clouded by game situation, playing style, and a heavy dose of randomness. Neither TD’s or INT’s say much about the quarterback’s ability, so when you put them together it just compounds the distortion. If we used TD-INT difference instead of ratio, it would be slightly less problematic, but still bad. The ratio method exacerbates randomness, as the more random of the two numbers (INT’s) is the denominator, which inadvertently gives INT’s several times the weight of TD’s. For example:

    QB A: 40 TD, 5 INT
    QB B: 10 TD, 1 INT

    Anyone in their right mind would choose the productivity of QB A over QB B, but TD/INT ratio thinks QB B is better. It’s a terrible metric, plain and simple, and football analysis would be better off without it.

  • Adam

    Another reason TD/INT sucks is because it’s focused exclusively on rare events, while ignoring the vast majority of a quarterback’s plays. Touchdowns and interceptions COMBINED only comprise roughly 6% of QB dropbacks. Do we really want to put so much emphasis on a metric that ignores 94% of a QB’s passing plays?

    • Richie

      An int could happen on any throw.

      • Adam

        True, but that’s what INT% is for. Generally when people use TD/INT ratio, it’s lazy shorthand for the entirety of a quarterback’s performance (while ignoring important stats like NY/A).

  • Tom

    I almost get the feeling that the stat was created to somehow “put a check” on thinking a QB is great because he throws a lot of TD’s. As in, “Man, Joe Blow is awesome, he threw 40 TD’s last year!” “Yeah, but hold on buddy, he’s not that great because he also threw a lot of picks”, etc.

    So it’s like, “Hey, let’s take the best thing that can happen when a QB throws the ball, and put it together with the worst thing that can happen”. In some kind of yin-yang way of thinking, maybe it’s alright, but it doesn’t seem like they can be related, as Adam notes.

    • Adam

      “but it doesn’t seem like they can be related”

      Exactly. If you’re going to use a ratio to measure something, the numerator should be a subset of the denominator (completions / attempts) OR the two numbers should be direct opposites (hot days / cold days). TD passes and INT’s aren’t a subset of each other, nor are they opposites; they’re two unrelated events.

  • WR

    I think it would be a mistake to stop using td-int ratio. If you want to argue that it isn’t particularly useful as a single season metric, I won’t argue too strenuously. But surely it still has use as a career measure. I don’t agree with Adam that “neither td’s or int’s say much about the quarterback’s ability”. Over the course of a full career, a QB who throws a lot of TDs while avoiding turnovers is going to really help his team. And the ability of players like Rodgers, Wilson, Brady, Mcnabb, etc. to maintain a strong ratio of TDs to picks is a major reason why their teams have won so many games.

    Baseball players hit a lot more home runs than they did fifty years ago. But a guy who’s consistently near the top in home runs is still likely to be one of the best players in MLB today.

    • Adam

      “Baseball players hit a lot more home runs than they did fifty years ago. But a guy who’s consistently near the top in home runs is still likely to be one of the best players in MLB today.”

      That’s true, but HR’s are a relatively pure measure of skill, whereas TD passes are not. Batters have roughly the same opportunity to hit a HR every time they come to the plate, and their chances of doing so are not influenced by teammate interaction. Furthermore, nobody in baseball is dividing home runs by an unrelated event and calling it a meaningful statistic.

      If TD/INT ratio had a baseball equivalent, it would be something like HR/GIDP (grounded into double play). Bill James would laugh at such a metric.

      • WR

        I see your point Adam, and I think you’re hitting upon the fact that all events in sports are the product of circumstance. I would argue this is also true of Home runs and GIDP, because a player sees different pitches depending on in game situations, and the talent in the lineup around him. Manny Ramirez playing with a bunch of AAA guys is going to see different pitches than Ramirez on the 99 Indians, because in the latter case, Thome and Sexson are coming up next.

        I agree with you that TD-INT ratio is a quick and dirty measure, and isn’t hugely reliable for a single season. But if a guy is near the top of the league in td-int ratio and differential for several seasons in a row, I can’t use that as evidence that he’s helping his team? Bryan has suggested TD-INT+, and I think that’s a good idea. All achievements need to be adjusted for era.

        I think we all recognize that football stats are impacted by teammates and play calling. For example, I believe that a QBs stats are heavily impacted by the quality of receivers and running backs around him. But I’ve yet to figure out how to quantify that reliably, but that doesn’t mean it shouldn’t be part of an evaluation. If there’s a baseball stat that’s similar to td-int ratio, it would probably be strikeout-walk ratio. The point being that pitchers who can strike out a lot of hitters while also limiting walks are the most valuable of all. Guys who have a lot of both, like Nolan Ryan, are a notch below guys like Kershaw and Pedro Martinez. That’s the same basic reason why Favre is a notch below Manning and Brady, who are way ahead of him for TD-INT ratio.

        • Adam

          “But if a guy is near the top of the league in td-int ratio and differential for several seasons in a row, I can’t use that as evidence that he’s helping his team?”

          This where I’m not sure which direction the causation arrow is pointing. Over a multi-year sample, I guess TD passes are a reasonably good indicator of how often a QB is moving his team down the field and into the endzone. But I don’t know about interceptions. All football stats are influenced by circumstance, but I think INT’s suffer from entanglement issues more than the rest. Brady, Rodgers, and Wilson have very low INT rates, but these three probably get to play from ahead more than any other QB’s in the 21st century. Wilson’s defense has only given up 16 PPG in an era where the average team gives up over 22 PPG, so he often gets to play with a nice cushion and doesn’t have to take risks; of course his INT% will be low.

          To be clear, my discounting of TD/INT is not a knock on the guys who excel at it. Guys like Brady and Rodgers are great for other reasons, namely their ability to move the ball down the field with high efficiency. If two QB’s have the same NY/A, 1D%, and rushing numbers, I think TD/INT is a good tiebreaker, but IMO it will always be a secondary statistic.

          • WR

            This is another case where we’re seeing things differently. There’s a chicken and egg problem here. Are Rodgers, Brady, and Wilson avoiding INTs because of the defense, or is their ability to avoid INTs one of the reasons why their defenses allow fewer points? Brady has about 100 fewer INTs than Peyton. Are you saying that a diff that big is because Peyton’s teams are more often behind, and he has to take chances downfield? I find that hard to believe. And despite his lower INT rate, Brady’s rate of TDs is nearly as good as Manning’s anyway.

            In 2010, Brady was the career record holder for td-int ratio. That season, he broke his own record for TD-INT ratio with 36 TDs and 4 picks. He also set the record for consecutive pass attempts without an interception. At the time, some of Brady’s critics claimed that it was all because of luck and circumstance, but at what point do we admit that avoiding interceptions is a skill? When Rodgers went 2 years without a pick in home games or whatever it was, it wasn’t just because his teams were always ahead late in the games. If Brady and Rodgers had weak TD rates but great INT numbers, I might agree with you. But they don’t. When guys like Alex Smith and Foles put up great ratios, that’s different. They’re not doing it with the same totals produced by Rodgers and Brady. But look at Wilson last year. In the last 8 games, he had a ratio of 25 TDs to 2 INTs while throwing over 2100 yards. That suggests to me that he’s gotten better as a passer since his first 2-3 seasons. And it’s not like Seattle was comfortably ahead in the 4th quarter of every game.

            As for the baseball stuff, I’m familiar with FIP, and I agree that k/bb ratio tells you more than td/int ratio. That doesn’t make td/int useless.

            • Adam

              I’m not trying to imply that INT rate is all a product of circumstance; there’s obviously some skill in it, especially over a full career sample. Brady is clearly better at avoiding INT’s than Peyton (and pretty much everyone else), and he deserves credit for this aspect of his play. My point is that we shouldn’t give FULL credit to the QB for his INT%, only partial, as the stat is very noisy. I disagree that Brady’s 2010 was special for INT avoidance, as 492 passes is too small a sample to say he played any differently than he normally does. He threw 13 INT the year prior and 11 the year after, so I tend to think the one season dip was mostly randomness.

              Do you think Damon Huard magically learned how to avoid INT’s in 2006, and forgot for the rest of his career? How about Josh McCown and Nick Foles in 2013? Or Josh Freeman in 2010? Or Steve DeBerg in 1990? Or David Garrard in 2007? If you’re going to attribute Brady’s 2010 INT rate to skill, then you have to do so for these other seasons, as well.

              As far as Rodgers and his no-INT streak at home, I think he’s way too risk averse, especially when playing from behind (yes, he had to play from behind on occasion). His failure to mount comebacks is well documented, and I believe his lack of willingness to risk throwing picks is part of the reason why.

              • WR

                I think your response actually reinforces my point. I agree that the examples you provided show that less skilled players can occasionally have low INT rates, and fail to reproduce that level. But what I’m saying is that rare players like Brady and Rodgers have repeated this skill, and therefore are more successful. Brady’s 2010 season was a historical outlier, even by his standards. But look at the rest of his career. He’s come close to reproducing those numbers, including in 2015. He has six other seasons apart from 2010, in which he had an INT rate of 2.0 or lower. He’s also gotten better at avoiding INTs as time has gone on, and the last season in which he wasn’t in the top 10 was 2005. I don’t think Foles and McCown magically got better in a particular season. But if they had maintained very good INT rates since then, I would judger their careers differently.

                Brady and Rodgers have the two lowest INT rates since the start of the 2007 season, and it’s not even close. Their skills appear to be repeatable. They’re also both in the top 4 for TD%. No one else can say that. And I agree with you that Rodgers may be too risk-averse. I’ve seen his weak numbers in 4QC/GWD stats. According to Packers fans who watch more of him than I do, that’s probably due to a lot of factors beyond his control, but it still doesn’t look good. Brady doesn’t have that problem. AFAIK, he still has better 4QC/GWD numbers than Manning and Montana, which doesn’t support the hypothesis that he avoids INTs by playing cautiously.

                • Adam

                  Tom Brady is probably the best INT avoider of all-time, I won’t argue that. But I still don’t agree that 2010 was anything more than randomness. His 0.8% that season was less than half his career rate, and QB’s don’t suddenly become twice as good at something for one particular season. If it makes you feel any better, I also think Peyton’s 55 TD season was mostly a product of randomness and ideal circumstances, not a true jump in skill on his part.

                  • Douglas Bath

                    I think you’d have to say Aaron Rodgers is above Tom Brady in that area. Not just because Rodgers has a 4.0 ratio while Brady has a 3.0 ratio. But also consider Tom Brady has 12 times as many career pick 6’s as Aaron Rodgers does.

        • Adam

          I think K/BB ratio for pitchers is a far better indicator of skill than TD/INT is for quarterbacks. Not sure if you’re familiar with FIP, but it theorizes that strikeouts and walks are the two stats a pitcher is most responsible for, while hits allowed is relatively random. Thus, we can say with confidence that Pedro is better than Nolan Ryan because his K/BB is noticeably better over a huge sample.

          It’s telling that Chase uses NY/A for predictive studies rather than ANY/A – he’s filtering out TD’s and INT’s because they’re not consistent from year to year. We can infer that TD’s and INT’s are noisy stats which are heavily influenced by factors other than the QB’s skill level.

      • Tom

        “…HR’s are a relatively pure measure of skill, whereas TD passes or not.” Yeah, this where I get hung up with TD passes as a stat in general. It’s a useful stat, but it seems like a lot of it has to do with play-calling, type of offense, etc. I haven’t looked at any numbers, have never played QB (besides this one awesome year in 6th grade during recess), and so I don’t know – are some QB’s just plain better at throwing TD’s? Or are they just better quarterbacks? Maybe a question for another post, but this is why the TD/INT stat doesn’t fly with me – I feel that avoiding INT’s is something that a QB can get better or worse at, but I don’t know if the same can be said for throwing a TD. I was thinking maybe a stat that showed how QB did when attempting to throw a TD pass, but that’s got all kinds of problems and wouldn’t be worth the effort, so I’m not sure.

        Again, it just seems like an attempt to put the best and worst thing that can happen on a pass play together in one stat. Although not quite the same thing, would we even consider a TD/fumble ratio for running backs? Of course not…TD’s for an RB are is an important stat, as is FUM%, but we don’t put them together in one stat (or do we? I’ve never seen it!)

        • WR

          But fumbles by a RB and INTs aren’t really the same thing, beyond the fact that they both count as turnovers. Throwing downfield is a much more complex task than just carrying the ball. A QB can avoid INTs by making pre snap reads, anticipating where DBs will be positioned, throwing accurately downfield, avoiding cross body throws, avoiding tips by DL at the line of scrimmage, etc. A QB can also cut down on INTs by avoiding risky throws altogether. But if he does this, he’s unlikely to maintain a high TD rate. Rodgers and Brady are the best at doing both. That’s a big part of why they have been the two most productive QBs in the league since 2007.

          • Tom

            WR – totally get what you’re saying. Yes, fumbles and INT’s aren’t quite the same, I brought it up because the TD/FUM thing smacks of the same kind of thinking as TD/INT. But they’re different, yes.

            But here’s the crux of the matter with your argument I believe – it’s your opinion that (not) throwing interceptions is related to throwing TD’s, based on your statement, “But if he does this [avoiding risky throws altogether], he’s unlikely to maintain a high TD rate.” And so this is what we need to look at. I looked at all QB’s from 1970-2015 who threw over 100 touchdowns (career) and the correlation between TD’s and INT’s for these passers is 0.70; meaning the more TD’s a QB has thrown, the more INT’s they throw (and vice-versa) which basically supports what you’re saying. Of course, the problem there is that the more passes of any kind that a QB throws, the more likely he is to throw both TD’s and INT’s anyway, so I’m not sure we can use that. How about rate stats? Here, we have a different story: looking at the same QB’s, we get a correlation of -0.03 when comparing TD% to INT%, which means there is no connection. And this makes sense to me, and it’s why comparing these two events is difficult. If you say to me that TD% and INT% aren’t the same because an INT can happen at any time, and TD’s usually don’t happen anywhere on the field, etc. and thus you can’t make this correlation, I’d completely agree with you.

            Take a look at the chart…some guys throw a lot of TD’s (per attempt) and a lot of INT’s. Some guys throw a lot TD’s and not that many INT’s, etc., etc. Yes, Rodgers and Brady are great because they don’t throw picks and they throw TD’s. I’m on board with that, and I agree that TD/INT perhaps has some interest in career stats, in a yin-yang/bad-good type of way. But I just can’t really take it seriously as anything other than that, in the sense that they aren’t really related.

            • WR

              Thanks Tom, that’s really interesting. If I”m reading the graph correctly, it suggests that TD% and low INT% are separate skills, that don’t correlate to each other. I agree with that. I guess what I am saying is that the ability to do both well is quite rare, and that when a player like Rodgers or Brady succeeds at both, it’s probably one of the major reasons why they have been successful overall. I’m not saying that a QB will throw more TDs BECAUSE he throws fewer interceptions, just that when someone does both, it should be recognized as an indicator that he’s probably a great QB.

              I did a quick and dirty study of my own, by looking at all QBs since 1950, and identifying the ones who had the best combo of TD%+ and INT%+. The top 8 names on the list were

              Rodgers
              Steve Young
              Tom Brady
              Joe Montana
              Roger Staubach
              Peyton Manning
              Len Dawson
              Dan Marino

              Russell Wilson has good enough numbers to make the list, but he’s only played 4 seasons. Now, it is interesting that 3 of the top 4 are guys who played in West Coast style offenses, so perhaps these results are the product of the system to some degree. Rodgers has never been described as a system guy, but he’s had some years with a lot of air yards, and others with a lot of YAC. His skills appear to be pretty diverse. I also think Brady’s reputation as a system guy is hugely overblown, since he’s had good air yards numbers in years like 2007 and 2004, and way back in 2003, he was throwing downfield a lot more than he does now. The guys on this list have pretty diverse styles. I do believe that avoiding interceptions is a skill, although it doesn’t appear that there are that many guys who really excel at it. Perhaps a high TD% isn’t a reflection of a particular skill but a series of skills, along with the fact that the QB in question ran an efficient offense. It appears that it’s quite difficult to run a high-scoring, pass-heavy offense that also minimizes interceptions.

              • Richie

                I’m surprised Marino made your list. His interceptions tend to hurt his career numbers but I think INT+ shows how much more common interceptions were in his day.

            • Richie

              Hi Tom,

              Do you have handy data to see if there is any correlation between yards per attempt and TD%?

              • Tom

                Richie – I’ve attached a chart showing TD% and Y/A. An era adjustment would probably be appropriate, but this gives us something to talk about. The correlation coefficient is 0.77…so as a QB’s yards per attempt increases, so does his TD’s per attempt…I suppose that makes sense?

                • Richie

                  Yeah, my instinct was that as a QB passes for more yards, his TD passes would also increase. As long as he isn’t racking up incomplete passes to end drives.

                • Adam

                  This makes sense. Without having the exact numbers, I know Y/A correlates strongly with red zone trips, and more red zone chances lead to more TD passes.

            • Adam

              Very interesting, Tom. The graph is a nice touch! This confirms both of our suspicions that TD’s and INT’s aren’t related. TD/INT is roughly the equivalent of something like IQ/40 time – it’s good to be strong in both categories, but they aren’t related and shouldn’t be shoehorned into one metric.

            • Do you have the info in front of you to do this using TD%+ and INT%+? The linked chart shows why era-adjusting is important here.

              https://pbs.twimg.com/media/Cp6G9cXWAAAk8el.jpg:large

              • Tom

                Bryan – I don’t have that data in front of me, I just quickly threw something together for the comment. Yes, the change in INT% is striking, era adjustment is definitely needed.

                • Here’s a quick and dirty chart I made using what I think is similar data to what you used.

                  https://pbs.twimg.com/media/Cp6PcdaWYAACKOu.jpg

                  • Tom

                    Bryan – yeah, this is much more useful than the chart I threw together. So just by eyeballing this, it looks like there is a bit more of a correlation; I can kind of imagine a northeasterly line going through those points. What is the correlation coefficient (I know we’re just throwing stuff around)?

                    • Not in front of the data, but I believe it was a +0.08. So not high but positive. I figure part of the reason is that a good QB is a good QB, and a guy isn’t usually going to reach 100 touchdowns if he is awful in other areas.

        • Adam

          I think TD throwing is essentially a byproduct of other aspects of QB play (in addition to a litany of other variables out of his control). Good QB’s throw more TD’s because they get into the red zone more often, not because they have some special ability to complete passes specifically in the end zone. As you alluded to, there’s nothing a QB can do to really get better or worse at throwing TD’s, other than just becoming a better or worse passer overall. QB’s can consciously choose to trade INT’s for sacks and vice versa, or trade Comp% for Y/C, but what can they trade for more TD passes? Nothing I can think of.

          Interesting that you bring up a hypothetical TD/FUM ratio for running backs. That never crossed my mind before…and maybe that’s the point!

  • Trepur

    I always thought it was meaningless…