Clayton Blog: Competition Committee recommends OT changes

[AbeBeta]

and thats slightly more than 50%...which is barely above half

50/50 is the coin toss odds

David

52% wins vs. 43% losses -- you are evaluating the 52% against 50-50 which isn't correct since teams don't win half the time and lose the other half.

50-50 is not relevant here as there are 3 outcomes. Since 5% were ties it would be 47.5-47.5-5 that would be the relevant comparison values if the coin flip made no difference.

 

[AbeBeta]

Of those 5% or roughly 12 games, currently about 4 will be won by the team winning the coin toss. That number used to be 3 prior to the kick off being moved 5 yards back.

And that's the important point -- that extra game being won by the team winning the toss makes the toss too important. It might sound like a small # of games but as we all know, 1 game can be the difference between playoffs or not.

 

[burmafrd]

Why are you guys ignoring the fact that on 71% of the games the opening drive of the OT does not score? THAT by itself says that in the great majority of OT games both sides get at least one shot. Out of maybe 5 games a year that go to OT, that means maybe only 1 or 2 do not see both teams getting a shot. Why change anyting- the kickoff bit is fluff.

 

[burmafrd]

The K ball thing should have been handled years ago. there is no way that one team should be able to mess with the football in any way. THAT should always be up to the officials.

 

[AbeBeta]

Why are you guys ignoring the fact that on 71% of the games the opening drive of the OT does not score? THAT by itself says that in the great majority of OT games both sides get at least one shot. Out of maybe 5 games a year that go to OT, that means maybe only 1 or 2 do not see both teams getting a shot. Why change anyting- the kickoff bit is fluff.

I'm interested in the outcome of the games in total, not who wins when. Although it is clear that the 29% of the games that end with a single possession are what is creating the imbalance in wins for the team who gets the toss.

Why not look at it this way. The team who wins the toss has a 29% chance of winning on the first possession whereas the team who loses the toss has a 0% chance of winning on the first possession.

 

[Mavs Man]

If you flipped a coin 100 times and came out with an outcome of 55 heads and 45 tails, would you consider the coin unbalanced? Probably not, since there is enough room for error with only 100 tosses. Now, 1,000 or 10,000 tosses - that could be significant.

There are ways to determine the probabilty of 52% win vs 43% loss, so I would like to know how many games were included in that study.

But you'd also have to consider this is much more complicated than a coin toss, and take into account which team was better (record) that season, which team was playing at home, etc. to obtain a figure that attempts to factor out what could be the real cause of the discrepancy.

Correlation doesn't necessarily imply causation.

 

[AbeBeta]

If you flipped a coin 100 times and came out with an outcome of 55 heads and 45 tails, would you consider the coin unbalanced? Probably not, since there is enough room for error with only 100 tosses. Now, 1,000 or 10,000 tosses - that could be significant.

There are ways to determine the probabilty of 52% win vs 43% loss, so I would like to know how many games were included in that study.

But you'd also have to consider this is much more complicated than a coin toss, and take into account which team was better (record) that season, which team was playing at home, etc. to obtain a figure that attempts to factor out what could be the real cause of the discrepancy.

Correlation doesn't necessarily imply causation.

But again. This isn't a sample. It is an entire population of data for the NFL. Folks here are screwing up the distinction between samples and populations -- don't worry, this is a common error made by folks who've had a stat class or two. You know about inferential statistics (a.k.a. signfiicance tests) and their importance -- but you miss the essential point. These statistical are used to make inferences based on samples because populations are usually to large to measure. There is no "inference" to make here. We have data for the entire population of games.

Correlation doesn't imply causation but correlation is a necessary condition for causation to exist. Causal claims like this are always sketchy but you don't throw out the evidence b/c you can't attribute causality. In this situation, we can however rule out a number of alternative explanations that occur in many correlational results -- first, we have established temporal priority -- the coin flip occurs before the overtime period. Second, we have ruled out common causes -- there is nothing that would both cause a team to a) win the flip and b) win the game. So we are a ton closer to causality than your claim suggests.

 

[Mavs Man]

Coin flips are also an entire population, which you sidestepped in your response. And you can measure that.

If it's a small POPULATION it can still be small enough to be within the margin of error, if you know what outcome you should be expecting (close to 50-50 in this case).

 

[Big Dakota]

I got your Statistical significance right here.

 

[dbair1967]

I'm interested in the outcome of the games in total, not who wins when. Although it is clear that the 29% of the games that end with a single possession are what is creating the imbalance in wins for the team who gets the toss.

Why not look at it this way. The team who wins the toss has a 29% chance of winning on the first possession whereas the team who loses the toss has a 0% chance of winning on the first possession.

that is not true...just one example off the top of my head is Seattle won the toss in a playoff OT game vs GB three yrs ago, they lost the game on their opening posession when Hasselback threw a pick that was returned for a TD

again, when barely half the time the teams winning the toss wins the game at some point, and when almsot 3/4 of the time both teams are getting chances on offense, you are crying over spilt milk

David

 

[AbeBeta]

Coin flips are also an entire population, which you sidestepped in your response. And you can measure that.

If it's a small POPULATION it can still be small enough to be within the margin of error, if you know what outcome you should be expecting (close to 50-50 in this case).

There is no margin of error in measuring a population. The MOE is the standard deviation of the population divided by the square root of the size of the sample then multiplied by some test statistic parameter (e.g., chi-square for 95&#37. No sample = no MOE.

The coin flip example is not an entire population. The population of coin flips in infinite. The population of games that went into OT is not. There were something like 350 and all were measured and reported.

 

[AbeBeta]

that is not true...just one example off the top of my head is Seattle won the toss in a playoff OT game vs GB three yrs ago, they lost the game on their opening posession when Hasselback threw a pick that was returned for a TD

An INT is a change of possession.

 

[dbair1967]

An INT is a change of possession.

you said the team that lost the toss had a ZERO % chance of winning on the first posession, and clearly that isnt the case

feel free to keep spinning

David

 

[Mavs Man]

There is no margin of error in measuring a population. The MOE is the standard deviation of the population divided by the square root of the size of the sample. No sample = no MOE.

The coin flip example is not an entire population. The population of coin flips in infinite. The population of games that went into OT is not. There were something like 350 and all were measured and reported.

So, let's say during the first year of overtime games there were three OTs and the team that won the toss won the game all three times.

Would you make the case that in that hypothetical league definitively the winner has a 100% chance of winning OT games if they win the coin toss, over the lifetime of the league?

 

[AbeBeta]

you said the team that lost the toss had a ZERO % chance of winning on the first posession, and clearly that isnt the case

feel free to keep spinning

David

It isn't the first possession if there is an INT. It is the 2nd possession. That isn't spin. That is the rules.

 

[AbeBeta]

So, let's say during the first year of overtime games there were three OTs and the team that won the toss won the game all three times.

Would you make the case that in that hypothetical league definitively the winner has a 100% chance of winning OT games if they win the coin toss, over the lifetime of the league?

No, I wouldn't. Because that was a single year. You do have a value for the population of games played in the league but only an idiot would draw a conclusion based on a population that was so small because it is likely to be unstable.

But we are talking about a population covering 30 years not 1.

 

[Mavs Man]

No, I wouldn't. Because that was a single year. You do have a value for the population of games played in the league but only an idiot would draw a conclusion based on a population that was so small because it is likely to be unstable.

But we are talking about a population covering 350 games not 3.

So you're saying 3 games is too small but 350 is big enough - where do you draw the line? Where does a population cross the threshold?

 

[AbeBeta]

So you're saying 3 games is too small but 350 is big enough - where do you draw the line? Where does a population cross the threshold?

You draw the line at context. Here is where critical thinking skills come in... E.g., is this result likely to change much over the next year? For 3 games, it could shift considerably. For data based on 30 years, it would be nearly impossible for the observed differences to fade based on a reasonble # of OT games.

 

[CCBoy]

you bring so much really good information to the board, that I'd just like to stop, and say thanks.....

 

[theogt]

same thing can happen in regulation games Theo, is that unfair?

David

That's a completely differenct scenario. During regulation, the opposing team always has the chance to get the ball back and win. In overtime, all a team has to do is get one single lucky break and poof the game is over. The opposing team isn't given an opportunity. The only comparison in regulation is when a team gets the ball with little time left and drives down to kick a game winning FG, but that's not an adequate comparison.

There is no margin of error in measuring a population. The MOE is the standard deviation of the population divided by the square root of the size of the sample then multiplied by some test statistic parameter (e.g., chi-square for 95&#37. No sample = no MOE.

The coin flip example is not an entire population. The population of coin flips in infinite. The population of games that went into OT is not. There were something like 350 and all were measured and reported.

So, 100 coin flips isn't entire population, but 300 football games is entire population? There's no difference. There are potentially infinite coin flips, and there are potentially infinite football games in OT. You can't use 55 out of 100 flips being heads to claim that the coin is rigged in heads favor, just like you can't use 52% wins out of 300 football games to claim that the coin flip decides the game. (Note: Obviously, I think there is some level of influence in the coin flip and I think this should be eliminated, but that's not what we're talking about at this point. My point is that I don't know whether the sameple size is large enough to know how correct the 52% number is.)

I can take a 10 year old company's stock and measure it's performance against the market for 10 years (which would be "an entire population" for the company's stock), and if the cumulative abnormal returns are small enough, then the number will not be statistically significant. I can similarly take 30 years of football games and measure the results against coin flips for 30 years (which is what we're doing here). If the abnormal results are small enough, then the difference will not be statistically significant.

To illustrate the problem, over the next 5 year span, it is entirely possible for the team that loses the coin flip to win every single OT game. That would likely change the results sigificantly. Let's assume that it would change the results to the coin-flip loser winning 55% vs. the coin-flip winner winning the game 40%. At that point, you couldn't make the argument that losing the coin helps win the game, because obviously there is some level of chance in the results. Statistical significance states whether the abnormal results (compared to the coin-flip having absolutely zero impact, i.e., teams spliting 47.5-47.5-5) of the sample size (which is all this is) are meaningful -- in other words, whether they're true.