A new point scale to improve the championship

[zac510]

The reason you want to change the scoring system is because you feel some driver or team you like has not been rewarded appropriately, which is not a criticism of you GM7, it is just an observation of human bias. But we can't go changing the points every year to boost up the team or driver we think has done well.

The teams receive their money on championship order, not points, so the team will be rewarded financially for good performances regardless of getting points down to 20th place or not. As studied in behavioural economics the point system should have scarcity value, so when Stroll or Alonso gets a point that's a talking point.

Shrieker, that rule might have been removed when the teams were contracted to attend every race. I last remember it in place some time in the 80s.

[rscsr]

The way I see it a points system should make a pass worthwhile, no matter which position you are running in. Therefore I calculated the cut off success rate for overtakes for some different points systems.

The way I calculated that value was to calculate the expected gains, for variable percentage of success and rearranged a function for the required percentage of succesful attempts.
rearranged to with e=0, where g are the gains and l are the losses for a succesful or unsuccesful pass.
I was calculating these values for a situation where a pass might result in an accident and no finish (dashed line) and where the overtaking car might lose 1 position (solid line). I don't include values for passes on direct competition because it results in basically the same curves with an offset, so fundamentally the same curve shapes.
The curve D uses an exponential point distribution, a succesful pass gives you 40% more points in this case and I included 2 old point systems as well.
A lower value for the curves means that you need a lower success rate for overtakes. That means a lower value should lead to riskier overtaking attempts.

A - current system - 25, 18, 15, 12, 10, 8, 6, 5, 4, 3, 2, 1, 0
B - op's proposal - 25, 20, 18, 16, 15, 14, 13, 12, 11, ...
C - linear - 19, 18, 17, 16, 15, 14, 13, 12, 11, ...
D - exp (1.4) - 600, 426, 304, 217, 155, 111, 79, 56, 40, 28, 20, 14, 10, 7, 5, 3, 2, 1
E - "old" system - 9, 6, 4, 3, 2, 1, 0
F - "old" system - 10, 8, 6, 5, 4, 3, 2, 1, 0



Since basically every proposed point system uses a linear scale for the lower positions you end up with the same cut off percentage for the backmarkers when you might lose 1 position.
I find it quite interesting that the old system is quite risk friendly with quite low required succes rates for overtaking attempts.
And while the exponential system might make for an ugly or difficult to remember point system, it seems to be pretty much position independent. So it would make sense to fight for positions no matter where you are running.

D yields constant values because everything but the factor cancel out. ( for accidents and for 1 position lost, f=1.4 for these values)

[ENGINE TUNER]

I believe that the driver with the most wins in a season should be champion NO MATTER WHAT.

Then in 2009 Button would of been crowned world champion in Istanbul...

I have absolutely no problem with that, but you are wrong, Button won 6 of the first 7 races but Malaysia only counts as half since the race was cut short, so he won 5.5 of the first 7 races, Vettel won 2 of the first 8. There were 10 races after Istanbul(#7) so anyone could have won the championship If they won 6 of those 10 races(if Button didn't win again). Vettel only had to win 4 of those 10, unless someone else won 7 of those 10.

Any driver that wins half plus 1 of all the races DESERVES the championship regardless of how much you or anyone wants to see the championship go to the last race.

Also you would generate a lot of broken carbon fibre as 2nd place would try a suicide manoeuvre since allowing a win is more painful than taking them both out.

Provided 2nd could even catch up with the leader. Any driver doing such a thing should face a season or lifetime ban, we could call it the schumacher rule.

[OneAlex]

If I were to design a points system I would immediately throw out the current system because of 3 major factors.

1) It does not incentivize winning enough.

If I remember right that was why the current points system (with bigger 1-2-3 differences) was brought in in the first place, because too many drivers were happy to finish second rather than risk trying to overtake for the win for not many more points.

[henry]

The way I see it a points system should make a pass worthwhile, no matter which position you are running in. Therefore I calculated the cut off success rate for overtakes for some different points systems.

The way I calculated that value was to calculate the expected gains, for variable percentage of success and rearranged a function for the required percentage of succesful attempts.
rearranged to with e=0, where g are the gains and l are the losses for a succesful or unsuccesful pass.
I was calculating these values for a situation where a pass might result in an accident and no finish (dashed line) and where the overtaking car might lose 1 position (solid line). I don't include values for passes on direct competition because it results in basically the same curves with an offset, so fundamentally the same curve shapes.
The curve D uses an exponential point distribution, a succesful pass gives you 40% more points in this case and I included 2 old point systems as well.
A lower value for the curves means that you need a lower success rate for overtakes. That means a lower value should lead to riskier overtaking attempts.

A - current system - 25, 18, 15, 12, 10, 8, 6, 5, 4, 3, 2, 1, 0
B - op's proposal - 25, 20, 18, 16, 15, 14, 13, 12, 11, ...
C - linear - 19, 18, 17, 16, 15, 14, 13, 12, 11, ...
D - exp (1.4) - 600, 426, 304, 217, 155, 111, 79, 56, 40, 28, 20, 14, 10, 7, 5, 3, 2, 1
E - "old" system - 9, 6, 4, 3, 2, 1, 0
F - "old" system - 10, 8, 6, 5, 4, 3, 2, 1, 0

http://i.imgur.com/5KMuIdZ.png

Since basically every proposed point system uses a linear scale for the lower positions you end up with the same cut off percentage for the backmarkers when you might lose 1 position.
I find it quite interesting that the old system is quite risk friendly with quite low required succes rates for overtaking attempts.
And while the exponential system might make for an ugly or difficult to remember point system, it seems to be pretty much position independent. So it would make sense to fight for positions no matter where you are running.

D yields constant values because everything but the factor cancel out. ( for accidents and for 1 position lost, f=1.4 for these values)

Thanks for a very interesting post. A points system that offers rewards for gaining places all the way through the field would be a welcome improvement. However, the risk/reward would I think still not be sufficient, since the risk is not losing one place but losing a finish and hence all points.

Last year, when there was a debate about whether Rosberg was a worthy world champion, I suggested a measure of merit might be to see which driver was number 1 over a series of virtual, 19 race, championships. Using the current scoring system Rosberg and Hamilton contested 60 such series and Hamilton won 37 of them, 62%. So that he won 66% of the world championships seemed, to me, fair. If we apply your scoring system he would have one 1 more virtual championship and still two WDC.

It seems that the occasions when a driver doesn't finish are much more important than gaining places, or being the winner in this case.

But the OP was more interested in midfield so I looked at Hamilton vs Button, since they didn't race together when the car was dominant. The results surprised me. They raced 40 virtual seasons. Under the current points system Hamilton won 21. Under your system, with substantially greater rewards for every place gained he won 17. Slow and steady Jenson Button did better than an acknowledged risk taker.

To really reward risk takers i think it would be necessary to use one of the best N results techniques. This would also better cater for no fault DNFs.

My personal analysis is flawed in that I really ought to discard all races where one of the protagonists had a no fault DNF. But that's a lot of work and I'm not that concerned.

EDIT: fixed an auto-miscorrect

[Phil]

The problem I see with these topics is that any point scoring could be better when applied mid season to improve 'something' in hindsight.

Every point scoring system has flaws. If you have one that rewards wins, it will fire back if you have a single team dominating. If you have equal teams fighting on the same level more consistent point system will reduce the reward in fighting for position. More points reward consistency and reliability while hurting technical issues and DNFs. Potentially, as DNFs become more costly, it discourages risk.

So you see, there is no perfect scoring system. Anyone can come up with something better to reward the drivers they like.

For example: I don't like it that #TeamNR beat #TeamLH last year to the WDC. Even if I could come up for a more "just" point scoring system for that year (that would alter the outcome to my perceived ideal of who deserved it most), that same point system would hardly be as just (by the same ideals) for any other year. It just doesn't work.

[henry]

The problem I see with these topics is that any point scoring could be better when applied mid season to improve 'something' in hindsight.


So you see, there is no perfect scoring system. Anyone can come up with something better to reward the drivers they like.

Agreed there is no "perfect" system but I doubt very much that your second point is correct. As I showed in the post immediately before yours a point system that appears to reward risk taking gives an unexpected result in the one case I examined.

But the OP wasn't concerned about the pointy end of the championship. The concern was that the points in the midfield don't fairly represent the skill and endeavour of the participants.

A system with a distribution further down the field, such as proposed by rscsr, with a few discards would, I think, encourage risk taking, if thats what we want to encourage.

Sadly Canada showed that no amount of encouragement is likely to regularly get people close enough to be in a position to try a risky overtake.

[ENGINE TUNER]

The problem I see with these topics is that any point scoring could be better when applied mid season to improve 'something' in hindsight.

Every point scoring system has flaws. If you have one that rewards wins, it will fire back if you have a single team dominating. If you have equal teams fighting on the same level more consistent point system will reduce the reward in fighting for position. More points reward consistency and reliability while hurting technical issues and DNFs. Potentially, as DNFs become more costly, it discourages risk.

So you see, there is no perfect scoring system. Anyone can come up with something better to reward the drivers they like.

For example: I don't like it that #TeamNR beat #TeamLH last year to the WDC. Even if I could come up for a more "just" point scoring system for that year (that would alter the outcome to my perceived ideal of who deserved it most), that same point system would hardly be as just (by the same ideals) for any other year. It just doesn't work.

There is a perfect system if it is designed as such, the current system is based on the original system than was mostly arbitrary. Throw the current system out and rethink the entire thing with the proper emphasis and incentivize winning and attacking rather than "points racing", minimize the penalty of mechanical failures(to a point) by allowing the dropping of races and points through the field to make sure all can be graded by the points system.

I always thought double was a good way to incentivize attacking, every position at least double the points. Again, starting with 26th position with the hope that eventually there will be 13 viable teams and to make the system work going forward regardless how many teams there are.

1) 1 billion
2) 62914560
3) 12582912
4) 4194304
5) 2097152
6) 1048576
7) 524288
8 ) 262144
9) 131072
10) 65536
11) 32768
12) 16384
13) 8192
14) 4096
15) 2048
16) 1024
17) 512
18) 256
19) 128
20) 64
21) 32
22) 16
23) 8
24) 4
25) 2
26) 1

Every position up is *2, from 4th to 3rd is *3, 3rd to 2nd is *5, 1st is 1 billion to make sure the driver with the most wins is always champion. Only the best 18 of your 21 races count for points.

Yeah it looks crazy, but it works.

[roon]

What if an element of qualifying or time attack affected the maximum points available per position? Based upon a driver's improvement upon their own laptimes during the race.

Such that the driver is not only racing the other participants in the race, but also themselves. This might discourage tactics which involve the driver slowing down, if he is racing his past self along with the rest of the field, in pursuit of the maximum points available for a given position.

Said another way: the degree to which a driver improves upon their own laptimes unlocks more points per position.

[Phil]

I always thought double was a good way to incentivize attacking, every position at least double the points.

Every position up is *2, from 4th to 3rd
Yeah it looks crazy, but it works.

So if the slowest team lucks into a ridiculously high finishing position at one race, they'd be virtually guaranteed finishing above their "skill level"?

Not quite sure that is better than any other point system, which was sort of the point i was making.

And instead of raising the incentive to overtake by rewarding more points and thus raising the cost of a potential DNF the higher the position you are battling for, why not simply make overtaking easier? If overtaking stays as challenging as it is, you are just promoting more dangerous driving.

[notsofast]

F1 is dependent on spectators. The points system should not only reward the driver, but the audience as well. If we're going to discuss changes to the points system, I would like to see a system that rewards drivers for doing the things that the audience wants to see. Let's say, Hamilton is in the lead, and Bottas is running in third, 50 seconds behind. In between them, there's Vettel, 10 seconds behind, with 15 laps to go. With the current points system, Vettel can just phone it in, and collect his second place points. But that's not what I want to see. I want to see Vettel chasing Hamilton. There ought to be some kind of reward for keeping the gap in front small and the gap behind large. A driver should be rewarded for keeping up the pressure to close the gap in front, and also for keeping up the pressure to stay far ahead of the driver behind.

[roon]

That's sort of what I'm getting at, as well. Each position could have a points range, rather than one value. Max points per position achieved via the degree to which a driver reduces their laptimes. A "slow second place" would reward fewer points than a "fast second place."

In your example, when the race is all but decided, drivers would still be incentivised to push to unlock more points available per position.

[zac510]

Why not give the winner infinite points? Then if they win two races, they get infinite+1 points.

[rscsr]

...

...
Last year, when there was a debate about whether Rosberg was a worthy world champion, I suggested a measure of merit might be to see which driver was number 1 over a series of virtual, 19 race, championships. Using the current scoring system Rosberg and Hamilton contested 60 such series and Hamilton won 37 of them, 62%. So that he won 66% of the world championships seemed, to me, fair. If we apply your scoring system he would have one 1 more virtual championship and still two WDC.

It seems that the occasions when a driver doesn't finish are much more important than gaining places, or being the winner in this case.

But the OP was more interested in midfield so I looked at Hamilton vs Button, since they didn't race together when the car was dominant. The results surprised me. They raced 40 virtual seasons. Under the current points system Hamilton won 21. Under your system, with substantially greater rewards for every place gained he won 17. Slow and steady Jenson Button did better than an acknowledged risk taker.
...

Do you mind sharing how you do those virtual seasons?

[henry]

...

...
Last year, when there was a debate about whether Rosberg was a worthy world champion, I suggested a measure of merit might be to see which driver was number 1 over a series of virtual, 19 race, championships. Using the current scoring system Rosberg and Hamilton contested 60 such series and Hamilton won 37 of them, 62%. So that he won 66% of the world championships seemed, to me, fair. If we apply your scoring system he would have one 1 more virtual championship and still two WDC.

It seems that the occasions when a driver doesn't finish are much more important than gaining places, or being the winner in this case.

But the OP was more interested in midfield so I looked at Hamilton vs Button, since they didn't race together when the car was dominant. The results surprised me. They raced 40 virtual seasons. Under the current points system Hamilton won 21. Under your system, with substantially greater rewards for every place gained he won 17. Slow and steady Jenson Button did better than an acknowledged risk taker.
...

Do you mind sharing how you do those virtual seasons?

Not at all.

I listed all the placings of the seasons they contested. I allocated points to each place. I used a number of points systems. The current one, a simple place based one, 22 for first, 1 for last, and most recently your, logical, approach.

Starting at race 19 I summed the points for those 19 races, essentially a calendar season. Then for race 20 I summed 2 to 20 making a virtual season. For each virtual season I compared the points and got a "winner".

To improve my data I probably ought to exclude all races where one of those under comparison had a no fault DNF. However this is tricky and somewhat subjective and quite a bit of work. Since no one was interested in my approach I didn't bother.

I based the idea on other long-season sports such as tennis or golf, which typically have world number one based on the previous 52 weeks and a special place for the leader at the end of the calendar year.

I think it could be a useful tool for evaluating drivers over a career. Since I already had the Hamilton data I completed it. Up to the end of last year he had completed 170 nineteen race virtual seasons and "won" 123. Which is pretty impressive. I contemplated doing Alonso and Vettel, two other contenders for pretty impressive, but thought it too much work.

I'd be interested in your views on this.