• ℹ️ Heads up...

    This is a popular topic that is fast moving Guest - before posting, please ensure that you check out the first post in the topic for a quick reminder of guidelines, and importantly a summary of the known facts and information so far. Thanks.

ElloCoaster Poll is live again, this time including STEEL

I believe the Hawker algorithm used purely head to head - my implementation of it certainly did and I was able to recreate his results pretty closely (barring a few wierd edge cases like people listing the same ride more than once).

I'm a fan of the pure head to head as total win percentages are affected by the rides you're comparing them with - lots of people voting on just UK rides could see Nemesis with a disproportionately small number of losses for example.
 
Hawker used total win percentage, then used head-to-head ONLY to break a tie. His calculations only went two decimal places, which would've resulted in a few ties in our results (although he presented those as "0.9934" where we take the same result and show "99.34"). But other than ties, the pure head-to-head didn't matter, as this explanation from his site shows (highlight is mine):
"Each coaster is compred one at a time to every other coaster to see whether
more people who have ridden both of them preferred one or the other.
A coaster is given a "Win" for each coaster that more mutual riders ranked
behind it, given a "Loss" for each coaster that more mutual riders ranked ahead
of it, and given a "Tie" for each coaster that the same number of mutual riders
ranked above it and below it.

Coasters are ranked by their overall winning percentage (where ties count as
half of a win and half of a loss). In the event that two coasters end up with
identical winning percentages, the tie is broken (if possible) by determining
which of the two won the mutual rider comparison between those two coasters."
 
Looking at our Top 25, the only tie (using Hawker's 2-decimal rounding) would be Dauling Dragon/ Mystic Timbers - and the H2H would've kept them in the same order.
On the steel side, only Iron Rattler (13) vs Superman (14) would've generated a tie and those would also have kept their current rankings.
Going past the Top 25, the next tie would be Wonder Woman (30) and Starry Sky Ripper (31) - which would've actually changed places... BUT they had only one common rider, so there's that.
 
Ouch.

Taron and Helix are great but my god. That's some overrated rubbish right there.

I've never personally found myself at odds with one of the Hawker polls to this extent before.



Sent from my SM-G973F using Tapatalk
 
Hawker used total win percentage, then used head-to-head ONLY to break a tie.
This is not true, as Mitch explained in your quote. The "win percentage" is not an absolute %age but is calculated from the number of wins/losses a ride had using the mutual riders method. Just look at Expedition GeForce in the
2012 poll - it had a score of 100% as it came out on top compared with every other coaster, not top of every ballot. Bizarro lost to EGF (16 to 21) and beat 362 other coasters head-to-head: 362/363 = .9972.

Both methods have their merit, but this poll is not a clone of the Hawker poll!
 
Both methods have their merit, but this poll is not a clone of the Hawker poll!

No, this is exactly the same. Have a look at Lightning Rod. It shows 182 PAIR WINS, not total wins on each ballot. That means that it won (had more wins than losses) against 182 other coasters. It lost to Fjord Flying Dragon and Python in Bamboo Forest (each with only one mutual rider) and tied with Jungle Trailblazer. You can scroll down the "Wood Individual Pairs" sheet to see it up close. For each pair, only the mutual riders for that pair get counted. Just like Hawker's method.
 
Wait, what?

So where do the absolute winning percentages you're talking about come in? Were they dropped after the 2017 poll?
 
Wait, what?

So where do the absolute winning percentages you're talking about come in? Were they dropped after the 2017 poll?
YES. The 2017 poll used total win percentages, but beginning with 2018 we switched to a direct pairs percentage Hawker clone in order to keep people with very long lists from having undue influence on the results.
 
Surely, people with longer lists *should* have more of a weighting on the list?

I mean... if you are European and have only been on coasters in Europe - of course Taron, Helix, Shambhala and Co are going to be in your top 10?

The most prized ballots are from those people who have been on all the best coasters in the USA, Europe, Asia etc.

Anyway, at the end of the day I guess it is a bit of fun. I am just very, very surprised that you have the likes of Helix and Co. above boundary pushing coasters.
 
Surely, people with longer lists *should* have more of a weighting on the list?
This was my thinking as well when we did the first poll in 2017.
HOWEVER, this point is only really valid when a poll asks (basically) "what are your favourite coasters" - this is how the GTAs work, as well as lots of other polls out there.
If that's how you approach it, then ABSOLUTELY the people with the longest lists should have a weightier impact on the results. After all, someone who puts [coaster] at his/her #1 spot is impressive if they've ridden 100 coasters. It's 10X more impressive if they've ridden 1000 coasters.
But...
This poll doesn't care if you rank [coaster] as your #1. It only matters whether or not you rank [coaster] above or below [another coaster that you've ridden]. That's it.
So in this case, it isn't like asking someone who's only been on two coasters which is their #1, it merely asks which of those two coasters is better. That person's opinion is therefore absolutely as valid as a more seasoned rider's opinion, because it's only asking about those two coasters.
So asking someone "what's your top ten" makes a longer credit list a distinct advantage... but asking someone "which of these two coasters is better" only matters that they rode both of them.
I hope that made sense. I feel like I'm rambling a bit.
 
Fair enough.

To be honest, I wouldn't be picking apart the system if I agreed with the rankings! :p

But, as it is, it's the Hawker poll I have least resonated with so far. Apart from the #1. And, as of this year, I have been on more coasters than ever. The more I seem to ride, the less I seem to agree with the list.

:D
 
I hear ya. My personal top ten wood was well-represented in the top 25 (although my #1 barely made the top 50) but holy crap, my top ten steel is pretty much nonexistent in the top 25. Only two of my top ten made it in. Several of my top ten steel didn't even make the top 100!
 
I am just really *really* shocked that I305 is so low. It's around the 5-10 mark for me, yet here...

Wow!

Also really gutted about Maverick's position. I think it's the perfect coaster. I was so disappointed that Taron just didn't have the bite or grunt of Maverick's wacky layout - even though Taron is such a beautiful machine. Those boring curves are unforgivable on top 10 coasters. There should *not* be any weak spot on a top 10 coaster!
 
I'm just baffled the likes of Skyrush, Maverick and i305 all fell short compared to Hyperion. I honestly believe it to have been the most over-hyped coaster of last decade.
 
I know a lot of people who have been on Hyperion. Nobody says it's bad. Everyone says, you know 'decent' - 'solid' but not amazing.

That is far from the case with the above named Intamin coasters. People get off those Intamin's like OMG*

*you know, I think restraints are playing a bigger role in how people judge coasters these days
 
I just thought of a good analogy to explain why those with really long credits lists shouldn't have their votes counted more than those with short lists...
Let's say that you've only seen two Hitchcock films: Vertigo and Psycho.
How other polls work: "What's the best Hitchcock film?"
You can see how this makes your vote less important than the vote of someone who's seen nearly all of them.
How the ElloCoaster poll works: "Which film did you like more, Vertigo or Psycho?"
Now you can understand why your answer to that should have just as much weight as the person who's seen more of the films, because we aren't concerned with any of those movies other than Vertigo and Psycho. And as long as you've seen both of those, your opinion should count.
 
Top