Probability of current seeds making the finals

[Howard Somerset]

This isn't a spoiler thread in the same way as others in this section, but I've put it here as it has info about recent games. Feel free to move it, or even delete it, if you want.

There's been a fair bit of speculation recently about whether various current seeds will make the finals, and in particular whether Richard will end up as #8 seed and thus probably face David in the QF.

I've tried to look at the probability of each of the seeds making it though to the finals.

David and Michael are both certainties, as there are insufficient games left for the required number of octochamps to pass them

Tim is virtually there. I believe there are 47 preliminary games left, which is the minimum possible for 6 further octochamps to be found. That would require Carole, who won earlier today, to go on to be an octochamp, and then be immediately followed by 5 others. Moreover, all of them would have to have a points score greater than Tim's. So I think that betting on winning the jackpot in the lottery is far safer than betting on Tim failing to make it.

With the rest of the current seeds it's much more interesting. The way I've looked at it is, for each or the seeds, to see how many people since the start of the 15 round games have managed to get a higher position. For example, in Jason's case, 126 people (in 1378 games) have got a better score. With 48 games to go include the game played earlier today, 3 people or more getting a better score will put Jason out. Since 3/48 is slightly less than 126/1378, that would suggest that Jason has a less than even chance of qualifying.

I've tried a statistical model to quantify the probability. Although not perfect, I believe it's not far from being right. And applying this I find the probability of each of the seeds making the finals as:
Barry 85.8%
Richard 68.4%
Jason 17.3%
Matt 3.4%
Tony 0.5%
(that's assuming that Matt is ranked higher than Tony)

In Richard's case, his actual position is very relevant, particularly to David, and applying the same reasoning gives his probability of achieving specific seeding positions as:
5th seed 5.3%
6th seed 16.2%
7th seed 23.9%
8th seed 23.0%

Of course the people concerned will know very soon; they may even already have knowledge about some of the games yet to be shown. But without that knowledge we can only speculate.

[Matt Coates]

3.4 has always been my lucky number

[Howard Somerset]

3.4 has always been my lucky number

Just to encourage you, Matt, your probability rises quite rapidly as each champion fails to pass you. Calculated immediately after your defeat, your probability of making the finals would have been 2.6%. And if Carole loses today it will rise to 3.7%.

Of course, once you get the call from Damien, or (let's not think of it) fail to get the call, you'll know for certain. But the rest of us wont.

[Charlie Reams]

With the rest of the current seeds it's much more interesting. The way I've looked at it is, for each or the seeds, to see how many people since the start of the 15 round games have managed to get a higher position. For example, in Jason's case, 126 people (in 1378 games) have got a better score. With 48 games to go include the game played earlier today, 3 people or more getting a better score will put Jason out. Since 3/48 is slightly less than 126/1378, that would suggest that Jason has a less than even chance of qualifying.

If I've understood your method correctly then (further good news for Matt), your probabilities are slight underestimates. I think you implicitly assume that the win count of each contestant is independent. That's not quite true because (for example) an octochamp reduces the total number of distinct winners in a series, and hence slightly decreases the chance of someone else racking up a decent number of wins. So, to give a concrete example, (probability of having five 4-time winners) is less than 5x(probability of one 4-time winner.) It's probably possible to compensate for this somehow but currently I'm too lazy to work it out.

[Howard Somerset]

You're correct Charlie. The probabilities are slight underestimates. I said that the model is not perfect, and was aware of the underestimates. A more accurate model would be much more complex and would've taken more than a couple of hours. Anyway better to predict slightly under than slightly over, I believe.

[Charlie Reams]

You're correct Charlie. The probabilities are slight underestimates. I said that the model is not perfect, and was aware of the underestimates. A more accurate model would be much more complex and would've taken more than a couple of hours. Anyway better to predict slightly under than slightly over, I believe.

Yep, nothing wrong with some simplifying assumptions. Just thought I'd give Matt a bit of a lifeline!

[JasonCullen]

Well I'm still holding out for going back for the finals. Although statistics don't lie, I am still optimistic especially the way this series is turning out I would even settle for 8th seed as long as I get a second chance even if that means taking on the almighty Mr. O'Donnell in my quarter final

[Joseph Bolas]

Well I'm still holding out for going back for the finals. Although statistics don't lie, I am still optimistic especially the way this series is turning out I would even settle for 8th seed as long as I get a second chance even if that means taking on the almighty Mr. O'Donnell in my quarter final

I think you are going to be back for the finals .

I know Ben who won yesterday might be an octochamp (the lad is good IMO) but you will still be in the seed table if this happens .

[Benji Hanks]

I know Ben who won yesterday might be an octochamp (the lad is good IMO) but you will still be in the seed table if this happens .

Well I'll certainly take that as a compliment!