1. <YouRang>
1. <just a kid>
3. <iron maiden>

I was thinking again of <EXPERIENCE> vs <YOUTH> which I have done before.

EXPERIENCE, looking at age but now also somewhat at amount of top tournament participation order:

<EXPERIENCE>
1. Ivanchuk
1. Anand
3. Aronian
4. Grischuk
5. Radjabov
6. Carlsen
7. Dominguez
8. Wang Yue

<ALPHABET> by cromat!:)

1. Anand
2. Aronian
3. Carlsen
4. Dominguez
5. Grischuk
6. Ivanchuk
7. Radjabov
8. Wang Yue

1. Anand
2. Radjabov
3. Carlsen
4. Aronian
5. Ivanchuk
6. Wang Yue
7. Grischuk
8. Dominguez

In particular, there is a new type of Final Standings ranker, which is similar to the "Betting Ranker" report that I use in the game predictions contest. It's called the <Pair Betting> ranker.

The idea behind it as as follows:

Suppose you submit your predictions for the final standings, such as:

1. Leko
2. Adams
3. Polgar
3. Dreev
5. Kamsky
6. Bacrot

This can be viewed as a series of bets on every *pair* of players. For instance for the pair "Adams-Kamsky", you have predicted that Adams will finish above Kamsky. For Dreev-Leko, you have Dreev finishing below Leko. For "Dreev-Polgar", you have predicted that they will finish tied.

If there are 6 players (like in MTel), there will be a total of 15 such pairings. If 8 players (like Linares), there will be 28 pairings. For 14 players (like Corus) there will be 91 such pairings!

In the "Pair Betting" ranker, each contestant who submits a final standings prediction is considered to have wagered a $1 bet on each pairing. So, if there are 20 contestants, then there will be $20 wagered on each pairing. The $20 wagered on a particular pairing will ultimately be divided up between those who predicted that pairing correctly.

If a lot of people (say, 15 of 20) got it right, then the payoff for that pairing will be small ($20/15 = $1.33). But if only one person got it right, then he or she wins the whole $20 (like a jackpot). Of course, if you predicted wrong, you lose $1 for that pairing.

Note that predicting ties is a little risky since the players must finish with the exact same score for you to win. Then again, if you're right, you have a good chance to win big.

It's also possible that there will be a pairing result that nobody predicted. In that case, all the money wagered on that pairing will be refunded.

The "Pair Betting" ranker can be very volatile. Conceivably, you can be in 1st place after 10 rounds, and in last place after 11 rounds!

~~~~~~~~~~~~~~~~
Other improvements:

- I think I've fixed the problem with people getting negative scores. I had a bug in the maximum score calculation.

- It now uses your original method of evaluating the rank of ties. For example, if two people are tied for 4th, then for the sake of calculations, they both have a rank of 4.5 (rather than 4).

- You can now keep the standings for all completed rounds in your YRC file, and select which round you want to use in the report. (I'll send you an example via email).

1. Carlsen
2. Anand
3-5. Aronian
3-5. Ivanchuk
3-5. Radjabov
6. Wang Yue
7-8. Dominguez
7-8. Grischuk

This is a pretty important change and means a whole new predicting attitude will be required, so I'd like to know clearly -

1) Does this apply to both predicted ranks and actual ranks?

<1) Does this apply to both predicted ranks and actual ranks?>

Yes, it applies to both.

<2) Does this mean we'll be seeing strange fractions in our scores soon?>

No. Since the "half ranks" always come in pairs, they sum to whole numbers when the differences are added together.

Even a tiny difference could be crucial. But the difference in optimal strategy will be more than tiny. And this is why.

Not squared.>