Criteria of an Expert

[Sportsguy_USA]

Been thinking of building a sports prediction game and trying to wrap my head around what variables or formula would you use to determine an expert. For example, if you had 1000 people participated, and wanted to create a leaderboard ranking them in chronological order of their 'expertise' of their sport prediction skills/accuracy, what criteria would you use?

I've narrowed the variables too.

1. win percentage
2. earnings/winnings
3. number of games predicted
4. time

What formula could I use from these variables to provide the most accurate list of experts?

I had first thought 'win percentage", but that would be skewed based on the number of games a person predicts (ex. I predicted 2 games and get 2 right = 100% vs. someone who predicts 100 games and gets 70 right = 70%). So then time and number of games should become a factor. And if these are two integral factors, what would be an appropriate value to use for these?

Throwing this on think tank wall to come up with the most ideal scenario.

[brettd]

The best two methods of predicting handicapper success:

1. Create inbuilt Z scores of how they each go against the spread.

2. Percentage of times their picks beat the closing line when released.

[Sportsguy_USA]

Thanks Brett. Bit new on this, but what are 'z-scores'?

But in the end, couldn't you then just base this on their win percentage over time and by sport, with the assuming their pick is set at time the book is closed?

[Duff85]

For what your describing $ won is going to be best suited to your needs.

[That Foreign Guy]

Time is not a factor.

Going 5-0 picking one game a year is no better or worse than going 5-0 picking a game a day for five days or any other combination of non-simultaneous games (going 5-0 on simultaneous games makes slightly less $ due to simultaneous Kelly so could arguably be considered worse).

Google Z score and either use that or money won.

Also learn what chronological means.

[nohookplayer]

You raise such good point, expecially considering how slim the difference is between a winner and a loser. I am a big fan of Al McMordie, he honestly tracks his results and has been well tested over time.

[cc440unn]

earning and winning have to be the first criteria for rate an handiccapper

a guys can be 3000-3500 in his prediciton but +100 unit i

and an other guys can be 3000-25000 and -100 unit

A guys who frequently take underdogs will loss more often than the guys who is taking the favorite. But he will win more each time. This is why the Win % canT be more important than the earning,

The number predicted will only affect the win % (60% on 5k prediction will be better than 65% on 1k )

The time doesnt really count IMO

[cc440unn]

Time is not a factor.

Going 5-0 picking one game a year is no better or worse than going 5-0 picking a game a day for five days or any other combination of non-simultaneous games (going 5-0 on simultaneous games makes slightly less $ due to simultaneous Kelly so could arguably be considered worse).

Google Z score and either use that or money won.

Also learn what chronological means.

False IMO

On a year you can choose the best game where you have the max chance of winning and you cant wait to find the best game for place your bet.

On one day you will have to make more hard decision.

For example if i had to bet only 5 game in a year I ill wait games like Philadelphie vs Neyyor (nhl) or something like that. In one day you will never have 5 game easy like this

Wish i dont amke too much error... im not really good in english

[u21c3f6]

I don't know if there is a single "best" answer.

You could actually design several different scenarios of the game using different winning criteria such as: Over the next 100 wagers the winner will be the one with the largest gain in bankroll based on flat wagers, progressive wagers, wagers of players choosing and/or within a range of size. You could then have a scenario based on ROI and/or win % with certain criteria. Each scenario would have a somewhat different strategy for trying to win which could make the game more interesting instead of playing the same scenario over and over.

From a personal point of view, when I try to decide what strategy and/or opportunities I want to wager on, I give overriding consideration to the Kelly growth rate of the strategy which essentially means that more often than not, I will be selecting the strategy with the highest win %. That might be a little complicated for your game (and maybe not).

Joe.