swagbirddragoln-blog

[swagbirddragoln-blog]

End of Round 3

As promised, here are the highlights of my rankings and predictions of the results of the crapshot bracket. If you want the full explanation of how I am ranking the crapshots, see here. (TL;DR Future match ups are predicted based on past performance. Points are awarded based on how closely it lost, and how many points the crapshot it lost to had.) At present, the winning crapshot is given 100 points. The results of each round are given for crapshots eliminated in that round. Enjoy

Best and Worst of Round 1

Some amount of good crapshots were almost guaranteed to be eliminated in the first round, as well as the worst of them.

The Best

93 [248]The Vine with 28.13 points

85 [235]The Prospector with 29.11 points

77 [461]The Lethargy with 30.16 points

72 [222]The Toast with 31.56 points

60 [352]The Opinion with 33.51 points

The Worst

508 [80]The Promotion with 2.63 points

509 [24]The Admission with 2.55 points

510 [184]The Slap 2 with 2.40 points

511 [12]The Amplifier with 2.39 points

512 [469]The Intervention 2 with 1.72 points

It is worth remembering that these numbers are not set in stone, and will continue to shift as the bracket continues. 

Best and Worst of Round 2

With half the crapshots gone, many more good ones were certain to go this round. It was almost inevitable however, that some not so good ones would slip through.

The Best

65 [453]The Backup with 32.98

55 [456]The Human 2.0 with 34.25

52 [460]The Transport with 35.14 points

51 [367]The Trainer with 35.16 points

29 [443]The Gaslight with 47.04 points

The Worst

450 [34]The Slam with 7.07 points

458 [20]The Gaming with 6.55 points

478 [194]The Hood with 5.56 points

481 [263]The Unboxing 2 with 5.35 points

488 [193]The Tokens with 4.69 points

Best and Worst of Round 3

Down to the last quarter, we’re mostly left with good crapshots. However, I think you’ll be surprised by just how low a few that made it this far are ranked. 

The Best

46 [419]The Damage with 36.69 points

38 [115]The Elite Hacks with 40.82 points

35 [242]The Prisoner with 42.21 points

33 [204]The Resolution with 44.07 points

25 [196]The DUI with 48.83 points

The Worst

292 [201]The Redial with 14.36 points

344 [477]The Taste Test with 11.95 points

348 [220]The Bush with 11.52 points

382 [169]The Pinatas with 10.43 points

421 [482]The Bet with 8.88 points

And that’s it for all the rounds so far. Now we’re on to the most exciting bit, the predicted winners.

The Top Ten

10 [130]The Weather with 65.75 points. Eliminated in Round 6

9 [268]The Supplement with 66.26 points. Eliminated in Round 7

8 [468]The Retail Therapy with 65.75 points. Eliminated in Round 8

7 [455]The Fruit with 68.60 points. Eliminated in Round 7

6 [200]The Edit with 69.06 points. Eliminated in Round 6

5 [483]The Hand with 73.98 points. Eliminated in Round 6

4 [348]The Lumber with 74.31 points. Eliminated in the Final Round

3 [446]The Tech Support with 77.52 points. Eliminated in Round 7

2 [411]The Farewell with 78.71 points. Eliminated in Round 8

1 [270]The Home Show with 100 points. The ultimate Winner

And that’s it for this round. I’ll be back again at the end of next round with more highlights. If anyone is interested in the current fate of particular crapshots or series, please feel free to message me. 

[swagbirddragoln-blog]

Explanation of Prediction/Ranking Algorithms

During the weeks that the crapshot bracket has been running, I have put together a spreadsheet that uses the results of each match up to both predict future match ups and rank each crapshot based on those results. Before I post the results of these algorithms, I thought I’d give an explanation of how they work.

Prediction

In order to predict the percentage of votes each crapshot will receive in each match up what I do is take the average percentage of votes it received in previous matchups -50% and divide that by the sum of the average number of votes received by both crapshots in the matchup -100%. So if we have two crapshots matching up and one has averaged 60% and the other 75%, what you end up with is (0.75-0.50)/(0.60+0.75-1.00) for the higher averaging crapshot. Simplifying that down, we get 0.25/0.35 which is approx 0.714 or 71.4%. If we run the same calculation in the other direction, we get 0.10/0.35 which is approx 0.286 or 28.6%.

The other important properties of this algorithm are what would happen if both crapshots have the exact same average and what would happen in the most extreme possible case. In the event of equal averages, it doesn’t matter what the number is, the equation will spit out exactly 0.50 or 50%. If you look at the equation we might write this case as (a-0.50)/(a+a-1.00) where a is the average. This can be simplified to (a-0.50)/(2a-1.00) which is equal to (a-0.50)/2(a-0.50), and then you can simplify out the (a-0.50) from the top and bottom, leaving 1/2 which is 0.50 or 50%. The other extreme is where one crapshot averages very close to 50% and the other 100%. If that were to happen, the equation would be (1.00-0.50)/(1.00+0.50-1.00) which simplifies to 0.50/0.50 which equals 1.00 or 100%. So this algorithm will always predict a number between 0.50 and 1.00 for the winning crapshot. 

Ranking

The only information we have to work with is that each crapshot is definitely better than the one it defeated and how much it defeated it by. What I went with was to assign each crapshot a point value based on how many points the crapshot it lost to had and how much it lost by. Part of this is also assigning an arbitrary number of points to the winning crapshot. It doesn’t matter what this value is, as any change to it is reflected exactly in all the other crapshots (ie. If I double this value, all the other crapshots are doubled in value) so it will not affect the order. 

The algorithm I use sets the ratio of the two point values equal to the ratio between the amount of votes received. As an example, if the winning crapshot in a match up had 75% of the votes and had 3 points, then the loser of that match up will receive 1 point. What percentage of the winner’s points the loser is to receive is calculated by taking 1 divided by the winner’s percentage of the votes and subtracting 1. Then multiply that by the number of points, and you have the points earned by the loser. That gives the following equation P(1/w-1) where P is the point value and w is the winner’s percentage. That can be simplified to P/w-P. To reuse the above example, it would be filled in as 3/.75-3 which is equal to 4-3 which is equal to 1. 

Once this algorithm has been applied to every match up, ranking order is simply determined by sorting the crapshots from highest points to lowest. In the event of a tie, the one which was eliminated later in the bracket will be placed higher.

I will be posting highlights of my results after each round of the crapshot bracket, and the full results after it has finished.