S.A.D. Predictions - Eights 2014

Iteration#0: 6th May

This is the initial 'flat' set-up assuming each crew has just its normal rating for its starting position and any predicted changes are simply due to random noise. Things will obviously look a little more interesting once people start sending me their estimates - see How to Contribute.

Predictions for the Saturday finishing order using the Statistical Analysis of Divisions (SAD) program based on collective estimates of boat speeds (top divisions only).

The following tables show the Wednesday starting positions of each crew, its score and average predicted change in position by the end of the week, followed by the crews most likely to occupy that position at the end of the week, listed in order of decreasing probability.

Click on crew name for graphics showing evolution of SAD score with each iteration and spread of Predicted Net Changes over the week.

The Evolution of SAD Score shows score each iteration as a red cross, the vertical extent representing the uncertainty. The Norm for the crew's starting position is shown as a blue dashed line. In the absence of any new information, each iteration the score will relax towards the "Norm" and the error bar expand towards ±1. If new estimates have been supplied the error bar will get smaller. This plot is updated as new scores come in, ie prior to the next formal iteration when the full SAD prediction is rerun. These intermediate scores are located half way between the last and the next iteration on the horizontal axis.

The Predicted Net Change histogram plot shows the probability of different net gains in position over the week (blue columns for rises, red columns for falls, white for remaining the same). The dashed line shows the mean change in position averaged over all possible outcomes (all the information for these histograms is extracted from the tables below).


Men's Divisions 1-3

[
Next]
Pos Start Crew Score Av.Rise Finishing Crew
1Pembroke 6.5 -0.9 Pemb 53% Chri 26% Orie 12% Univ 6% Wolf 3% SCat 0%
2Christ Church 6.4 -0.6 Chri 32% Pemb 23% Orie 20% Univ 12% Wolf 9% Magd 4% Trin 0%
3Oriel 6.3 -0.4 Orie 26% Chri 18% Univ 18% Wolf 14% Pemb 12% Magd 8% SCat 5% Trin 0%
4University 6.2 -0.3 Univ 21% Orie 17% Wolf 17% Chri 12% Magd 12% Pemb 8% SCat 8% Trin 5%
5Wolfson 6.2 +0.1 Wolf 20% Univ 18% Magd 17% Orie 12% SCat 11% Chri 8% Trin 7% Ball 4%
6Magdalen 6.1 +0.1 Magd 22% SCat 16% Wolf 15% Univ 11% Trin 11% Orie 8% Ball 6% Worc 6%
7St Catherine's 6.0 +0.0 SCat 21% Trin 16% Magd 16% Wolf 10% Ball 10% Univ 8% Worc 8% Hert 5%
8Trinity 5.9 -0.1 Trin 20% SCat 17% Ball 15% Worc 12% Magd 11% Hert 8% Wolf 6% Univ 6%
9Balliol 5.8 -0.2 Ball 20% Trin 17% Worc 17% SCat 11% Hert 11% SEdm 7% Magd 7% Kebl 5%
10Worcester 5.8 +0.1 Worc 19% Ball 17% Hert 16% Trin 11% SEdm 10% SCat 7% Kebl 7% Wadh 6%
11Hertford 5.7 +0.0 Hert 20% SEdm 15% Worc 15% Ball 12% Kebl 10% Wadh 8% Trin 8% SCat 5%
12St Edmund Hall 5.6 -0.1 SEdm 19% Hert 16% Kebl 14% Wadh 12% Worc 11% Ball 9% Linc 7% Trin 5%
13Keble 5.5 -0.2 Kebl 17% SEdm 16% Wadh 16% Linc 11% Hert 11% Worc 7% SJoh 7% Ball 7%
1Wadham 5.5 +0.1 Wadh 20% Kebl 20% Linc 16% SEdm 14% SJoh 10% Hert 8% SAnn 5% Worc 5%
2Lincoln 5.4 +0.1 Linc 18% SJoh 16% Wadh 15% Kebl 13% SAnn 11% NewC 9% SEdm 8% Bras 5%
3St John's 5.3 +0.0 SJoh 17% Linc 16% SAnn 15% NewC 13% Wadh 12% Kebl 9% Bras 9% SEdm 5%
4St Anne's 5.2 -0.1 SAnn 17% NewC 17% SJoh 16% Linc 13% Bras 12% Lady 8% Wadh 7% Kebl 6%
5New College 5.2 +0.2 NewC 18% Bras 17% SAnn 17% Lady 13% SJoh 12% Linc 8% Exet 8% Wadh 4%
6Brasenose 5.1 +0.1 Bras 18% Lady 16% NewC 15% SAnn 14% Exet 12% SJoh 8% Mans 7% Jesu 5%
7L.M.H. 5.0 +0.0 Lady 18% Exet 16% Bras 16% NewC 11% Mans 11% SAnn 9% Jesu 9% SJoh 5%
8Exeter 4.9 -0.1 Exet 18% Lady 16% Mans 16% Jesu 13% Bras 12% ChC2 8% NewC 7% SAnn 5%
9Mansfield 4.8 -0.2 Mans 18% Jesu 17% Exet 17% ChC2 12% Lady 12% Some 8% Bras 8% SPet 4%
10Jesus 4.8 +0.1 Jesu 17% ChC2 16% Mans 16% Some 12% Exet 12% SPet 8% Lady 8% Pem2 5%
11Christ Church II 4.7 +0.0 ChC2 17% Jesu 16% Some 15% Mans 13% SPet 11% Exet 9% Pem2 8% Ori2 5%
12Somerville 4.6 -0.1 Some 17% ChC2 16% SPet 14% Pem2 12% Jesu 12% Mans 10% Ori2 8% Corp 5%
13St Peter's 4.5 -0.2 SPet 16% Some 16% Pem2 15% ChC2 12% Ori2 11% Corp 8% Jesu 7% Mans 6%
1Pembroke II 4.5 +0.0 Pem2 20% SPet 18% Ori2 16% Some 15% Corp 10% ChC2 9% Quee 5% Jesu 4%
2Oriel II 4.4 +0.1 Ori2 16% Pem2 16% Corp 16% SPet 14% Quee 12% Mert 10% Some 9% ChC2 5%
3Corpus Christi 4.3 +0.0 Corp 16% Ori2 16% Quee 15% Mert 14% Pem2 12% SPet 10% SHug 9% Some 5%
4Queen's 4.2 -0.1 Corp 16% Mert 16% Quee 16% SHug 13% Ori2 13% Pem2 9% New2 8% SPet 5%
5Merton 4.2 +0.2 SHug 17% Quee 16% Mert 16% Corp 14% New2 13% Ori2 8% SEH2 8% Lina 4%
6St Hugh's 4.1 +0.1 SHug 17% New2 16% Mert 15% Quee 14% SEH2 13% Corp 9% Lina 8% Ori2 4%
7New College II 4.0 +0.0 SEH2 16% New2 16% SHug 16% Mert 13% Lina 12% Quee 9% Uni2 8% SAnt 5%
8St Edmund Hall II 3.9 +0.0 SEH2 17% New2 17% Lina 15% Uni2 12% SHug 12% SAnt 9% Mert 7% Quee 5%
9Linacre 3.8 -0.1 Lina 17% SEH2 17% Uni2 15% SAnt 14% New2 13% Mag2 9% SHug 8% Tri2 4%
10University II 3.7 -0.2 Uni2 16% Lina 16% SAnt 16% Mag2 13% SEH2 12% New2 8% Tri2 8% Wad2 5%
11St Antony's 3.7 +0.1 Uni2 15% SAnt 15% Mag2 15% Lina 13% Tri2 11% Wad2 9% SEH2 9% Bal2 6%
12Magdalen II 3.6 +0.0 Mag2 15% Uni2 14% SAnt 14% Wad2 13% Tri2 13% Lina 10% Bal2 9% SCa2 5%
13Trinity II 3.5 -0.2 Mag2 15% Tri2 15% Wad2 14% SAnt 12% Bal2 12% Uni2 10% SCa2 9% Lina 5%
1Wadham II 3.5 +0.1 Wad2 19% Tri2 18% Mag2 15% Bal2 15% SCa2 10% SAnt 10% Uni2 6% Wor2 5%

Women's Divisions 1-3

[
Next]
Pos Start Crew Score Av.Rise Finishing Crew
1St John's 6.5 -0.9 SJoh 50% Wadh 27% SEdm 14% Pemb 6% Magd 3% Chri 0% Ball 0%
2Wadham 6.4 -0.6 Wadh 28% SJoh 25% SEdm 21% Pemb 13% Magd 9% Ball 4% Univ 0% Chri 0%
3St Edmund Hall 6.3 -0.4 SEdm 23% Wadh 18% Pemb 18% Magd 15% SJoh 13% Ball 9% Chri 4% Univ 0%
4Pembroke 6.2 -0.3 Pemb 18% Magd 17% SEdm 17% Wadh 13% Ball 13% SJoh 8% Chri 8% Univ 4%
5Magdalen 6.2 +0.1 Magd 18% Pemb 18% Ball 16% SEdm 13% Chri 12% Wadh 8% Univ 7% Hert 4%
6Balliol 6.1 +0.1 Ball 19% Chri 17% Magd 15% Pemb 13% Univ 13% SEdm 9% Hert 7% Wadh 4%
7Christ Church 6.0 +0.0 Chri 19% Ball 16% Univ 16% Hert 12% Magd 11% Pemb 9% Mert 7% Kebl 6%
8University 5.9 +0.0 Univ 18% Chri 17% Hert 16% Mert 12% Ball 12% Kebl 9% Magd 7% Pemb 5%
9Hertford 5.8 -0.1 Hert 19% Univ 17% Mert 16% Kebl 13% Chri 12% Orie 8% Ball 7% Wolf 5%
10Merton 5.7 -0.2 Mert 17% Hert 16% Kebl 16% Orie 12% Univ 12% Wolf 8% Chri 8% Some 6%
11Keble 5.7 +0.1 Kebl 17% Mert 15% Orie 15% Hert 13% Wolf 11% Some 9% Univ 8% Osle 5%
12Oriel 5.6 -0.1 Orie 16% Kebl 15% Wolf 14% Mert 14% Some 13% Hert 9% Osle 8% Worc 5%
13Wolfson 5.5 -0.2 Orie 15% Some 15% Wolf 15% Kebl 12% Osle 11% Mert 10% Worc 8% Hert 6%
1Somerville 5.5 +0.2 Some 19% Wolf 19% Osle 15% Orie 15% Worc 10% Kebl 9% Mert 6% SCat 5%
2Osler House 5.4 +0.1 Osle 17% Worc 16% Some 15% Wolf 14% SCat 12% Jesu 9% Orie 9% NewC 5%
3Worcester 5.3 +0.0 Worc 16% Osle 16% SCat 15% Jesu 14% Some 12% Wolf 10% NewC 9% Orie 5%
4St Catherine's 5.2 -0.1 Worc 17% Jesu 16% SCat 16% Osle 13% NewC 13% Exet 8% Some 7% Wolf 6%
5Jesus 5.2 +0.2 Jesu 17% SCat 17% NewC 16% Exet 13% Worc 13% Osle 8% SAnn 8% Some 4%
6New College 5.1 +0.1 NewC 17% Exet 15% Jesu 15% SCat 14% SAnn 13% Worc 9% Linc 8% Lina 5%
7Exeter 5.0 +0.0 Exet 17% SAnn 16% NewC 16% Jesu 12% Linc 12% Lina 9% SCat 9% Worc 5%
8St Anne's 4.9 +0.0 SAnn 17% Exet 16% Linc 15% Lina 14% NewC 13% Trin 8% Jesu 8% SCat 5%
9Lincoln 4.8 -0.2 SAnn 16% Linc 16% Lina 16% Trin 13% Exet 13% Mans 9% NewC 8% Wad2 4%
10Linacre 4.8 +0.1 Linc 16% Trin 16% Lina 16% Mans 13% SAnn 12% Exet 8% Wad2 8% Quee 5%
11Trinity 4.7 +0.0 Trin 16% Lina 15% Linc 14% Mans 14% Wad2 12% Quee 9% SAnn 9% SHug 5%
12Mansfield 4.6 -0.1 Trin 16% Mans 15% Wad2 14% Lina 13% Quee 12% Linc 10% SHug 8% Corp 5%
13Wadham II 4.5 -0.2 Mans 16% Quee 15% Wad2 15% Trin 12% SHug 12% Corp 8% Lina 8% Linc 5%
1Queen's 4.5 +0.1 Quee 19% Wad2 18% Mans 16% SHug 15% Corp 10% Trin 9% SHil 5% Lina 4%
2St Hugh's 4.4 +0.2 SHug 16% Quee 15% Corp 15% Wad2 14% SHil 12% Bras 10% Mans 9% Trin 4%
3Corpus Christi 4.3 +0.0 SHug 15% Corp 15% Bras 15% SHil 15% Quee 12% Wad2 10% Lady 9% Mans 4%
4St Hilda's 4.2 -0.1 Corp 16% Bras 16% SHil 15% Lady 14% SHug 13% Quee 9% SJo2 8% Wad2 5%
5Brasenose 4.2 +0.2 SHil 16% Lady 16% Bras 16% SJo2 14% Corp 14% SHug 8% Wol2 8% Quee 3%
6L.M.H. 4.1 +0.1 Lady 16% SJo2 15% Bras 15% SHil 14% Wol2 13% Corp 9% SAnt 8% SPet 5%
7St John's II 4.0 +0.0 Wol2 16% Lady 16% SJo2 15% Bras 13% SAnt 12% SPet 10% SHil 10% Corp 4%
8Wolfson II 3.9 -0.1 SJo2 17% Wol2 16% SAnt 15% SPet 14% Lady 13% Pem2 9% Bras 8% SHil 5%
9St Antony's 3.8 -0.2 SAnt 16% SPet 16% Wol2 16% Pem2 14% SJo2 14% CCh2 9% Lady 8% Lna2 3%
10St Peter's 3.8 +0.1 SAnt 16% SPet 15% Pem2 15% Wol2 14% CCh2 13% SJo2 8% Lna2 8% Wor2 4%
11Pembroke II 3.7 +0.0 Pem2 15% CCh2 15% SPet 15% SAnt 14% Lna2 12% Wol2 9% Wor2 9% Her2 6%
12Christ Church II 3.6 +0.0 CCh2 16% Pem2 15% Wor2 13% SPet 13% Lna2 13% SAnt 11% Her2 9% Uni2 5%
13Linacre II 3.5 -0.3 Lna2 16% CCh2 16% Wor2 15% Pem2 13% Her2 12% Uni2 9% SPet 8% SAnt 5%
1Worcester II 3.5 +0.1 Wor2 19% Lna2 18% CCh2 16% Her2 15% Uni2 10% Pem2 10% Mer2 5% SPet 4%

Predicted Gains

Table shows the crews in decreasing order of predicted Gain in places for the week. Start is the crew's Wednesday starting position. (Eg if a crew has a 50% probability of rising one place and a 50% probability of rising two places, the predicted gain is +1.5.)

[Next]
Men's Divisions Women's Divisions
CrewGainStartCrewGainStart
1Mert+0.231 1Bras+0.231
2NewC+0.218 2Jesu+0.218
3Ori2+0.128 3Some+0.214
4Jesu+0.123 4SHug+0.228
5Wadh+0.114 5Lina+0.123
6Wolf+0.1 5 6SPet+0.136
7Worc+0.110 7Magd+0.1 5
8Linc+0.115 8Osle+0.115
9Wad2+0.140 9NewC+0.119
10SHug+0.13210Wor2+0.140
11Bras+0.11911Quee+0.127
12SAnt+0.13712Lady+0.132
13Magd+0.1 613Ball+0.1 6
14SJoh+0.01614Kebl+0.111
15Pem2+0.02715Corp+0.029
16Corp+0.02916Worc+0.016
17New2+0.03317Chri+0.0 7
18Lady+0.02018Exet+0.020
19Hert+0.01119SJo2+0.033
20SCat+0.0 720Pem2+0.037
21Mag2+0.03821Trin+0.024
22ChC2+0.02422Univ+0.0 8
23SEH2+0.03423SAnn+0.021
24Exet-0.12124CCh2+0.038
25Trin-0.1 825Wol2-0.134
26Quee-0.13026Orie-0.112
27SAnn-0.11727SCat-0.117
28SEdm-0.11228SHil-0.130
29Some-0.12529Mans-0.125
30Lina-0.13530Hert-0.1 9
31Mans-0.22231Linc-0.222
32Uni2-0.23632Wolf-0.213
33SPet-0.22633Wad2-0.226
34Ball-0.2 934SAnt-0.235
35Tri2-0.23935Mert-0.210
36Kebl-0.21336Lna2-0.339
37Univ-0.3 437Pemb-0.3 4
38Orie-0.4 338SEdm-0.4 3
39Chri-0.6 239Wadh-0.6 2
40Pemb-0.9 140SJoh-0.9 1


Contributor Statistics

The following table shows the statistics for each contributor's score compared to the crew scores after the latest SAD iteration. Average error is the average difference, irrespective of sign, between the contributor's score and the current score, while the Bias is the average difference including sign (positive bias means contributor scores are higher than current scores).
[
Next]

[nothing to show yet]


How to Contribute Information

To supply a score, just decide what sort of standard each crew seems to be, based on the following scale:

Within the SAD program, a difference of 0.4 corresponds to 1.5 lengths quicker over the course, ie a bump is expected at the finish line.

Don't feel obliged to supply an assessment for every crew - just anything you're confident about, even if they are similar to the numbers already listed (it adds weight to the estimate). Also, you can revise your estimates at a later date, in which case I'll remove your earlier estimate for any particular crew. I don't accept scores of your own college.

Normally I will list contributors by their initials (or any other 2-letter acronym you wish to be known by) at the top of this page, but I won't give out any other information on your identity or which crews you have assessed.

Anyone who contributes information will receive the current list of scores for their own college (scores only - contributor codes removed).

New contributors will have a weight +/-0.5 assigned to their estimate. Those who contributed in Eights 2013 will have their scores weighted by their average error, limited to range 0.2-0.5

Email scores to dudhia@atm.ox.ac.uk


The Gory Details

The SAD program calculates a statistical average of the results of 10000 runs of Bumps. In each run the crew speed is varied by some random amount about the mean speed listed (in brackets), the 1-sigma variation depending on the number, accuracy and time of estimates supplied by contributors. This allows for errors in the assessed speed, but within each `day' of the Bumps, there is an additional term to allow for random variations in a crew's performance from day to day, varying from +/-0.2 for 1st Division crews to +/-0.4 for 3rd Division.
The program allows for overbumps, Sandwich boats, and separate rules for Torpids and Eights, but makes no allowance for crews below those listed.
FORTAN source code.

Contributor estimates are each given a weight of +/-0.5, but those who proved more accurate last Torpids will have those errors used instead. These weights apply for the iteration in which the score was used, but on subsequent iterations all the older scores are reduced in weight (i.e. recent estimates carry more weight).


SAD BaStaRD

(SAD Bayesian Statistical Revised Distribution)

Once racing gets under way, I'll start running SAD BaStaRD (PS Thanks to Tom Fleming for the acronym) to `tune' the actual estimates to values that best fit the results of each day's racing.

The SAD BaStaRD program runs the SAD program and calculates the probability of the actual outcome of each day's racing using the initial scores. It then perturbs each crew's score by +/-0.1, and also the uncertainty in the score by +/-0.1, and recalculates the probability. Whichever perturbation leads to the greatest probability is kept, and the process then repeated. This is continued until, for a given set of scores, all further perturbations result in a lower probability for the actual outcome, and these scores are then taken as the best.

The final result should be scores which best represent each crew's true speed as well as give a means of evaluating the accuracy of each contributor.

If a crew bumps each day, the score that would give the actual outcome with the greatest probability is infinity, so to keep things sensible the original SAD estimates and their uncertainties are used as an a priori constraint.