S.A.D. Predictions - Eights 2014

Iteration#1: 7th May

This is the first iteration containing contributions from SG, AA, JW, CW and AD.

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

[
Prev] [Next]
Pos Start Crew Score Av.Rise Finishing Crew
1Pembroke 6.5 -0.5 Pemb 63% Chri 25% Orie 12% Univ 0% Wolf 0% SCat 0%
2Christ Church 6.5 -0.1 Chri 48% Pemb 25% Orie 24% Magd 1% Wolf 1% Univ 1%
3Oriel 6.5 +0.3 Orie 51% Chri 19% Pemb 9% SCat 7% Univ 6% Wolf 5% Magd 4% Ball 0%
4University 6.1 -0.8 Univ 41% Wolf 20% Orie 10% Magd 9% Trin 6% SCat 6% Chri 6% Pemb 3%
5Wolfson 6.2 -0.3 Wolf 35% Univ 30% Magd 14% SCat 6% Trin 6% Ball 5% Orie 2% Chri 2%
6Magdalen 6.2 -0.1 Magd 33% Wolf 25% Univ 15% SCat 9% Trin 7% Worc 5% Ball 5% Orie 1%
7St Catherine's 6.0 -0.4 Magd 24% SCat 23% Trin 15% Wolf 11% Worc 10% Ball 10% Univ 6% Hert 1%
8Trinity 5.9 -0.4 Trin 20% Worc 20% SCat 19% Ball 16% Magd 10% SEdm 6% Hert 4% Wolf 3%
9Balliol 5.8 -0.5 Worc 24% Ball 17% Trin 16% SCat 12% Kebl 11% SEdm 8% Hert 8% Magd 3%
10Worcester 6.0 +0.7 Kebl 20% Worc 19% Hert 13% Ball 13% Trin 11% SEdm 10% SCat 9% Wadh 4%
11Hertford 5.6 -0.7 Kebl 21% Hert 17% Ball 11% Worc 11% SEdm 10% Trin 10% SCat 8% Wadh 7%
12St Edmund Hall 5.6 -0.3 Kebl 22% Hert 21% SEdm 13% Ball 12% Wadh 11% Trin 9% Worc 7% SJoh 6%
13Keble 6.0 +1.6 Hert 19% Kebl 17% SEdm 17% Wadh 16% Ball 12% SJoh 8% NewC 6% Worc 3%
1Wadham 5.5 +0.1 Wadh 28% SEdm 18% SJoh 14% Hert 13% Linc 12% Kebl 8% NewC 5% SAnn 2%
2Lincoln 5.1 -1.0 Linc 24% SJoh 16% Wadh 15% NewC 13% SEdm 10% Bras 10% SAnn 6% Hert 5%
3St John's 5.3 +0.3 Linc 29% SJoh 14% NewC 13% SAnn 12% Bras 11% Wadh 9% SEdm 7% Lady 5%
4St Anne's 5.0 -0.8 Linc 23% SAnn 20% SJoh 14% NewC 13% Bras 12% Lady 11% Wadh 6% Exet 1%
5New College 5.2 +0.5 SAnn 26% Lady 17% Bras 15% NewC 15% SJoh 11% Linc 10% Wadh 4% Exet 2%
6Brasenose 5.1 +0.4 Lady 26% SAnn 22% Bras 18% NewC 15% SJoh 8% Mans 5% Exet 4% Linc 2%
7L.M.H. 5.1 +0.9 Lady 23% Mans 17% Bras 14% ChC2 11% NewC 10% SAnn 9% Exet 8% SJoh 5%
8Exeter 4.3 -1.7 Mans 33% Exet 13% ChC2 12% Lady 12% Bras 10% Jesu 8% NewC 7% SAnn 3%
9Mansfield 4.8 +0.6 Mans 26% Jesu 18% Exet 15% ChC2 15% SPet 7% Bras 6% Some 5% Lady 4%
10Jesus 4.5 -0.7 Jesu 18% ChC2 15% SPet 13% Exet 13% Mans 12% Some 10% Pem2 7% Bras 5%
11Christ Church II 4.7 +0.4 SPet 19% Exet 17% Jesu 15% Some 11% Pem2 11% ChC2 10% Ori2 7% Mans 5%
12Somerville 4.4 -0.8 Exet 26% Jesu 15% SPet 15% Some 12% ChC2 11% Pem2 8% Ori2 6% Corp 5%
13St Peter's 4.7 +0.8 SPet 19% Some 18% Pem2 15% Jesu 13% Ori2 11% ChC2 11% Corp 10% Quee 4%
1Pembroke II 4.5 +0.0 Some 20% Pem2 17% SPet 16% Ori2 14% Corp 10% ChC2 9% Jesu 8% Quee 5%
2Oriel II 4.4 +0.2 Quee 17% Ori2 17% Corp 16% Pem2 16% Some 15% SPet 8% ChC2 6% Mert 4%
3Corpus Christi 4.3 +0.2 Quee 27% Corp 16% Ori2 15% Mert 14% Pem2 12% Some 7% New2 4% SPet 3%
4Queen's 4.3 +0.5 Mert 27% Quee 21% New2 12% Corp 12% Ori2 11% Pem2 8% SEH2 5% SHug 3%
5Merton 4.2 +0.3 Mert 24% New2 20% Quee 13% Corp 10% Ori2 8% SEH2 8% SHug 6% Pem2 5%
6St Hugh's 3.7 -1.7 New2 23% Mert 20% SEH2 11% SHug 10% Corp 9% Lina 8% Quee 8% Ori2 6%
7New College II 4.2 +0.8 New2 19% SHug 19% SEH2 15% Lina 11% Uni2 8% SAnt 8% Mert 8% Corp 8%
8St Edmund Hall II 3.9 -0.4 SHug 25% SAnt 16% SEH2 15% Lina 14% New2 12% Uni2 11% Mag2 4% Mert 2%
9Linacre 3.8 -0.5 Tri2 24% SHug 19% SAnt 16% SEH2 10% Lina 9% Uni2 9% Mag2 7% New2 6%
10University II 3.7 -0.7 Tri2 18% SHug 16% SEH2 15% SAnt 14% Lina 13% Uni2 11% Mag2 7% New2 3%
11St Antony's 4.0 +0.7 Tri2 18% SAnt 17% Lina 15% Uni2 13% SEH2 12% Mag2 10% Wad2 8% Bal2 5%
12Magdalen II 3.6 -0.7 Lina 16% Uni2 14% Tri2 13% SAnt 12% Mag2 12% Wad2 10% SEH2 10% Bal2 7%
13Trinity II 4.2 +1.7 Uni2 15% Mag2 15% Wad2 13% SAnt 10% Bal2 10% Tri2 10% Lina 9% SCa2 8%
1Wadham II 3.5 -0.3 Mag2 20% Wad2 19% Bal2 15% Uni2 14% SCa2 10% Tri2 9% SAnt 5% Wor2 5%

Women's Divisions 1-3

[
Prev] [Next]
Pos Start Crew Score Av.Rise Finishing Crew
1St John's 6.5 -1.0 SJoh 52% Wadh 27% SEdm 12% Pemb 8% Magd 0% Ball 0%
2Wadham 6.5 -0.2 Wadh 40% SEdm 21% SJoh 19% Pemb 11% Ball 6% Magd 2%
3St Edmund Hall 6.4 -0.1 SEdm 31% Wadh 23% Pemb 16% SJoh 13% Ball 9% Magd 7% Chri 1%
4Pembroke 6.2 -0.2 Pemb 23% SEdm 22% Magd 18% Ball 15% SJoh 9% Wadh 8% Chri 4% Univ 2%
5Magdalen 6.0 -0.5 Magd 25% Ball 19% Pemb 18% Chri 10% Univ 10% SEdm 9% SJoh 7% Wadh 1%
6Balliol 6.1 +0.3 Univ 25% Magd 22% Chri 20% Ball 17% Pemb 11% SEdm 4% Hert 1% Mert 0%
7Christ Church 5.9 -0.2 Univ 30% Chri 22% Magd 13% Ball 11% Kebl 10% Pemb 7% Hert 3% Mert 2%
8University 6.1 +1.0 Univ 21% Chri 21% Hert 10% Kebl 9% Ball 9% Magd 9% Orie 8% Mert 6%
9Hertford 5.6 -0.8 Hert 26% Mert 15% Chri 15% Kebl 12% Univ 10% Orie 9% Ball 8% Magd 4%
10Merton 5.6 -0.6 Hert 30% Mert 24% Kebl 13% Orie 13% Wolf 6% Ball 6% Chri 5% Univ 2%
11Keble 5.7 +0.2 Mert 25% Hert 18% Wolf 17% Kebl 16% Orie 16% Worc 5% Osle 2% Chri 1%
12Oriel 5.6 +0.1 Wolf 27% Mert 17% Orie 15% Kebl 14% Worc 11% Hert 9% SCat 3% Osle 3%
13Wolfson 5.6 +0.5 Wolf 26% Worc 17% Orie 13% Kebl 10% Mert 8% SCat 7% Some 7% Jesu 6%
1Somerville 5.1 -1.2 Some 22% Worc 19% Wolf 19% Orie 13% Kebl 10% Osle 6% SCat 6% Jesu 4%
2Osler House 4.7 -1.9 Some 25% Worc 16% SCat 15% Jesu 11% Orie 9% Osle 7% NewC 7% Kebl 6%
3Worcester 5.5 +1.5 Some 25% Jesu 15% SCat 15% Worc 14% NewC 14% Osle 10% Orie 5% Exet 1%
4St Catherine's 5.2 +0.3 Jesu 19% SCat 18% NewC 18% Osle 17% Some 15% Worc 10% Exet 2% SAnn 1%
5Jesus 5.2 +0.6 Osle 22% NewC 20% Jesu 15% SCat 13% Linc 9% SAnn 8% Exet 6% Worc 5%
6New College 5.1 +0.6 Osle 29% NewC 12% Lina 12% Linc 11% Jesu 10% SCat 10% SAnn 9% Exet 5%
7Exeter 4.2 -2.1 Lina 24% SAnn 17% Linc 16% NewC 13% Jesu 11% SCat 8% Exet 8% Trin 1%
8St Anne's 4.7 -0.3 Lina 26% SAnn 21% Linc 11% Trin 10% Exet 10% NewC 8% Jesu 6% Mans 4%
9Lincoln 4.8 +0.0 Trin 20% Lina 18% SAnn 17% Exet 13% Linc 12% Mans 9% NewC 6% Jesu 4%
10Linacre 5.1 +1.8 Trin 26% Mans 22% Exet 18% SAnn 11% Lina 9% Linc 8% Wad2 2% NewC 2%
11Trinity 4.9 +0.7 Exet 35% Mans 17% Trin 16% SAnn 10% Linc 8% Lina 5% Wad2 3% SHug 2%
12Mansfield 4.9 +0.7 Mans 26% Trin 19% Linc 12% Wad2 12% SHug 9% Quee 5% SAnn 5% Corp 5%
13Wadham II 4.5 -0.8 Wad2 22% Mans 15% Linc 13% Quee 12% SHug 12% Corp 11% Trin 6% SHil 4%
1Queen's 4.4 -0.9 Wad2 31% Quee 23% Corp 16% SHug 16% Mans 6% SHil 4% Bras 2% Trin 1%
2St Hugh's 4.5 +0.0 Quee 23% Corp 21% Wad2 18% SHug 15% SHil 9% Bras 8% Lady 4% Mans 1%
3Corpus Christi 4.6 +0.7 Quee 21% Corp 19% SHug 17% SHil 11% Bras 11% Wad2 9% Lady 8% SJo2 5%
4St Hilda's 4.2 -0.5 SHug 16% Corp 16% Bras 15% SHil 14% Lady 13% Quee 12% SJo2 8% Wol2 5%
5Brasenose 4.2 -0.1 SHil 20% Bras 18% Lady 16% SJo2 14% Wol2 9% SHug 8% Corp 8% Quee 3%
6L.M.H. 4.1 +0.0 Lady 18% SHil 17% Bras 17% SJo2 17% Wol2 15% SPet 7% SAnt 5% Corp 2%
7St John's II 4.0 +0.0 Wol2 16% Lady 15% SJo2 14% Bras 13% SPet 13% SHil 11% SAnt 11% Pem2 6%
8Wolfson II 3.9 +0.0 SJo2 14% Wol2 14% SAnt 14% SPet 14% Lady 12% Pem2 11% Bras 9% SHil 6%
9St Antony's 3.7 -0.5 SAnt 16% Pem2 14% Wol2 13% SJo2 13% SPet 13% CCh2 11% Lady 9% Lna2 5%
10St Peter's 3.8 +0.1 SAnt 18% Pem2 13% CCh2 13% SPet 12% Wol2 12% Lna2 10% SJo2 9% Lady 5%
11Pembroke II 3.7 +0.0 SAnt 17% Pem2 13% CCh2 12% SPet 12% Lna2 11% Wor2 10% Wol2 10% Her2 6%
12Christ Church II 3.6 -0.1 Wor2 16% CCh2 13% SPet 13% SAnt 12% Pem2 12% Lna2 11% Her2 8% Uni2 6%
13Linacre II 3.5 -0.3 Wor2 17% Lna2 13% Pem2 13% CCh2 13% Her2 11% SPet 10% Uni2 9% Mer2 5%
1Worcester II 3.6 +0.4 Wor2 21% CCh2 16% Lna2 15% Her2 12% Pem2 12% Uni2 9% SPet 6% Mer2 6%

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.)

[Prev] [Next]
Men's Divisions Women's Divisions
CrewGainStartCrewGainStart
1Tri2+1.739 1Lina+1.823
2Kebl+1.613 2Worc+1.516
3Lady+0.920 3Univ+1.0 8
4New2+0.833 4Mans+0.725
5SPet+0.826 5Corp+0.729
6Worc+0.710 6Trin+0.724
7SAnt+0.737 7Jesu+0.618
8Mans+0.622 8NewC+0.619
9Quee+0.530 9Wolf+0.513
10NewC+0.51810Wor2+0.440
11Bras+0.41911SCat+0.317
12ChC2+0.42412Ball+0.3 6
13Orie+0.3 313Kebl+0.211
14SJoh+0.31614Orie+0.112
15Mert+0.33115SPet+0.136
16Ori2+0.22816Wol2+0.034
17Corp+0.22917SHug+0.028
18Wadh+0.11418SJo2+0.033
19Pem2+0.02719Linc+0.022
20Magd-0.1 620Pem2+0.037
21Chri-0.1 221Lady+0.032
22Wolf-0.3 522Bras-0.131
23SEdm-0.31223CCh2-0.138
24Wad2-0.34024SEdm-0.1 3
25SEH2-0.43425Wadh-0.2 2
26Trin-0.4 826Chri-0.2 7
27SCat-0.4 727Pemb-0.2 4
28Ball-0.5 928SAnn-0.321
29Lina-0.53529Lna2-0.339
30Pemb-0.5 130Magd-0.5 5
31Jesu-0.72331SHil-0.530
32Uni2-0.73632SAnt-0.535
33Hert-0.71133Mert-0.610
34Mag2-0.73834Wad2-0.826
35SAnn-0.81735Hert-0.8 9
36Univ-0.8 436Quee-0.927
37Some-0.82537SJoh-1.0 1
38Linc-1.01538Some-1.214
39Exet-1.72139Osle-1.915
40SHug-1.73240Exet-2.120


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).
[
Prev] [Next]

       Pos#   Name   Avg.Error   Bias   No.estimates supplied
         1     SG      .053     -.032        19
         2     TT      .122      .005        80
         3     AA      .167      .133         6
         4     CW      .218     -.045        22
         5     AD      .236     -.036        14
         6     JW      .356      .044         9

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.