Academy Cup
Dates 2015
> Sun 25 Jan (Spring)
> Sun 28 Jun (Summer)
> Sun 18 Oct (Autumn)
> Sun 20 Dec (Winter)
Check in Close
> U12 09:30
> U14 13:30
> U10 15:30
Seymour Leisure Centre
Seymour Place, W1H 5TJ,
+44 20 77238019
Get directions

> Boys Foil (U14), (U12), (U10)
> Girls Foil (U14), (U12), (U10)
Download Poster

Data Analysis
> 2010 Match Analysis
> 2010 Ranking Analysis
> New Format comparison
> Old Format breakdown

Data Analysis

No sport will find a place in the league of reputable sports if it does not use scientific analysis and testing. This is intended to be an official study paper on one aspect of fencing, with scientific analysis based on real data.

The Data

Firstly we will present the calculated data which is based on three Academy Youth Cup competitions in 2010 (Winter, Spring & Summer) in three categories (U10, U12, U14) boys and girls. This is data taken from DE matches only. All of these competitions were run based on the usual single DE tableau format. After recording, organising and grouping this data we were able to see three categories of match results distinct from each other.

1) The no-contest matches: These are matches lost with a final score of 0, 1 or 2 only for the loser. Because we have DE matches to 7 hits, 10 hits and 15 hits depending on age category we had to work on a percentage to get an even set of data from all the matches. Therefore, matches in this category were fixed at 20% or below, representing the score of the loser compared to that of the score of the winner.

2) The tightly fought matches: These are matches that could have gone either way. This means there were only two or three hits in it at the most. ie. 14-15 score. Once again we applied the percentage format to get to an even set of data. These are matches with 70% or higher of the score of the loser compared to that of the score of the winner.

3) The average match: These are matches that are not a walk over but are reasonably one sided with various level. This is the percentage between 20% & 70%.

Looking at these categories the first and the second are particularly interesting. It is easy to conclude that the first category cannot be either positive or useful. Therefore we will call them "the useless matches". These are matches which are better not to have taken place. It is also easy to conclude that the second category matches are highly contested, which make the fencers inspired to produce their skills and inspiring for the spectators. These matches are full of benefits for the fencers. Therefore we will call them "the inspiring matches".


Data Projection : Score Analysis
Below is a chart representing the percentage of matches according to the three categories above after analysing all of the 190 DE matches.

As this chart shows, almost a quarter of the matches are of the first category. All of these matches are not only useless but they are in fact the source of a problem for the losing fencer. It is therefore obvious that the fencers will be at a far greater advantage if we could reduce the number of useless matches. At the same time if we could possibly increase the number of inspiring matches this would hugely increase the experience and positiveness gained from the competition.

There are many formats on the market. Plate competition, two wave system and so on. More DE matches, as in other systems such as the repechage system and so on, are not the right answer for the young category of fencers. The length of the competition should not increase significantly. It is also important to keep the overall time of the competition short. After testing many formats the multi DE tableau format came up with the right answers. However the question is: what group size should the DE tableau be divided into? Divide the competition into two, three, or more equal numbers? Should it be perfect tableau of eight?

Well, we looked at the data available and the answer was there.

Data Projection : Ranking Analysis (1)

Firstly we had to record the seeding of the fencers after the first round and see what is the likelihood of lower seed fencers beating higher seed fencers.

The total number of these matches was 40 matches out of a total of 190 matches. As you can see in this chart, just over one fifth of matches were won by lower seed fencers. This is a reasonably substantial percentage and any format must preserve this.

Now that we have determined the importance of this, we can start to analyse the data to find out where is the breaking point.

Data Projection : Ranking Analysis (2)

This chart calculates the percentage of lower seed fencers winning DE matches 1-6 places above and those further apart. This clearly shows that the overwhelming majority of matches won by lower seed fencers were against higher seeds not more than six places above.

Therefore, we have our answer. The competition should be divided into groups of six to preserve these victories and keep the format fair to all.


Data Projection : Ranking Analysis (3)

However, after checking more data we discovered that most of the lower seed victories happened in the top 8 tableau: 27 out of 40 matches were in the top 8. This leaves only 13 matches where lower seed fencers won in the rest of the competition stages. This chart shows it in relevant percentages.

Therefore it was important to reserve the top 8 tableau of the competition. But thereafter divide the competition into groups of six.

Testing New Format

With all this data and analysis pointing to the right format for this competition, we had to put it to the test and then scrutinise the results to determine if all of this works in practice as well as on paper.

Therefore the last Academy Cup competition (Autumn) was run with this new format which incidentally was the largest competition of all and therefore gave a substantial amount of data to be analysed. A total of 84 DE matches were recorded and calculated. Below are the results with comparisons. New format v old format.

Match Analysis
The first objective of this format was to achieve less "useless matches" and more "inspiring matches". After data analysis we can see clearly a huge success in the new format. The percentages are clarified in the chart below, but in terms of numbers there were only four matches lost with scores of 3 or less hits. This was great news. Also there were 17 matches with close scores ie 14-15. This is more inspiring matches than all the previous three competitions put together which had a total of 15 inspiring matches. Again a huge success there.
Ranking Analysis
This is all good. But it is important to make sure that we do not lose the number of lower seeds beating higher seeds. Once all the data was crunched after the last competition with the new format the results were very encouraging indeed. The percentages below explain it all. In terms of numbers, in the last competition with the new format, there were 20 matches whereby lower seed fencers beat higher seed fencers. This more than all the last three competitions put together, where there were 13 such matches. Again the new format wins hands down.
Final Consideration

With all this encouraging data, there are only a small number of matches that may be at loss with this new format. So the final stage was to examine these and see what conclusions can be drawn.

Firstly let us look at the data from the three competitions with the old format and identify those matches which will be at a disadvantage with the new format.

Out of 190 DE matches from all three competitions forty of these included lower seed winning against higher seed. But 27 of these matches were from the top 8 tableau. Because we preserved the top 8 tableau in the new system these were not affected. Therefore we are left with 13 matches. Out of these 13 matches, only four matches were won by lower seed fencers against higher seed fencers more than 6 above. Because in the new format we have divided the competition into groups of six, this will continue to work for 9 other matches.

Now we only have a problem with four matches out of 190 original matches from the three competitions and three age groups each. A very negligible percentage which is more than acceptable. Nonetheless, we looked into those four matches. There was only one of those four matches which a lower seed fencer beat someone significantly higher and managed to get through two DE's. This match was in the U14 boys category. Therefore we have decided not to include the U14 in this new format for 2011 until we collect more data and analyse it further.

This left us with the following three matches:

> The first number is the original seed of the losing fencer.
> The second number is the original seed of the winning fencer.
> The third number in red is the difference between the seeding of the fencers.
> The following three figures are the scores and score analysis, which is not relevant for this particular part.

We looked further into those three matches and we found that none of them went beyond the first DE. As everyone knows, this could be just one style against another which can happen in fencing. But when we looked into the data further, we found that in these cases the poule round was not very representative of final results. This meant that either the winning fencer was in a particularly weaker poule or the losing fencer was in a particularly stronger poule.

As we are now seeding the first round of poules from the ranking list, as well as having large poules, this should not happen in future competitions.

Medal Analysis

Once we had finished all the analysis we thought, hold on, the most obvious question that everyone will have about the new system has not clearly been looked at. Will this new system reduce the chances of fencers who do not do well in the first round getting into the medals?

We recorded and grouped all the finals DE matches and kept the original seeding of the fencers which they received after the first round. The result was conclusive. The winners of all the 18 finals were fencers who were seeded fisrt and second after the poules. Only on three occasions were they seeded in third and fourth place. As the data is not too long we have included it below:

Original Seed
Silver medal
Original Seed
Gold Medal
Winter 2010 U12 Girls Finals 2 1
Winter 2010 U12 Boys Finals 4 3
Winter 2010 U14 Boys Finals 1 2
Winter 2010 U14 Girls Finals 2 1
Winter 2010 U10 Girls Finals 1 2
Winter 2010 U10 Boys Finals 7 4
Spring 2010 U10 Boys Finals 6 1
Spring 2010 U10 Girls Finals 2 1
Spring 2010 U12 Boys Finals 4 2
Spring 2010 U12 Girls Finals 2 1
Spring 2010 U14 Boys Finals 1 2
Spring 2010 U14 Girls Finals 1 2
Summer 2010 U14 Boys Finals 2 1
Summer 2010 U14 Girls Finals 5 2
Summer 2010 U10 Boys Finals 4 3
Summer 2010 U12 Boys Finals 7 1
Summer 2010 U12 Girls Finals 5 2
Summer 2010 U10 Girls Finals 1 2

Next, we analysed the two bronze medal winners to see if any low seed fencer had a chance to make it. Again the evidence was conclusive. The overwhelming majority of the time the bronze medals were won by seeds 1-4 and on the odd occasion it was by seeds 5-8. Never has a seed below 8th ever made it to Bronze medal.

Therefore the results are clear:
Gold Medal: Never won by lower than 3rd seed
Silver Medal: Never won by lower than 7th seed
Bronze Medal: Never won by lower than 8th seed

Therefore, in the new system, by making the first group include the top eight fencers, we have virtually made sure anyone whoever has a chance to get a medal will still have that same chance to get a medal. No one is at a disadvantage and no fencer will lose their chance of a medal with the new system. The results have been very conclusive on this. Also by using the new seeding system for the first round based on fencer ranking will further ensure that the seedings after the poules are even more accurate.

This new format will deliver better, more exciting and much more beneficial matches to the fencers. The new format will be more fair and the final results will be a lot more representative of the standard of the fencer. The majority of the matches will be meaningful and reflect positively on the attitude of the fencers. That said we will continue to monitor and analyse data to make sure we can continue to deliver a format that works well for the fencers. We hope that you appreciate our efforts and enjoy these findings.
(C) The fencing Academy