For decathlon lovers: analysing decathlon performances


Main message board: for the discussion of topical track & field items only.

For decathlon lovers: analysing decathlon performances

Postby olorin » Sun Oct 14, 2012 3:22 am

Decathlon is arguably track and field most complicated event for both the athletes that compete in it and us track and field fans who try to evaluate the performances of these amazing athletes. Unlike individual events a decathlete will never set the world a light in any of the events he compete in. There are only a few performances in the decathlon that are worthy of a place in an Olympic final. For example, one of the best ever performances in the decathlon was Suarez’s 77.47 in the JT. This would rank him in 27th place in the London Olympic just between Dayton Marquez (COL) and Jarrod Bannister (AUS) both unknown to most T&F fans.

So how can we judge the performances of top decathletes? One alternative, is trying to compare their performances to the performances of top athletes in individual events. However, this method explicitly assumes that on average decathletes compete on the same level in all ten events. However, this isn’t true in real life. For example, most of the top decathletes will jump ~8.00m, which is sometimes enough to make an Olympic final in the long jump. In contrast, the best ever DT is 55.87 (Clay) that would rank him in the last place in the qualification round in the last Olympics.

So, in order to evaluate the performances of decathletes we must compare them to other decathletes across the ten events. With that in mind, I created an “average decathlete” that in each event scores the average result of the top 19 decathletes in the all time list. The marks of our hypothetical decathlete are: 10.63-7.77-15.56-2.07-47.80-14.03-47.99-4.99-65.35-4:33.36 that sum up nicely to 8,799 points. Then, I compared the performance of each of the decathletes to our hypothetical decathlete. For example, while setting the WR Eaton ran 10.21, which is 98 points better than our hypothetical decathlete. Similarly, he jumped 8.23 that is 118 points better, while his mark in the SP (14.20) is 87 points less than the average decathlete. Eventually, all these differences across the 10 events should sum up to 240 points, the difference between Eaton’s WR and the score of the hypothetical average decathlete.
I use the above method to answer some interesting (well at least for me) questions about the decathlon.
olorin
 
Posts: 707
Joined: Sat Oct 08, 2005 4:32 am

Re: For decathlon lovers: analysing decathlon performances

Postby olorin » Sun Oct 14, 2012 3:27 am

What makes a great decathlete?
So how do you become a superstar in the decathlon? Before this year only seven decathletes scored more than 8,800 points (Sebrle, Dvorak, O’Brien, Thompson, Hingsen, Clay and Nool). Do these seven have extremely good event(s) or maybe the key is consistency throughout the ten events? In order to answer this question I compare their performance to the 14 athletes that are ranked below them in the all time list. First, I divided the performance of all decathletes to two groups. The six events in which he scored the most in comparison to the hypothetical decathlete and the four in which he performed the worse.

For example, while setting his previous WR Sebrle scored 10.64 (-4 compared with the hypothetical decathlete), 8.11 (+87), 15.33 (-14), 2.12 (+47), 47.79 (+8), 13.92 (+14), 47.92(-2), 480 (-57), 70.16 (+73), 4:21.98 (+75). Thus, his top 6 will include the LJ (+87), HJ (+47), 400 (+8), 110h (+14), JT (+73) and 1500 (+75) for a total of +304. The other four events are labelled bottom4 and sum to -77.

I repeat the same procedure for the top 21 decathletes in the all time list (except Eaton) and divided them to three groups: top (the fabulous seven), medium (ranked 8-14) and bottom (ranked 15-21). For each of the three groups I present the average for the top 6 events and the bottom 4.

Group..................Top 6......................Bottom 4..............................Total
Top Seven..............235.4.......................-143.4..................................92.0
Medium 7...............197.1.......................-239.5.................................-42.3
Bottom 7................175.6.......................-288.2................................-112.7
Diff (Top-Med)..........38.3........................134.3..................................168.6
Diff (Top-Bott)..........59.9........................144.9..................................204.7

The table gives a very clear answer. Prior to Eaton the road to greatness was mainly through limiting your loses in your weak events rather than excelling in a few selective events. For example, the overall difference between the Top and Median groups is 168 points. More than two thirds of this difference is related to the performance of the decathletes in their bottom 4 events. The same is true when we compare the performances of the Top and Bottom groups. In statistical terms the correlation between the Bottom 4 events and the overall performance is 0.85 whereas the correlation between the top 6 events and the overall performance is only 0.39.
olorin
 
Posts: 707
Joined: Sat Oct 08, 2005 4:32 am

Re: For decathlon lovers: analysing decathlon performances

Postby olorin » Sun Oct 14, 2012 3:34 am

So what about Eaton (or why is Eaton so good?)
Same information with Eaton:
Group.................Top 6......................Bottom 4..............................Total
Top Seven............235.4.......................-143.4..................................92.0
Medium 7.............197.1.......................-239.5.................................-42.3
Bottom 7..............175.6.......................-288.2................................-112.7
Eaton (11).............305.........................-375...................................-70
Eaton (12).............546.........................-306................................... 240

In 2011 Eaton emerges as a potential GOAT in the decathlon. Unlike Sebrle, Dvorak and company his performances were not base on damage control in his weak events, but on excelling on large number of events. His score in 2011 in his top 6 events (305) was already the fourth best ever and two points above Sebrle when he set the previous WR. This year he improve this top 6 score by no less than 241 points to take set a new level of 546. This result is 236 better than the second place decathlete.

But the event is decathlon and not Hextathlon so why should anyone care except statistical geeks like myself. In other word what does it matter which way Eaton achieved his WR. In order to answer this question lets go back to 2011. In his Eugene performance Eaton scored 8,729 points. Assume that someone makes the bold prediction that Eaton will score 9,200+ points within three years. Surely most fans would disregard it as overenthusiastic prediction like to one that suggested that Felix can break the WR in 200 this season or that SAFP can go sub 21.8. Then, would continue and predict that Eaton will improve by 250 points both of his strong six events and bottom 4.

Of these two predictions the improvement by 250 points of his bottom 4 events seems much easier. Improving his bottom 4 by 250 points means that Eaton will be 125 points below our hypothetical decathlete. This has been done on the past by four other decathletes: Sebrle (-77), Dvorak (-85), Thompson (-117) and O’Brien (-124) with Hingsen (-132) not far behind. The other part of the equation - improving by 250 points on his strong events - seems much harder as that will take him 240 points above anyone else.

In Eugene 2012 Eaton achieved the harder part of this “prediction”. He also improve his bottom 4 by 70 points. In London he further improved his bottom 4 by additional 70 points. A good DT (~46m) would have seen him getting very close to the require mark. Eaton is so good because in order to take the event to a whole new level he doesn’t need to break new ground, he simply need to do what other top decathletes did before him. A modest set of marks of: 14.75 SP, 2.08 HJ, 46.0 DT and 62 JT will see Eaton threatening the 9,200 barrier each time he competes.
olorin
 
Posts: 707
Joined: Sat Oct 08, 2005 4:32 am

Re: For decathlon lovers: analysing decathlon performances

Postby Marlow » Sun Oct 14, 2012 6:07 am

olorin wrote:Eaton is so good because in order to take the event to a whole new level he doesn’t need to break new ground, he simply need to do what other top decathletes did before him. A modest set of marks of: 14.75 SP, 2.08 HJ, 46.0 DT and 62 JT will see Eaton threatening the 9,200 barrier each time he competes.

Not sure what the the up-side is for his HJ, but I totally agree that his throws can and will improve, both from a technical and raw strength standpoint. I also see a 9200 in his future (though certainly not on a regular basis; it's just too hard to string ALL the events together every time).
Marlow
 
Posts: 21079
Joined: Thu Jan 24, 2008 5:00 pm
Location: Somewhere over the . . . hill

Re: For decathlon lovers: analysing decathlon performances

Postby 26mi235 » Sun Oct 14, 2012 8:33 am

One slight weakness in the analysis is that the assumption of trying to maximize the score in each event during the competition is actually not quite right. The event exception is, not surprisingly, the 1500. Often, the athlete will be far enough ahead or secure enough in the position that they are in (e.g., second), that they run 'adequately' in that event. Only when they really have to push (e.g., Eaton to get the WR) will they dig as deep as they are capable.
26mi235
 
Posts: 16315
Joined: Sat Oct 08, 2005 4:32 am
Location: Madison, WI

Re: For decathlon lovers: analysing decathlon performances

Postby DecFan » Sun Oct 14, 2012 2:21 pm

Thanks, olorin, interesting stuff. 26mi235 is right that the unusual nature of the 1500 distorts the analysis a bit. There's no easy way to adjust for that that I can think of, however.
DecFan
 
Posts: 1409
Joined: Sat Oct 08, 2005 4:32 am

Re: For decathlon lovers: analysing decathlon performances

Postby olorin » Mon Oct 15, 2012 1:47 am

26mi235 wrote:One slight weakness in the analysis is that the assumption of trying to maximize the score in each event during the competition is actually not quite right. The event exception is, not surprisingly, the 1500. Often, the athlete will be far enough ahead or secure enough in the position that they are in (e.g., second), that they run 'adequately' in that event. Only when they really have to push (e.g., Eaton to get the WR) will they dig as deep as they are capable.


There are two questions here:
1. Is the result for the average decathlete is biased because not all decathletes put their "A game" when running the 1500?
2. When analysing an individual performance, is there a bias because of lack of effort in the 1,500?

For the first question the answer is that there may be a bias but it is not that large. Since I calculated the average based only on the best performance of each decathlete it means that all of the decathletes line up to 1,500 when a PB was on the card. While I agree that a PB is not the same motivation as a WR (or a medal) it still suggest that they are not going to jog to the line. Indeed, most of the decathletes set relatively good score in the 1500 compared with their normal form in the decathlon.

The second question is a real issue that I simply don't know how to solve. I thought about puting each decathlete third best performance in the 1500 as minimum mark in this event. But because I'm not sure about the validity of the method (and because my database is only from 8700+ perfromances) I decided to stick with the actual results.

To illustrate the problem, let’s consider Eaton's performance in London. His top 6 events in London he scored +308 (third best ever). But these events include the 1,500 in which he lost 2 points on the average decathlete. If I used his third best performance (4:20.56) then the 1,500 will be +85 and the top 6 +395 which will be second best ever by a clear margin.
olorin
 
Posts: 707
Joined: Sat Oct 08, 2005 4:32 am

Re: For decathlon lovers: analysing decathlon performances

Postby Dave » Mon Oct 15, 2012 6:29 am

Marlow wrote:
olorin wrote:Eaton is so good because in order to take the event to a whole new level he doesn’t need to break new ground, he simply need to do what other top decathletes did before him. A modest set of marks of: 14.75 SP, 2.08 HJ, 46.0 DT and 62 JT will see Eaton threatening the 9,200 barrier each time he competes.

Not sure what the the up-side is for his HJ, but I totally agree that his throws can and will improve, both from a technical and raw strength standpoint. I also see a 9200 in his future (though certainly not on a regular basis; it's just too hard to string ALL the events together every time).


Given that he has gone 2.11 in the HJ, 14.45 SP, he can really focus on two events to get to olorin's base.

Who knows what he is doing with the PV. A 5.30 vaulter with 10.2 speed and 8.23 LJ has a lot more potential than 5.30.

I do think it will be interesting to see if he puts the mark out of reach for a generation at a meet like the Texas Relays where the winds tend to be quite friendly(if not overly so).
Dave
 
Posts: 2119
Joined: Sun Aug 20, 2006 3:16 pm

Re: For decathlon lovers: analysing decathlon performances

Postby tgs3 » Thu Oct 18, 2012 5:34 am

Marlow wrote:Not sure what the the up-side is for his HJ, but I totally agree that his throws can and will improve, both from a technical and raw strength standpoint. I also see a 9200 in his future (though certainly not on a regular basis; it's just too hard to string ALL the events together every time).


While I certainly think Eaton is capable of 9200, I also think people expect a bit too much out of him. Not counting people that did it in the last few years, there have been 25 athletes to score 8500 or more before they turned 25 (Eaton turns 25 in January). More than half of these, 14, never improved upon their best score after turning 25. Only 5 of them (Seberle, Dvorak, Hardee, Nool and Degtyaryov) improved by more than 100 points.
tgs3
 
Posts: 209
Joined: Sun Jul 25, 2010 5:19 am

Re: For decathlon lovers: analysing decathlon performances

Postby Marlow » Thu Oct 18, 2012 5:57 am

tgs3 wrote:While I certainly think Eaton is capable of 9200, I also think people expect a bit too much out of him.

I could agree that his 100, LJ, and 1500 are maxed out, but I think he's nowhere near his max in the throws. The PV also gets better with experience. The HJ and 110H are probably near max (tho his hurdles could still surprise us). 46.x is all we should expect henceforth in the 400. The question therefore is: can he re-approach the speed he showed in the OT 100/LJ? That would take some great conditions (despite the conditions in Eugene).

If he stays healthy, expect at least 9200 by Rio, but that's probably the end of the improvement curve.
Marlow
 
Posts: 21079
Joined: Thu Jan 24, 2008 5:00 pm
Location: Somewhere over the . . . hill

Re: For decathlon lovers: analysing decathlon performances

Postby olorin » Fri Oct 19, 2012 6:00 pm

Marlow wrote:I could agree that his 100, LJ, and 1500 are maxed out, but I think he's nowhere near his max in the throws. The PV also gets better with experience. The HJ and 110H are probably near max (tho his hurdles could still surprise us). 46.x is all we should expect henceforth in the 400. The question therefore is: can he re-approach the speed he showed in the OT 100/LJ? That would take some great conditions.

If he stays healthy, expect at least 9200 by Rio, but that's probably the end of the improvement curve.


Was Eugene a one off?

In Eugene Eaton achieved two wonderful marks in the LJ and the 1,500. Each of these marks gave him more than 100 points compared with the average hypothetical decathlete. Was this the special feature of his WR performance or was it the series of good performances ?
In order to answer this question I examine all performances above 8,700 points (total of 55 performances). For each performance I calculate the top 2 events and the events ranked 3-6. The table below present the top 5 in each category:
...................Top 2........................................Top 3-6
1. Eaton........+245....................................1. Eaton...........+301
2. Nool..........+234....................................2. Sebrle..........+142
3. Clay..........+219....................................3. O’Brien.........+136
4. Hingsen......+198....................................4. Clay. ...........+105
5. Dvorak .......+181....................................5. Hämäläinen....+102

While the Top 2 performance in Eugene is ranked best all time, it is not that different from other greats in the past. In his WR Eaton scored 245 points above the average decathlete. Nool in second place is only 11 points behind, whereas Clay in third is only 26 points behind. Hingsen and Dvorak are not that far off the pace and trailing Eaton only by ~50 points.

In contrast, Eaton performances in events 3-6 are by far superior to anything we have seen in the past. His score (+301) is more than double of that of the second place decathlete Sebrle. To put it in perspective, the difference between Eaton and Sebrle is larger than the difference between Sebrle and the Apaychev (that score +14) that is in ranked last place among all 8,700+ decathletes.

Thus, Eaton performances are extremely good in the top 2 events his real talent is to be able to perform is very high level across six events.
This is exactly why Eaton is so good. He doesn’t need 8.23 or sub 10.2 in order to get to 9,200. As I pointed out before he only needs to improve his weak events. For example:
10.28-8.10-14.8-2.08-46.5-13.56-46.5-5.30-63.5-4:18 will see him over the 9,200 barrier.

Surely, there will be times when Eaton will have a bad event that will derail him to the low 9,000. But every time he will get it right he will be on the verge of something special and unlike Marlow I do not think that 9,200 is necessarily where he will stop.
olorin
 
Posts: 707
Joined: Sat Oct 08, 2005 4:32 am

Re: For decathlon lovers: analysing decathlon performances

Postby Dave » Sat Oct 20, 2012 9:38 am

Any thoughts of applying this same analysis to the Heptathlon?
Dave
 
Posts: 2119
Joined: Sun Aug 20, 2006 3:16 pm

Re: For decathlon lovers: analysing decathlon performances

Postby olorin » Sun Oct 21, 2012 1:30 am

Dave wrote:Any thoughts of applying this same analysis to the Heptathlon?


Yes, but not in the near future. The most consuming (and boring) part is the data collection. So, unless someone can provide me with the data (in excel format) I will stick to decthlon for now. Dvorak is next on my list.
olorin
 
Posts: 707
Joined: Sat Oct 08, 2005 4:32 am


Who is online

Users browsing this forum: Bing [Bot], El Toro and 9 guests