OK, as promised here is a little analytic loop that I mined out of my Triathlon Race Database. I wanted to see how competitive I was over time and one of the basic ways of doing so is to look at the USAT ratings I've achieved in my races over time. The USAT has employed two main rating systems over the 19-year span of my triathlon racing. The old system assigned a maximum rating points for a race based on the length of the race and the number of participants. As I recall, a local sprint race awarded the winner 90 points, whereas a local half-IM distance awared a max of 92 points. Larger races had higher possible points as well (I don't remember the details). In the case of the 90 point local sprint, if one finished 5% slower than the winner you'd get a 85.7 rating for the race (1/1.05 X 90).
The new USAT system adopted in 2006 uses what's called a "pace-setter"approach. Basically for any given race, they take all participants who had a prior rating (who raced at least 3 times the prior year) and assigns an expected or par score for the race. Then, every person racing in this season's event gets a relative score to the par. For example if there was just one pace setter who had an 80 rating and you raced the same time as that pace-setter, you'd score an 80. It's a bit like the NASTAR system where a know skier (one who has a rating) skies a course and establishes the par and everyone else is compared to that par. The difference here is there isn't one pace-setter in a race but as many as who had a prior year's rating.
In any event, I've recorded ratings with the Old system up through mid-2014 and have the new system for 2006 on. I should also note that the USAT only rates USAT sanctioned races with the New system (so no ratings for foreign races or for local non-sanctioned races). As it turns out I have 142 races rated with the Old system and 102 with the New (current) system. Many of my races between 2006-2014 have two ratings (Old and New). My averages across all races, over time are:
Old Rating System: 76.5
New Rating System: 76.6
For just the 88 races where I have both Old and New ratings the averages are:
Old Rating System: 76.4
New Rating System: 77.4
So it seems that Ratings for the average race are slightly higher with the New system. However, in looking through the Database, it appears that there is more variability in the Old system (probably because it was all driven off the performance of only one pace-setter--the winner). Since the annual rankings are driven not by an average of all races, but just the top 3, in practice the Old system yielded the highest (and lowest) individual race ratings and a higher annual rating to determine year-long rankings.
OK, enough of that. How have my ratings changed over time?
This first chart below shows the average (Old and New, or as listed "Current") for ALL of my races each year. The first data point is the combined 2001/2002 season and the last is for my "come-back" years of 2016-2019 (no 2015 rating as I was recovering from my back issues). A couple of observations:
-I peaked in 2007/2008. This seems pretty clear as you will see in many subsequent charts. That was the year (2007) I worked with Peter Reid and was the only year I was an All-American (back then it meant I was in the top 5% of my Age Group in the country). I raced my first race at the age off 44 and improved for the first 6-7 years, reaching my peak at the age of 50.
-Since my peak I seem to have fallen about 9% on average over the last 11/12 years with a sharper decline after 2013, when I was 56.
The new USAT system adopted in 2006 uses what's called a "pace-setter"approach. Basically for any given race, they take all participants who had a prior rating (who raced at least 3 times the prior year) and assigns an expected or par score for the race. Then, every person racing in this season's event gets a relative score to the par. For example if there was just one pace setter who had an 80 rating and you raced the same time as that pace-setter, you'd score an 80. It's a bit like the NASTAR system where a know skier (one who has a rating) skies a course and establishes the par and everyone else is compared to that par. The difference here is there isn't one pace-setter in a race but as many as who had a prior year's rating.
In any event, I've recorded ratings with the Old system up through mid-2014 and have the new system for 2006 on. I should also note that the USAT only rates USAT sanctioned races with the New system (so no ratings for foreign races or for local non-sanctioned races). As it turns out I have 142 races rated with the Old system and 102 with the New (current) system. Many of my races between 2006-2014 have two ratings (Old and New). My averages across all races, over time are:
Old Rating System: 76.5
New Rating System: 76.6
For just the 88 races where I have both Old and New ratings the averages are:
Old Rating System: 76.4
New Rating System: 77.4
So it seems that Ratings for the average race are slightly higher with the New system. However, in looking through the Database, it appears that there is more variability in the Old system (probably because it was all driven off the performance of only one pace-setter--the winner). Since the annual rankings are driven not by an average of all races, but just the top 3, in practice the Old system yielded the highest (and lowest) individual race ratings and a higher annual rating to determine year-long rankings.
OK, enough of that. How have my ratings changed over time?
This first chart below shows the average (Old and New, or as listed "Current") for ALL of my races each year. The first data point is the combined 2001/2002 season and the last is for my "come-back" years of 2016-2019 (no 2015 rating as I was recovering from my back issues). A couple of observations:
-I peaked in 2007/2008. This seems pretty clear as you will see in many subsequent charts. That was the year (2007) I worked with Peter Reid and was the only year I was an All-American (back then it meant I was in the top 5% of my Age Group in the country). I raced my first race at the age off 44 and improved for the first 6-7 years, reaching my peak at the age of 50.
-Since my peak I seem to have fallen about 9% on average over the last 11/12 years with a sharper decline after 2013, when I was 56.
The second chart below shows the more familiar Official USAT annual Rating--this number is calculated using just the three best races each year. Observations:
-You can see that the Old system yields a higher Official Rating due to the higher variability in rating scores as compared to the current system.
-The peak in 2007 is very evident
-The decline is my official rating is less...about 7% over the last 12 years
Below are my top 10 races under both Rating systems:
-The 2007 White Lake Half-IM was clearly my best race. In the race I (and Anders) qualified for the USAT National Long Course team. I set my Half/70.3 PR of 4:43 as I rode at 23.2mph (2:24 for 56 miles) and ran 7:29 miles (1:38 half marathon)
-I finished first in my Age Group beating Louis Almekindres, who at the time was ranked 4th in my Age Group in the United States
-Note the preponderance of 2007 races (half in the New system)
-Another key performance for me was SkipJack 2011 (this race was 1.2/66/10, so not an "official" Half-IM). In that race I finished 2nd overall, getting passed with just 3 miles to go. Shortly after that race I qualified for Kona at IM Arizona in 2011. SkipJack will almost certainly be the latest race that I ever crack my personal top 10 in.
This next chart shows the distribution of my ratings over time (here I use the New system when I have both and either the Old or New, when I have just one. My most common rating has been 77.0-77.9:
Another useful way to look at relative performance is to track my percentile ranking against everyone in the race and more narrowly relative to just my Age Group. I can also do this for each discipline (Swim, Bike, Run) as well as for the Overall result. Because the mix of races I race affects these average annual %tiles (I have a lower %tile ranking at Kona than I would at a local Sprint for instance), I measure %tiles for all races as well as just for "local races" (local races exclude IM, IM70.3, World and National Championships, as well as races with more than 1,000 finishers):
Observations:
-The 2nd chart of Local races only I feel gives a better comparison over time. For example, in the top chart for all races, 2009 and 2013 seem to be better years for me but these averages are relatively inflated since I did primarily local races in those years. You can see they were relatively less highly rated when we compare more like-to-like Local races in the 2nd chart.
-My bike is my clear strength. It was from the very beginning and I've averaged in the top 10% of finishers in local races each year throughout my career.
-My swim showed strong improvement up through about 2010/2011 and has declined a bit since. I've definitely deemphasized swim training, especially over the last few years and this data would seem to indicate that I might consider more swimming investment this year.
-My run improved as well up through 2007 and has been in decline since with a very rapid fall off over the last 5 years--I now am lucky to finish in the 50 %tile in a run these days.
-Overall, I've averaged in the top 20% Overall and top 10% in Local races for most of my career, with a steady, run-driven decline over the last 5-8 years.
The same charts can be generated for my annual %tile averages against just my Age Group competitors:
-The same basic general conclusions can be drawn from these two Age Group charts as was observed with the Overall %tile charts
-My Age Group bike performance is even more striking--basically I've been in the top 5% for just about all of my career. From 2007 through mid-2009, I was first in my Age Group on the bike is 17 consecutive races
-The decline in my run is, if anything, more striking and is the biggest change in my relative competitiveness over the most recent timeframe.
-It looks like I'm having a bit of a nice comeback after 2014/2015 and my back problems.
OK, that's enough for now. More later!
No comments:
Post a Comment