Four different algorithms, giving different race predictions for my 10 km tomorrow. Which will be closer?
I can’t believe it’s Saturday again, already. It was only a week ago I was preparing for my last race and wondering what my first race of the season would hold. This is the first time ever that I have run races on consecutive weekends, and I’m not 100% sure my legs are completely ready – they certainly did not feel like it on my final pre-race run this morning. Though, I suspect that another 24 hours and all the buzz of a race will wake them up.
The Valentine’s 10k is a 10 km road race organised by the Norfolk Gazelles. The route starts at Easton College, Norfolk, and heads out into the countryside, taking in the villages of Colton and Marlingford before returning to the college. Similar to last week, I have not run this event before. However, I did run the Larking Gowen City of Norwich Half Marathon around the same area a good number of years back, and whilst the area is largely flat, I do seem to recall that there were a few hills dotted about.
As one might expect this week has been a light week for training, all courtesy of TrainAsONE. Some 30 minute economy (easy) runs, an Interval session (the 3 minute repeat kind, with 3 minutes fast and 3 minutes of rest). Finishing off the week with 12 minute easy runs on the last two days.
My Race Predictions
So what are my predictions for tomorrow?
|Algorithm||Pace (min/km)||Time (mm:ss)|
|Riegel (6 min Assessment)||4:21||43:30|
|Riegel (3.2 km Assessment)||4:25||44:10|
|Riegel (10 mile Race)||4:10||41:40|
So following on from last week’s race predictions, I have carried out calculations using Riegel’s formula on my recent 6 minute and 3.2 km assessments. Additionally, given that I now have a recent race (i.e. last weekend’s 10 mile), I have used Riegel’s formula on this event. As this is applying the Riegel formula a little more in line with its intention, one would surmise that it might be more accurate than the values based on the assessment runs. This is further supported as both the assessments predict paces slower for tomorrow’s 10 km than I raced the 10 mile last week. In addition to the Riegel calculations, TrainAsONE (TAO) has provided its prediction.
So will I do nearer a 4:10 pace, i.e. the 10 mile based Riegel prediction, or the 4:15 pace, i.e. TrainAsONE’s calculation?
Who will be closer? Reigel’s 4:10, or TrainAsONE’s 4:15?
I think the box plots are interesting this week. Concentrating solely on the 10 mile Riegel and the TrainAsONE prediction. The interquartile range (the box) is slightly larger for TAO, whereas the whiskers are slightly shorter.
I was discussing this observation with a research associate until the early hours of this morning… It raises lots of questions and lots of avenues for further investigation and exploration. These result from how the data is gathered and predictions made. Whilst on the surface the errors (i.e. the box / whisker sizes) are allowing direct comparison of the algorithms as they are drawn from the same dataset, many aspects mean this is not completely or necessarily true. And what could be most interesting is that TrainAsONE is picking up on the negative effects of a race only 7 days prior, lowering race pace and making prediction itself more difficult. All a working hypothesis to be explored at a later date.
Back to tomorrow’s race. Last week I targeted the faster pace prediction, which was TrainAsONE’s. So this week I’m thinking that it seems only right to aim the faster pace again, this time being Riegel’s (based on the 10 mile race). Which will be closer?
4:10 here we come!
I’ll let you know how I get on.