Well, this hasn't been going the way I wanted since the end of January...

The last I posted was right after the New Bedford Half Marathon. I was beat up from that race but feeling optimistic based on the 1:20:50 result. Since then two weeks have past and March has come to an end. Only a month and a half now remains before I put my toe on the line for the Maine Coast Marathon and all I can think about is how far below my targets my training has fallen since January.

I’ll start with the quick review of my training since my last post:

3/18/13 – 3/24/13

• Monday: 3.06 miles in 22:55. Easy recovery run around Spy Pond in Arlington.
• Tuesday: rest.
• Wednesday: rest.
• Thursday: rest.
• Friday: 7.52 miles in 59:00. I ran down the bike path and around Fresh Pond. My right Achilles and my left calf were still in agony for the entire run. After the run, I noticed that the sole of my Adios had completely worn off the outside of my strike area. I guessed that this meant that I was torqueing my ankle with each landing so I decided to retire the Adios and get some new trainers to keep going in.
• Saturday: 10.00 miles in 1:17:51. I ran to City Sports in Porter Square in my old NB880s (2.5 miles) and bought new Brooks Cadence (1^st generation because they were on sale for $59 and change) and did the 5 mile run club route then ran home in the Cadence. They felt a little better than my end-of-life Adios but the damage had been done by racing New Bedford in dead shoes so my Achilles still wasn’t elated with the idea of running.
• Sunday: rest due to Achilles pain.

3/25/13 – 3/31/13

• Monday: 10.17 miles in 1:19:01. I ran this one on Mass Ave, the bike path and Fresh Pond. I still wasn’t feeling 100% but I could tell that my legs were healing up now that they were in new shoes.
• Tuesday: 10.17 miles in 1:18:07. This was the exact same route as Monday but a little bit stronger.
• Wednesday: 10.17 miles in 1:16:04…in keeping with the trend for the week.
• Thursday: rest. Sarah and I went out to the Painted Burro in Somerville; it was pretty good but nothing to write a post about.
• Friday: rest. I thought about doing the hill workout that had been scheduled for Thursday but was le tired so I took a nap.
• Saturday: 10.00 miles in 1:11:28. Started with an easy 2.5 miles to Porter Square where I hooked up with the City Sports Run Club for a Marathon Pace 5 miler (I ended up running the 5 in more like 6:40 than 6:30 but it was my first real MP run off the treadmill and first workout since New Bedford, so I’ll take it). After the 5 with the Run Club I ran easy for the last 2.5 home. It was super nice out so in the afternoon, Sarah and I walked from Arlington Center through Harvard Square, along the Charles and over to Copley Square with a stop for beers in Cambridge, coffee and shopping on Newbury Street and Sushi on Boylston Street. It was a really nice first day that felt like spring!
• Sunday: 16.44 miles in 2:08:15. I ranged all over on this one. I started heading north on route 3 until I ran out of sidewalk in Woburn and did a loop through the Wholefoods parking lot before coming back south on route 3 until Wildwood Street which I followed to Church Street in Winchester. Church took me to Main Street through Winchester Center. I was completely lost and went north on Main Street until I saw a sign for 38 South towards Somerville. I took 38 South past the Oak Grove Cemetery to the Mystic Valley Parkway in Medford which I followed through Somerville and a bit of Cambridge to get back to Mass Ave. I took Mass Ave back to my apartment but only had about 1:46:00 on my watch when I got home so I decided to add on the 3.06 mile loop around Spy Pond which added Belmont to my list of cities that I passes through on this run.

March as a whole was a difficult month to characterize so I invented some math to compare it to other months. My gut feeling was that it was a particularly bad month of training due to all of the time I had to take off because of the Achilles problems. I ran 194.78 miles in 24:10:22 this month which is definitely up in distance since February (146.72 miles) but my pace was down from 7:10 min per mile to 7:27 min per mile.

I decided that there are three things that are important to any runner’s training: volume, intensity and frequency. I defined volume as the average distance per run (monthly mileage divided by the number of runs during the month). I defied intensity as the equivalent age graded percentage of the average run distance and time (monthly mileage divided by number of runs during the month, monthly time divided by number of runs during the month, age during the month all put into a age graded calculator to get a decimal from 0.00-1.00). I defined frequency as the percentage of days during the month that a run was completed (total number of runs divided by total number of days; this is expressed as a decimal but can exceed 1.00 if you are running doubles etc).

These three variables (volume, intensity, and frequency) are combined by taking the geometric mean to get a raw training score. The geometric mean is used because it allows you to calculate a meaningful average for multiple factors on different scales. The raw score will typically fall between 0.00 (for a runner who does not complete any run in a given month) to 2.971 for a runner who runs a World Record marathon every day during the month. As the quality of workout clearly does not scale linearly between these two boundary conditions, I raise e to the power of the raw training score to get what I’m calling (for lack of a good name) the ζ-value (zeta). The ζ-value ranges from 1.00 for a runner who does not complete a run to 19.5 for the runner who runs a world record marathon every day.

I went back and calculated my historical ζ-values since I started logging my running in January of 2010 (I actually started in December 2009, but January 2010 was my first month with the entire month logged).

zeta-value Control Chart...I need to clean this thing up...

 

The SPC control chart of my running over this 39 month period shows that my training has been very poorly controlled. My month-to-month 1σ non-uniformity is ~22.6%. The good news is that 12 of my past 13 months have been above average. The one month in that set of 13 months that fell below the SPC chart centerline was November 2012 and I spent two weeks of that month in South America with Sarah on our honeymoon. Also in November, I ran a marathon on the 4^th and then basically took the rest of the month off.

According to my ζ-value, this past March was my 4^th best month of training. While I don’t feel like it was my fourth best month, it does help to explain how I ran a 2:31 PR at the Half Marathon on a swollen Achilles. Looking at my moving average, I think I should be on track for some more PRs:

6 Month Moving Average of zeta-value since January 2010.

 

So March wasn’t everything that I wanted it to be, but the math suggests that if I keep working in April and get my ζ-value for the month over 5.0 then I should be in good shape for the marathon.

I should note that the failure of the ζ-value as a quantification of training is that average distance is incorporated into the result twice: once directly in volume and once indirectly in intensity. Also distance can very on a much larger range than either frequency or intensity. Were a runner to run one mile at World Record pace every day for a month, the equivalent zeta value would be ~2.72. This suggests that my current training is better than the theoretical mile world record every day. I’m still working on a model that will better account for this, but for the time being I am okay with my ζ-value model as it stands as far as marathon training goes. Due to the demands of the marathon for high volume training, the type of training that I am doing right now is probably more effective at preparing for a marathon than running a single mile every day even at World Record pace.

I’m also thinking about how to merge total elevation gain into this model so that there is some sensitivity to hill workouts. The drawback is that almost all of my runs start and end at my front door so my net elevation change while running over the course of the year is essentially zero and because of this I do not track elevation in my log. Lacking data to manipulate, it is difficult to develop a model that uses elevation as a parameter. Confounding this, I would need to track total elevation gain and total elevation loss separately as well as grade because clearly running up a short, steep hill is different than running down a long gentle decline.

Aside from this weekly/monthly update I am due for a postmortem on my Adidas Adios and an introduction of my Brooks Cadence. We’ll see when I actually get a chance to write those up though.

Here's another picture of a graph that I'm inserting here so that I have a URL for it elsewhere. It is a natural log regression of my racing speed as a function of distance for my recent races.

 

Happy running!

11 weeks down and just 6 weeks remaining until race day!!!

 

May your training miles be ever hilly and may all your races have free beer!