Wednesday, September 28, 2016

Sigh. So Much for That.

Well, folks, I think it's clear the model for prediction did not work.

And to be honest, I am thinking this is an exercise in futility because the BAA accepted 818 fewer qualifiers. Given the BAA can and will decide to accept any variation of 23,000 to 24,000, the prediction can never be close to accurate, even if we built in a whole bunch more feeders and demand predictions.

Over at bqrun.com, he has an even more complicated model with 6 years of data. His last update came up with 48 seconds. Still quite far from 2:09.

2957 applicants where rejected. At 129 seconds of cut off, that is 23 runners per second. With 818 fewer spots than last year, you have about 36 seconds of additional cut off. Apples to applies with entry spots the same year over year, the cut off would have been 1:33. So had they accepted the same number as last year, it would have been around 1:33.

All I can say is thank you for reading, and I'm sorry if I gave you false hope. I'm not sure if I'm gonna do this again for 2018 given the uncertainty with the number of qualifiers the BAA will accept. With the fact that they don't publish the data until the day of the cutoff announcement, makes it really hard to come up with a good number.

Sunday, September 11, 2016

Last 2017 Feeders: Lehigh Valley and Erie Marathon

We're at the end of the list of marathons I have in my list of feeders for the analysis. Lehigh Valley and Erie. And they followed the trend of most of the marathons this year. Making this update somewhat anti-climactic.

Lehigh Valley has the notoriety of being the race that had no intermediate mats in 2013, allowing what is the most fascinating story of probable marathon cheating I've seen in my lifetime (not so much the cheating part but the behavior of the accused cheater). After 2013, the race management added several timing mats along the course.



Erie at Presque Isle is a super flat, double loop course on a peninsula that usually has good weather (though can be windy) and a high number of qualifiers.


Lehigh Valley


AG Group2016 Qualifiers2016 AG TotalPercentage2017 Qualifiers2017 AG TotalPercentage
F18-342914819.59%3515722.29%
F35-39117315.07%95815.52%
F40-44186726.87%106814.71%
F45-4994918.37%94420.45%
F50-54113729.73%53514.29%
F55-5951729.41%32015.00%
F60-6411010.00%1425.00%
F65-692633.33%1425.00%
F70-74000.00%010.00%
F75-79000.00%000.00%
F80+000.00%000.00%
M18-343313724.09%131717.60%
M35-39199719.59%4805.00%
M40-44178420.24%5776.49%
M45-49237829.49%139114.29%
M50-54106515.38%137417.57%
M55-59123732.43%135523.64%
M60-6442317.39%62128.57%
M65-692728.57%41723.53%
M70-74020.00%1333.33%
M75-79010.00%010.00%
M80+000.00%000.00%
Totals20693821.96%14598114.78%

Big drop: nearly 30% fewer qualifiers.

Margin2016Percentage2017Percentage
<1 minute188.74%1611.03%
1-2 minutes167.77%1913.10%
2-3 minutes167.77%106.90%
3-4 minutes178.25%128.28%
4-5 minutes2512.14%53.45%
5-10 minutes4823.30%3826.21%
10-20 minutes4220.39%3624.83%
20> minutes2411.65%96.21%
Totals206145

My goodness, the Squeaker Pack in this race is huge, both years. 44.66% vs 42.76%. Lots of folks eking in under 2 minutes this year (I'm guessing the weather might have had something to do with it).

Erie Marathon


AG Group2016 Qualifiers2016 AG TotalPercentage2017 Qualifiers2017 AG TotalPercentage
F18-345521725.35%5619528.72%
F35-394312035.83%278531.76%
F40-443413325.56%3410632.08%
F45-494410442.31%3310033.00%
F50-54246934.78%235938.98%
F55-5972825.00%52619.23%
F60-6461931.58%72133.33%
F65-694757.14%1425.00%
F70-74020.00%030.00%
F75-79000.00%000.00%
F80+000.00%000.00%
M18-344816828.57%5818132.04%
M35-393212326.02%1710017.00%
M40-442811225.00%169816.33%
M45-494511937.82%5112540.80%
M50-545111743.59%3411928.57%
M55-59368144.44%257832.05%
M60-64185632.14%124228.57%
M65-69153148.39%93129.03%
M70-7431030.00%2540.00%
M75-79000.00%010.00%
M80+000.00%000.00%
Totals493151632.52%410137929.73%

16.84% fewer qualifiers.

Margin2016Percentage2017Percentage
<1 minute448.92%317.56%
1-2 minutes5511.16%276.59%
2-3 minutes6312.78%348.29%
3-4 minutes367.30%338.05%
4-5 minutes275.48%245.85%
5-10 minutes11322.92%12931.46%
10-20 minutes10320.89%9222.44%
20> minutes5210.55%409.76%
Totals493410

Squeaker Pack: 45.64% vs 36.34%.

These races definitely have a last minute feel just looking at the under 5 minute margins. I also wonder if people decided they didn't need to try to extend their margins and DNS'd as the cut off is looking much better than last year.


Final Totals


AG Group2016 Qualifiers20156 AG TotalPercentage2017 Qualifiers2017 AG TotalPercentage
F18-3453414200112.72%45344083811.10%
F35-3924661689614.60%22091710312.92%
F40-4424341607515.14%20781577713.17%
F45-4924791172621.14%21701220617.78%
F50-541548773220.02%1462825817.70%
F55-59816388521.00%721399318.06%
F60-64362161122.47%366173021.16%
F65-6910951321.25%11955621.40%
F70-742114015.00%2116812.50%
F75-7953016.67%1224.55%
F80+1425.00%41040.00%
M18-3448113750512.83%38253569910.71%
M35-3923941984012.07%19701914910.29%
M40-4427292152012.68%22512036411.05%
M45-4932861824718.01%27691828915.14%
M50-5425751453417.72%21501456714.76%
M55-591705889419.17%1489927116.06%
M60-641023473621.60%964487519.77%
M65-69465197423.56%439211420.77%
M70-7414165521.53%11465817.33%
M75-792517314.45%3018815.96%
M80+73917.95%104323.26%
Totals3474322873015.19%2969622587813.15%

14.53% decrease in qualifiers.
13.45% decrease in the rate of qualification.

Margin2016Percentage2017Percentage
<1 minute19555.63%16935.70%
1-2 minutes19575.63%17015.73%
2-3 minutes19715.67%16325.50%
3-4 minutes17735.10%15075.07%
4-5 minutes16674.80%13844.66%
5-10 minutes732721.09%640521.57%
10-20 minutes968527.88%822227.69%
20> minutes840824.20%715224.08%
Totals3474329696

Squeaker Pack: 26.83% vs 26.66%

Pretty much the same year over year.

Final bit of data: for the 26 races, taking all the finisher for the 2016 qualification window achieving 2:28 or better, we get a total number of 29689

29689 > 29696

The total number of 2017 qualifiers regardless of margin, is still less. This implies no cutoff, and anyone achieving their age group standard will be in.

In reality?

Really hard to say. The buffer in this calculation/model is at 173 which is a very small percentage of the whole pie. Easily gobbled up by lightly more demand.

The other interesting theory I've heard is that last year was the 120th running and, therefore, in addition to high qualification numbers, there was higher demand. So the demand of last year might be mucking up the numbers a bit.

My personal feeling is that if you have 1:00 or more, you have a pretty good chance. If you have more than last year's margin, you're almost certainly in (and, boy do I hope that's true because I have only 9 seconds more than last year's cutoff).

Now we* begin the agonizing wait 2+ week wait.

* we is really the "Squeakers" - there is near zero chance it doesn't open up for a second week of registration