Run-Down Forums

[Conway]

Wow ... We may want to go straight IAAF scale for the purchase points ... As much as Iwould love to buy Maurice for that amount,thta is way out of wack ... Mo and Chambers become total bargains while Obikwelu is severely over priced ...

Went through the women's sprints on the scale .. They take somewhat of a hit becasue theri records are so tough .. Harder for them to garner points than the men ... But it seems fairly good within their own scope though ... So is very usable ...

[Dan]

[Dan]

Let's see, this may not be too bad to come up with for the women. Using the lists at:

http://www.algonet.se/~pela2/women.htm

we would probably go with the following best marks per event for relative performance comparisons:

100m: Marion Jones 10.65
200m: Marion Jones 21.62
400m: Marie-Jose Perec 48.25
800m: Ana Quirot 1:54.44
1500m: Hassiba Boulmerka 3:55.30
mile: Svetlana Masterkova 4:12.56 (WR, but post cold war...)
3000m: Sonia O'Sullivan 8:21.64
5000m: Olga Yegorova 14:29.32 (EPO?) or Gete Wami 14:30.88
10000m: Ingrid Kristiansen 30:13.74

Would Quirot's time in '89 be considered Eastern bloc due to Cuba's status? Same for the 3:53.96 by Paula Ivan (Romania) in '88 (Seoul)...

It's hard to find someone other than the Chinese on that 1500 list! I had to go down to the 14th spot for Hassiba...

Now let's plug those times in and see what we get... In the following table, I selected the best men's and women's mark of the year and compared the women's marks to the above "records", listing the men's points and the women's before and after points (w1 is relative to the WR, w2 is relative to the second tier records).

============================
100m - 9.82 (m) - 10.82 (w)
(m) 96.57 (w1) 62.41 (w2) 80.19

============================
200m - 19.88 (m) - 22.23 (w)
(m) 70.50 (w1) 64.52 (w2) 66.96

============================
400m - 44.28 (m) - 49.59 (w)
(m) 75.97 (w1) 57.07 (w2) 70.64

================================
800m - 1:42.47 (m) - 1:56.85 (w)
(m) 87.81 (w1) 68.81 (w2) 78.71

=================================
1500m - 3:26.12 (m) - 3:59.35 (w)
(m) 99.49 (w1) 64.34 (w2) 83.43

==============================
3k - 7:30.53 (m) - 8:23.26 (w)
(m) 81.49 (w1) 68.96 (w2) 96.99

================================
5k - 12:56.72 (m) - 14:29.32 (w)
(m) 81.37 (w1) 98.70 (w2) 100.0

================================
10k - 27:04.20 (m) - 30:55.80 (w)
(m) 79.05 (w1) 59.75 (w2) 79.47


It's obviously not practical to try and get the values to match up perfectly for men and women, as they are relative to the performances of the year, which varies greatly by event, but the w2's seem to be a much better approximation. Agree or disagree?

Dan

[Conway]

I agree that the adjusted records give a much better approximation ... What if you went with ..

10.76 - pre FloJo WR
21.72 - pre FloJo WR and equalled by Torrence
48.83 - 84 Olympic result and seemingly free of drug using competitors
1:54.92 - Kazankina's best 800
3:55.74 - avg of Kazankina's 5 top races

and then follow what you have from there ... The problem with Marions sprints is that they were at altitude ... The 400 had the Atlanta track effect ... And I used Kazankina as the middle distance standard as I think that drug use at that time (70's) was more oriented towards the sprints/speed/strength events ... I think these might bring those events in line with the men and give the women a fighting chance at earning similar points ... Does it work ??

[Dan]

Sounds reasonable. Plugging those numbers in (some of which are quite close to the ones I went with in the previous test), we get:

100m: 92.90
200m: 72.19
400m: 83.11
800m: 82.88
1500m: 85.20

That doesn't seem too bad.

Dan

[Conway]

Was just looking for something that would be more real world representative ... And it seems to work relatively well ... What do you think ??

[Dan]

Yeah, I think it's definitely good enough to get us started. If this whole thing goes well enough to do it again next year, then we'll have a much better picture of how well we dialed in all our pricing and scoring estimates.

Dan

[Conway]

I think we are close enough to start ... At least for a trial run type set up ...

[Dan]

We've got enough for a trial run (scoring a practice meet), but I'd like to get a better handle on athlete selections before doing a full test. I'm also still working on putting together a program that will automate all the scoring and display of results/rosters.

Dan

[Conway]

Yes ... How do we want to handle athlete selection ??? How many events ?? How many athletes per event ??? Total number of available purchase points ??? Which events ??? Will have to figure out how to distribute my "money" !!! Focus on speed ?!? Stars ?!? Have we decided on cross over athletes (an athlete belonging to more than one individual)??? And doublers ?!?

Just some tings to think about !!!

[Dan]

My feeling is we set an order of preference as to what needs to be resolved:

1) select the meets and events
2) price the athletes based on #1
3) allow roster selections in any order

As long as we can carry out step #2 in a fashion that accurately reflects the athletes' potential scoring (the biggest "if" I've thrown around in a while ), then I don't see any reason we need to restrict shared athletes across rosters, order of selections, or place many rules on roster makeup.

We should probably require at least one athlete per event for each roster, just to keep things interesting, but the idea of the pricing system theoretically would equalize the value of athletes (sort of like 2 for 1 trades in the NBA where they have to match salaries) and allow you to spread your dollars wherever you deem best. If the pricing is somewhat accurate, then Maurice isn't necessarily the best choice for everyone, so why complicate things further by limiting how many can choose him?

The cash available for purchasing will depend on how many events we choose, which will in turn determine the minimum number of athletes per roster to cover them, and the average cost per athlete.

For event doublers, I think the best route is to make the athlete selections be for a single event (again, considering the 1500 and mile to be the same event; if a meet holds both and we have athletes split over the two, then we'll have to figure something out for places...). If someone thinks Ato is the steal in the 100 and 200, then that's two separate selections. He often seems to have a split personality under pressure.

For meets, obviously the Golden Leagues and maybe a few additional ones to give us a nice even 10. How's this sound for an event list:

men:
100, 200, 1500, 5k, 110h, long jump, pole vault

women:
100, 400, 800, 1500, 3k, high jump, long jump

edit: I decided to yank the women's 5k and replace it with the 800, giving more event variety.

I haven't yet tested the accuracy of the scoring methods for the women's high jump and long jump, so those are yet to be determined... Also, the 3k/5k may not be wise, as they are probably an either/or. I suppose we don't have to have an even number of events for men and women, as the IAAF certainly isn't concerned about such equality... We could drop the 5k and add the men's 400.

Dan

[ This Message was edited by: Dan on 2001-10-22 23:23 ]

[Dan]

Here are a few more comparisons for the women's field events:

comparison 9:
women's LJ WR (7.52m) = 1351.94 pts, top mark of the year (7.12m) = 1213.42 pts ==> 89.8% of WR points
7.12m is comparable to a 45.61 400m (relative to % of WR)
43.18 = 1111.49 pts, 45.61 = 996.91 pts ==> 89.7% of WR points

comparison 10:
women's HJ WR (2.09m) = 1359.52 pts, top mark of the year (2.04m) = 1291.59 pts ==> 95.0% of WR points
2.04m is comparable to a 44.24 400m (relative to % of WR)
43.18 = 1111.49 pts, 44.24 = 1059.96 pts ==> 95.4% of WR points

comparison 11:
women's shot put WR (22.63m) = 1387.78 pts, top mark of the year (20.79m) = 1253.00 pts ==> 90.9% of WR points
20.79m is comparable to a 47.00 400m (relative to % of WR)
43.18 = 1111.49 pts, 44.24 = 936.70 pts ==> 84.3% of WR points

comparison 12:
new event: women's javelin WR (71.54m) = 1281.34 pts, 2nd best mark of the year (69.53m) = 1241.82 pts ==> 96.9% of WR points
2.04m is comparable to a 44.43 400m (relative to % of WR)
43.18 = 1111.49 pts, 44.24 = 1050.99 pts ==> 94.6% of WR points


Observations:
- women's long jump is near perfect
- women's high jump is very close; predicts within half a percent of purdys
- women's javelin predicts too high (2-3%)
- women's shot put predicts much too high (6-7%)

Once again, the throws calculations seem to scale very poorly, but the jumps work quite well. Won't be a problem including the women's high jump and long jump.

Dan

[Dan]

Another thing to consider in a metric event like the high jump where the units of measurement are fairly large relative to the overall distance, the percent separation for our scaled scoring system becomes rather exaggerated. I updated the demo page so you can play around with it and see what you think:

http://run-down.com/fantasy/score.php?

Dan

[Conway]

OK ... I think we look pretty good ... Events look good .. Too bad no men's 400 but can't have eveything ... Looks well rounded the way it is ...

Scores are looking pretty good as well ... Can't think of any other way to work with the field events as there are no real standards to shift to like I did with the running events ...

I guess we need to start looking at values and costs ...

[Dan]

Well, there's no reason we couldn't add the men's 400... It would make things a bit sprint heavy, but the sprints are also the events where casual fans are more likely to know some of the names they are selecting from (more media attention). I was aiming for an equal number of men's and women's events, although I doubt we'd hear many complaints if we tipped the scales some. We could keep it even by also adding the 200 for women...

Strange that all the multis calculations scale differently for the throwing events. It seems like they were constructed to yield a comparable world record level of points to the non-throws multis field events, but less attention given to consistency at lower point levels. Don't know why that would be, but it's hard to argue with 5 out of 5 throwing events displaying this behavior while the other 6 events were all nearly perfect!


I'm open to suggestions from anyone as to how we can accurately price the athletes by event. I'd really rather not have to go through last year's results for everyone like I did with the men's 100m example above, but it almost seems like that sort of cumulative valuation is needed...

Dan

p.s. Another thought I had earlier: I had mentioned before that if things work smoothly, we could have a pool ($5 or $10 per person) for prize money at the end. Well, to avoid deterring people that would rather not bet, we could also have a non-paying option that would simply exempt them from prizes.

[ This Message was edited by: Dan on 2001-10-23 00:48 ]