Run-Down Forums

[Micah Ward]

You guys have lost me. Remember, I am something of a neanderthal. What is all this talk of Purdy points? Is this to score the meets or to rate athletes values for the draft?
For meets I was hoping for something simple like:
1st 10 pts
2nd 8 pts
3rd 6 pts
etc.,etc.,etc.

For the draft use the IAAF Rankings. You pick 2 athletes per event. For illustration lets say 2 athletes for 8 events. That means 16 athletes on your team. Put a limit of no more than 4 per team can be ranked number 1 on the IAAF list, no more than 4 ranked 2nd and no more than 4 ranked 3rd. That leaves you with 4 that have to be picked very intelligently. The numbers apply to both your mens and womens team.

As far as fall backs for injuries or cancelled events. For events we need to pick the most contested events so that does not become a frequent issue. For injuries...you just have to eat them...same as you would if you were coaching a real world team.

A simple plan concocted by a simple guy. By the way, I don't have a bad side...I'm a big old softy.

[Conway]

Dan is looking at quantifying a win ... OR a loss ... A close loss in one event could be better than a lousy win in another .. The Purdy tables are a way of saying how good a performance is .. So obviously a WR would be an outstanding performance and win .. Let's take the mile .. A win of say 3:48 ... Would get more points that a win of 3:50 ... By the same token (and correct me if I am wrong Dan) a win of 3:58 ... May not be as good as a loss of 3:50 .. So scoring would take into account just how well you perform - not simply whether you win or lose ... ... That lousy 5000 down under for example would have gotten lousy points for the participants had anyone had them on their team ... At least I think that is where Dan is headed ...

[Dan]

Precisely.



Patrick Hoffman just got back to me and cleared up an area I was misunderstanding the calculations, so it appears I now have two working models.



Micah, ironically, I think your system for choosing athletes is actually less simple than allowing players to choose whoever they want within their point allowance. If I have enough money to buy up the top 15 Kenyan and Morroccan steeplechasers, then so be it! I wouldn't do well in other events, but I'd be assured of good points in the steeple...



The key is coming up with a good system of pricing the athletes, which seems best accomplished by using the 2001 IAAF rankings points (not sure what we would do for an NCAA game). I suppose that means newcomers could not be accurately added to the pool of available athletes. It also means someone that improves a lot this year could be a real steal.



Dan

[Micah Ward]

How about these ideas:

For the draft assign purchase points based on the IAAF ranking. For example:
1 100 pts
2 80 pts
3 60 pts
4 40 pts
5 30 pts
6 20 pts
7 10 pts
Any one ranked lower than 7 is free.
If a team has 16 athletes then figure 50 points per athlete and that gives you 800 points to use in picking your team.

For scoring the meets use place only. If Kalamazoo College and Topeka Tech are having a meet they get scored on place only, regardless of how fast the runners run or how high the jumpers jump or the shot putters put. This way we keep it as realistic as fantasy can be...and simple enough that even I can understand.

[Dan]

I agree with the basic idea of assigning a purchase price to athletes based on their IAAF ranking, but I think it needs to be the points, not just the rank. Maurice Greene is ranked #1 in the 100m, as is Shawn Crawford in the 200m, and I know which one I'd pick if they cost an equal amount! The difference is 74 points, which isn't a lot to differentiate them (Crawford ran against such weak competition that someone had to win the meets and score rankings points...), but at least it places some value on the relative worth of their respective #1 rankings.



It appears from skimming the rankings lists that no one has a score below 1000, so we could subtract 1000 from everyone's IAAF ranking, giving purchase prices from roughly 80 to 470, making a 74 point difference like in the example above significant enough to maybe make Tim Montgomery a smarter pick as the #2 100m man than the #1 person in another event.



Of course, one might also look at the "cheaper" top ranker as a bargain, reflecting the level of competition in the event. That's where the Purdy points come in, though. A 20.3 for 200 might be enough for first place points, but it won't earn many Purdy points. Making the two 50/50 in the overall points equation would take into account several factors.



They also have an interesting table (APPENDIX IV PLACING SCORES) two-thirds of the way down the rankings explanation page:



http://www.iaaf.org/Results/Rankings/how.html



It appears they inflate the point system enough to be able to place an increasingly large number of points as the place improves.



Dan

[ This Message was edited by: Dan on 2001-10-17 08:55 ]

[Dan]

This may answer one of my above questions:

percent of women's world record to men's, based on Purdy points (avg is 81.88%), using my latest un-verified calculations:

100m: 85.02%
200m: 80.52%
400m: 82.03%
800m: 80.27%
1500m: 81.43%
3k: 84.19%
5k: 79.12%
10k: 82.48%

So, dividing women's Purdy points by 0.82 should give a fairly comparable ranking to the men when it comes time to score events.

I'm starting to think I should have started a separate thread just for Purdy talk...

Dan

[Conway]

I like the idea of incorporating the Purdy table as a way of grading performances .. I do think that one key will be WHICH events we choose to contest !!! As relative level of competition last year does have an effect on drafting and how one uses ones allotment of purchase points .. AND the surety of whom one is going ot draft ..

For example the 100 is easy .. Mo, Montgomery, Williams, Boldon, Chambers .. The 400 not so easy .. Moncur, Schultz ?? One season wonders or are they legit ?? Washington and Harrisons come back ?? 110 hurdles have lots of good performers but they are almost interchageable .. 3 different events .. 3 different ways to look at a draft ..

Also we really have to take a look at the field events as they were sparsely available last season !!!

Another question .. Do we want to award bonus points for records ??? Just a thought ... AND ... Do we have relays ??? As in put together a foursome and combine their times to create a relay ... And they wouldn't have to all be in the same event !!! Again just a thought .. If you had two people per event and we had two different sprints (or hurdles) you could combine for example the 100 and 110H times of the 4 individuals to get the time for your Run-Down relay team !!! Sort of like the drama of a final 4x4 ...

Anyway just some ideas ... But I like where we are headed

[Dan]

Bonus points for world records seems quite reasonable, although I think that would require us to skip events such as the women's steeple and pole vault, as world records are far too likely at their stage of development and would skew the scoring of other events.

Along the same lines, I'm all for sticking mostly to running events. Both because field events being held seem a bit more spotty and because I can't calculate point values for them as easily. I am looking at some multievent scoring tables that might work for approximating some of the included field events. They don't match up with the Purdy point totals, but another idea I'm working on (more in a minute) might make that a moot point.

The IAAF rankings points partly take into account the so-so events, but it's definitely a gamble choosing athletes from the likes of the 200 and 400 last year. For guys like the Harrisons and Tyree, we might want to come up with agreed upon price for them, which would allow them to be chosen even though they aren't really reflected in the 2001 rankings.

Relays, hmm. Could get tricky to pull off, but it's an interesting idea. Everyone would have to have 2 or more people running in each event making up the relay. We wouldn't be able to do it with our season starting rosters, as we'd have no idea the x number of people we chose in the 100 will actually run the 100 that meet... Even the meet start lists may not be accurate enough to be of much help, so I'm not sure when we could choose relay teams with any certainty.

Ok, here's where I'm at right now on scoring: First, the Purdy tables seem to seriously undervalue the 100 relative to other events, especially to the 200. So, instead of comparing the Purdy points across events, we could make relative comparisons (to the WR points) within the event and score that way. Here's an example, based on Purdy points that may not yet be accurate:

In the 100m, 9.79 is 1,063.09 points and 9.85 would be 1,048.55, which is 98.63% of the WR level. a 10.00 would be 1,012.96 points which is 95.28%. If we then take the percent difference (1.37% and 4.72%, respectively) and square it, then subtract that from 100 (100 points assigned to a WR level mark for any event; the equation would also need to account for scoring above the WR, which would be 100+ points), we end up with scores of 98.1 for 9.85 and 77.7 for 10.00.

We could approach the actual equation other ways, but that example does a nice job of separating out performances, since the Purdy scale differentiates performances by such a small amount (a few points here and there on an 1100 point scale). Maybe cube the percent difference, as a 10-flat arguably isn't worth as much as 77 points?

Next, the scoring based on places within an event could go like:

1st: 100
2nd: 80
3rd: 60
4th: 50
5th: 40
6th: 30
7th: 20
8th: 10

Someone winning a race and matching the world record gets 200 points for their effort, making the two halves of the equation a tidy 50/50.

Dan

[Dan]

An amendment to the above suggestion percent scalings of the Purdy points per event: I think it might be better to multiply the percent difference by 10 instead of squaring it, as squaring a small number results in a very small number... Going this route would make the above example 86.3 points for the 9.85 and 52.8 points for the 10.00.

I think that gives a much more accurate interpretation of how the times compare to the world record, especially considering the points system is basically a bonus to the points received for placing. It shouldn't reward mediocre performances all that much...

The exact manner in which we set up points scoring probably isn't too important, so long as we agree on it ahead of time and stick with it throughout the game.

Dan

[Dan]

Here's a work-in-progress sample to play with:

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

(note: if something doesn't work, i'm probably testing...)

Dan

[Conway]

Played with it ... Works pretty good ... What next ?? Should we try to establish rules of play ??

[Dan]

Just as you posted that, I made a tweak to accurately (as best I can tell from other lists) reflect world record performance levels in the Purdy points, which slightly affects relative % performances.

I found someone in town that has an old copy of the Purdy tables book (probably hasn't been revised, since it's based on 1930's performance levels and somehow extrapolated). I'll probably take a look at that Friday and [hopefully] confirm the work I've done thus far.

Rules of play would be a good thing to tackle. I've started outlining how to put a program together to make all this magically happen with each event. I'm only putting off half a dozen or so other projects I should be working on...

What did you think about my idea above about pricing athletes by subtracting 1000 points from their IAAF ranking? We would then do something like Micah suggested by deciding roughly how many athletes should be on each roster, then come up with an available number of points to spend on however many athletes you want to blow your money on. I don't see any reason whey we need to require x number of athletes on each roster. If someone wants to hedge their bets on a dozen middle of the pack athletes, that would be an interesting comparison in strategy relative to someone else spending everything on three or four guaranteed top rankers.

Also, we'll need to decide whether an athlete can be on more than one player's roster. If we base selections on the pricing system, then I see no reason why athletes cannot be shared across rosters (makes trading a somewhat frivolous thing, but still feasibly practical). Makes it easier to maintain selection lists for who is chosen and available, and it also doesn't limit how much participation we can draw. If we instead go with one roster per athlete, then we'll need to devise some selection method.

Also, what to do about an athlete in multiple events? If we have the women's 1500 and 5k as two of our events, we expect Szabo will score well in both, more so than her IAAF rankings points are likely to predict. I can think of two lines of thought for how to approach this:

1) If you pick Szabo for your roster and her purchase price is based on her 5k marks, any time she runs the 1500 you get bonus points there, which is probably nullified by her having not run the 5k in that meet. On the other hand, Ato Boldon typically doubles well in the 100 and 200, easily resulting in points double what his rankings would cost a player.

2) Require each athlete selection to be pertinent to a single event (the 1500 and mile could be considered the same event, since no one is going to double those in the same meet!). If you choose Ato for your 200 team and he runs a 9.80 in the same meet, too bad.

I'm not quite sure which approach I like better. They both have some holes in them...

Dan

[Micah Ward]

I may be a lone voice here but I think we are going in the wrong direction. We are trying to make it too complicated. I can see giving athletes point values for the purpose of selecting teams. But when it comes to scoring the meets let's just keep it simple. For Example:

If Conways 100 man wins then Conway has 10 points. If Dan's comes in second then Dan has 8 and if Conway's comes in third now Conway has 16 points and if mine comes in fourth I have 6 points, etc.

Then my guy who ran the 100 (maybe Ato) doubles in the 200 and wins. Now I got 16 points, 10 for the 200 and the 6 from the 100. Just like in a regular meet.

If Dan has Jenny Adams and she wins the long jump (10 pts) and finishes second in the 110 hurdles (8 pts) Dan has 18 points to add to his total.

This is the way a real meet is scored and ElG gets 10 points for winning the mile no matter how fast he runs.

Save the Purdy points for when we race each other next year!

[Dan]

I don't mean to dismiss your suggestion -- I have considered the straight 1-8 scoring method, which was actually my original intent -- but I just don't think that's a very accurate reflection of scoring a non-team effort in an event based on raw performances. If we get a repeat of the 15+ minute 5k nonsense, that surely isn't worth the same number of points for a win as a 12:40!

As far as athletes in multiple events, the issue isn't whether or not to get points for them, it's how to assess the value of such athletes when pricing them initially? If someone is likely to double, then that has to be reflected in their cost, otherwise they're worth twice as much as another athlete and everything is thrown out of whack.

Dan

[Conway]

I have ot agree with Dan on this one .. If it were a real type of competition where we were coaches and had more control over who was going to run what and when, then I would be more inclined to go with a straight scoring system .. However once we pick names we are out of control !!! So I think there needs to be a premium on the value of individuals as well as their level of competition .. I also like the idea of you spned your "money" and see what it garners you !!! Mo has got to be worth more than Dwain Chambers ... And if you spend all of your money on Mo and Ato then you better hope they double and double fast ALL the time !!!

Not sure how I feel about more than one person owning an athlete ... Guess it could work .. Although might have to be a limit depending on how many people are playing ... Everyone can't own Szabo for instance ... Then there is no competition so to speak ..