Run-Down Forums

[Conway]

You know that gives me an idea (actually two) .. One is obvious - that we should choose fantasy relay teams for the 4x4 .. But the other is that we should have a fantasy league for track and field period .. Perhaps could even start during the indoor season (would give time to prepare).. Dan,Micah, Justin, Myself and anyone else who would like to join .. Would be very simple .. We have a draft in which you would pick individuals for every event .. each week we would score a fantasy meet based on the resxzults of the individuals for that week .. Scoring just as you do a regular meet ... 10-8-6-4-2-1-0 (depending on how many people wanted to participate you would draft one or two people per event) ... Rules may need to be little more defined than that, but that would be the basic structure ... Anyone interested ??? What do you think Dan ??? Something to separate you from the ohter sites out there !!! Not that we aren't approaching elite status as it is ...

[Micah Ward]

Count me in!

[Dan]

Funny you should mention that. That's something I really wanted to do last year, but I balked at the thought of organizing it and was a bit tentative to take on something of the sort after watching participation in the monthly contests steadily decline to the point that it wasn't worth my time and effort.

If we set forth a set of rules, such as which events would be chosen from (can't really do all events since most meets are only holding certain events, which is the topic of another conversation), which meets (preferably ones where results are readily available! ), how to award points for athlete performances, how to allow choosing of fantasy teams (points per athlete based on last year's IAAF scoring system, where each player has so many points to spend?), if athletes can be traded, etc., then I could maybe put together some sort of a program that would tabulate scores on a rolling basis and keep us up to date as to how things are shaping up. It could probably be done just as easily, or easier, in a spreadsheet and posted to the web after each meet.

If we can brainstorm some way to devide work and how big of a thing to make this, then that would certainly help convince me it's worth embarking on.

Dan

[Micah Ward]

Just some quick thoughts off the top of my head:

Let's pick maybe 5 running events and 3 field events. The ones that are most commonly contested. Possibly:
60 meter dash
60 meter hurdle
200 dash
400 dash
1500 or mile, depending on meet
Pole vault
high jump
Long jump
Then we decide on maybe 5 major meets. Millrose, LA, Nationals, Worlds? Then we chose two athletes per event. We will choose by taking turns picking an athlete from any event until we all have 2 per event. And maybe this as the best idea of all...since I'm the dumbest of the bunch I get to pick first!

[Dan]

One problem I see with doing indoor meets is that it's so unpredictable which athletes will actually participate. With the state of indoor meets in this country, there's no guarantee they'll even take place...

Besides, waiting until outdoors gives us more time to decide how to organize this.

I'm not sure if picking a set number of athletes per event would work well, because the order we pick in would have no way of taking into account things like Szelzky winning nearly every time out in the 1500, making picks 2 through 5 pretty much a toss up, or an event like the 10k taking more out of one's legs and allowing less frequent attempts (significant if we change the order of picks per event and someone's #1 pick lands on the 10k and gets nowhere near the value out of it as in the 1500, for example).

Two ways I see around that problem are:

1) How every many people are playing, set an order of picks that we follow until however many athletes/events we choose to follow are chosen, with the order of picks being independent of events. For example, Justin might choose Zhanna with his first pick in the women's 100 , while Micah would counter with a number one pick (#2 overall) of Dragilla in the pole vault. That way, each contestant can evaluate events and athletes however they see fit.

2) Devise a points system like mentioned above. I think this would be the coolest, and would also allow the most participants (including after the league starts), but would take more work to pull off.

Other ideas?

Dan

p.s. I'll probably move this discussion to a separate thread to allow it to take on a life of its own. Just so you know if you come back looking for it here and it's gone...

[Conway]

Well I like the way it is shaping up already .. I think the idea of setting up a spreadsheet would be the easiest .. IT could show both weekly and cumulative scoring ... I also like Dan's idea of setting up the draft so that you choose your team based on what is important to you and you is available at the time !!! Would enable everyone involved to have a few aces on their team .. I like the indoor season, but Dan has a point in that who shows up becomes unpredictable .. But I do like the idea of targeting which mets will be scored .. For example all Golden League meets could be scored in addition to say certain Level I Euro meets .. Would like to find a way to include American meets, but not enough competition from Europeans .. Could do an NCAA season and then a "PRofessional" one .. But could more work than Dan is willing to put in .. And i do think we could help him on this so that it doesn't all fall on poor Dan .. Those are my initial thoughts .. What next fellas ??

[Dan]

Are you intending this to be something that just a few people participate in, or open it up to whoever wants in? If just a few people (i.e. those of us in this discussion), then I imagine each athlete could only be chosen by one player.

Picking certain meets is definitely the way to go. For one thing, I shudder to think of the difficulty in tracking down results from every obscure European venue and South African early season high altitude meet... Also, it adds an additional layer of strategy in that certain athletes typically perform well at certain meets year after year.

If we get to one of the pre-selected meets and suddenly discover that one of the events we're following is not being included there, should we implement some fallback plan for replacement athletes (similar point value?) so any player with key athletes in that event doesn't get shafted?

Trades? I don't see any reason to place restrictions on who can be traded or how often. Just add up the points for anyone on your roster at the point of each tabulation.

Once we have the rules defined and the scoring system and everything set up, maintaining it shouldn't be too bad. I think the major source of work at that point will be collecting and entering meet results, which is where assistance would be the most helpful.

Dan

[Dan]

Here's a good outline of rules for the Performance Bike fantasy league that was started up last year:

http://www.performancebike.com/inform/fantasy/fantasy.html

Cycling and running are similar enough that that might provide a good example to follow, at least initially.

It would be really cool if we could run the meet results through Purdy tables and score the game based on the resulting points, as that would still reward higher placing in events but would also take into account relative worth of performances. I just can't seem to find anywhere that lists the calculations that go into the tables... I've found a few programs that can calculate Purdy points, but no sign of the underlying data. Anyone know where that can be found and if it is feasible to work with? Two things I found are:

Patrick Hoffman's calculators:
http://www.cs.uml.edu/~phoffman/ex3.html
http://www.cs.uml.edu/~phoffman/ex1.html

and QSP Points:
http://www.uvm.edu/~msargent/qspoints.htm

Might be worth an email to Hoffman (another t-and-f lister) to see if he has any advice.

Dan

[Dan]

Check that, it may be on Hoffman's site afterall:

http://www.cs.uml.edu/~phoffman/xcinfo3.html

Dan

[Conway]

My thought was to open it up to anyone that wanted to participate .. Would love to have more people take part and this might be a fun way to increase site participation .. And if it went well and word got out !!!!

I like the idea of the Purdy tables ... Just ont sure how feasible it is .. Not sure if there might be an actual set of tables (sort of like the Big Green Book idea) ... That would be the ultimate .. But calculators coudl work too .. If they were available .. An email would be a good idea ..

Trading could certainly be allowed ... Might add some spice .. I quickly read the link you provided on the bicycle league .. Looks good as a start ..

would definitley have to select which meets wold be scored .. And like the idea of a fall back if an event were not offered or cancelled .. Would just have ot think of best way to do that ..

Any other ideas ???

[Micah Ward]

I'm not so sure I like the fall back idea. If an event or meet is canceled then that is just the disapointment of life in the real world. See we get to mix reality with fantasy.

Trades are ok.

Dan we may need to do as you suggested and get the rules and organization down in preparation for the outdoor season. We could still run maybe some "test" meets with the indoor season to see what kind of bugs would need to be worked out.

[Dan]

I thought of another issue: What to do about athletes that are in multiple events? Two or three options that I can think of:

1) Require an athlete-event combo to be specified, meaning if you pick ElG for the 1500 and he runs the 5k in one of the selected meets, you don't get points for that.

2) Make the athlete's "purchase" points the sum of their rankings across league-chosen events based on likelihood of running. Not sure what to do if we again use the case of ElG, for whom there is no current 5k ranking because he hasn't run it yet but plans to...

3) Lucky you!

I've been working much of the day trying to get the Purdy calculations working based on the stuff on Hoffman's site. I've got it close, but I'm not confident the equations are right in the first place, seeing as how his two different calculators yield different time/point equivalents when working backward, plus I'm currently at a loss to understand how a key step of the equations can possibly work...

If it were off by a consistent amount, then we could probably go with that. However, it is too high for some events and too low for others, usually +/- 10-40 points on a scale of 1000+. Maybe not a big deal, but I'd like to get it at least matching known WR values...

Another consideration is that the Purdy tables apparently only work for non-hurdle/steeple running events, which means no field events (not an objection for me ), unless we can come up with a similar scoring system for those that matches the points we're assigning for the running events. I'll take a look at the decathlon/heptathlon tables and see what those do.

On the off chance I do get these Purdy calculations right, anyone have a list of point equivalents for different times? I know some of them have been posted to the t-and-f list before, but I couldn't find them in the archive. All I've got to go on right now is a list of world records from a few years ago:

http://www.cs.uml.edu/~phoffman/postwr.html

Running a test season is a good idea. Just have to get something put together in time for an indoor meet or two...

No fall backs? Hmm, I suppose the real world approach could work, although I wouldn't want the unpredictable nature of the new "TV friendly" meet formats to discourage participation in the game.

Dan

[Conway]

I agree that 10 - 40 points is a pretty good sized spread .. Even over 1000 points .. Especially in the shorter events .. Time becomes magnified ...

I haven't seen lists in a long time ... And am afraid I didn't save them before .. ...

Test runs would be fun and should probably be done .. Course easy for us to say ...

Boy Micah is pretty tough there .. No fall back .. But hey if your guy / gal gets injured is kind of the same thing .. Could work either way ...

[Dan]

Yeah, I'd hate to see Micah's bad side.

I haven't figured out the Purdy thing yet, but I'll post an update if and when I do... It's got to be something wrong with my calculations (or so I hope), but my once great math prowess has eroded to the point where basic arithmatic is sufficiently challenging, so finding the error of my ways is proving elusive.

If we do pull this together, a test run during the indoor season seems quite reasonable. If that works, we could maybe do a small NCAA tourny and see what kind of participation that pulls in. If it is a success, then maybe we could really do it up for the Golden League series and charge $5 or $10 to enter, with a nice sized jackpot for the winner(s).

Dan

[Dan]

I've got a potentially good update!!! After much monkeying around, I have what appears to be some good numbers resulting from the Purdy points calculations:



http://run-down.com/statistics/purdy.php



Method #1 is the apparently "real" method of basing points on a standard distance/velocity comparison. It's actually pretty close now to correct values posted on Patrick Hoffman's site, but the disconcerting thing is that I don't recall changing anything on that method in the last round of tinkering and suddenly it became more accurate...



Anyway, the Purdy point values I've been trying to match are posted here:



http://www.cs.uml.edu/~phoffman/postwr.html



Additionally, the following page allows you to enter a distance and time and get the resulting Purdy points:



http://www.cs.uml.edu/~phoffman/ex3.html



(the Purdy points are the second column of the 4th line when you submit the form) It's a bit overwhelming to work with all that page shows you, but all the distance/time combos I've tested seem to be in close agreement with those on my demo page above.



The big question remains if those values are accurate in the first place... So, if anyone runs across somewhere that lists Purdy points corresponding to world records or anything else not covered on the postwr.html above, by all means let me know!



Oh yeah, method #3 is an approximation based on some estimates of startup time and slowdown based on running the curves. That's taken pretty much directly from Hoffman's examples.



Method #2 is the one that appears dead-on, and it's sort of a bastardization of method #1. I worked backwards to find some seemingly bogus standard-time adjustments for every distance (I say bogus because to make things work, I ended up plugging in 9.2 seconds for a 950 point base level 100m -- 9.2 is obviously worth more than 950 points!!) and skipped the actual Purdy table lookup and corresponding calculations. Unless it turns out to be inaccurate, I'm happy with it...



One thing this doesn't address is what to do with women's events? All the Purdy stuff I've seen deals with a single scale and only shows how men's performances rate. Anyone know if there is a separate scale for women or a factor that can be multiplied by their performances to yield a comparable number? Most Purdy table usage probably has no use for comparing men's and women's marks directly, but for the purposes of compiling scores for our game, it would be rather important!



Also, in looking through some example times for each given distance, it seems that the number of points don't change signficantly unless there is a huge change in time. So, I'm a bit undecided how much value the points have in that form for use in scoring the game, since my hope was that the Purdy points would differentiate good performances more so than the standard 8-6-4-3-2-1 scoring.



If used, it looks like the two items (place and Purdy points) will need to somehow be used in a number that meaningfully combines them. For example, Reynold's 400m record of 43.29 is worth 1,106.03 points, whereas MJ's 43.18 is worth 1,111.48. A mere 5.5 points separates the two, which is a very small percentage of the number. So, maybe we could do something like "scale" the numbers down to 106.03 and 111.48, multiply the event places by 10 or so (resulting in 80-60-40-30-20-10 type scoring), and add the two together. That way, a typical 400 win might result in 80 points on the adjusted Purdy table and 100 points for the win.



We could go a step further by treating the scaled down Purdy value as a percentage of the world record time/points and put that on a scale of 1-100. Hmm, I suppose we could do that regardless of Purdy calculations... In which case, what did I just spend the last day and a half working on??? If nothing else, I may have something new to add to the 'statistics' section of Run-Down...



Dan