over weights auction values

[ericr77 0]

i have used the compiler for about 3 years now and am happy to see an auction feature. However based on auction values from my 2 leagues last year I feel the compiler over estimated the max bid (especially for RB's) a bit too much. In my league with $200 cap the most spent for any player was $63 but the compiler has a max bid of $103 for Ricky with my scoring and roster systems. So this does not provide accurate values for me to go by based on the past history of my league. The compiler also recommends I spend $43 for the Bucs D but then $1 for the other D's. The highest price D went for $13 in my auction last year.

I guess what I want to know is how I could tweak this to get more accurate bid amounts and make this more valuable to me.

 

Thanks for any help.

 

Eric

[Mike MacGregor 20]

Hi Eric. I'm glad someone is finally responding about the auction values, because I would love to discuss them.

 

It was back in Feb/Mar I was researching and working on how to properly calculate the auction values. I read articles such as one by Greg Floyd on converting value calcs to auction dollars, and also researched fantasy baseball leagues which commonly use auction based drafts. I also discussed different methods/theories with some guys who had been doing auctions for a number of years. Here is what we finally came up with:

 

1. Do the value calcs as the Compiler has always done (these drive the overall list in a non-auction league).

 

2. Sum these values for all players above the chosen baseline, which will depend on number of players at each position and overall ranking method.

 

3. Divide the total league wide cap, less minimum bids for all players to be drafted, by the sum from step 2. This represents the dollar worth of each value point a player is calculated to have from step 1.

 

4. Multiply the dollar per value point amount from step 3 times the total value of each player from step 1. Add to this the minimum bid.

 

Restrictions/Assumptions:

 

- All players falling below the baseline for whatever method is being used will automatically be given a bid value equal to the minimum bid.

- The max bid for a player cannot be greater than a single team's cap number less minimum bids required to fill the roster.

 

The thing to remember is that the auction value being calculated is the maximum bid for any player. At the top end, if the bids are considered too high based on past experience, then you certainly don't need to bid this much.

 

But, based on expected performance and the difference in value between say Ricky and the bottom rated RB, I think it is fair to say Ricky is worth half your team cap.

 

Of course, there is an inherent risk in putting so much cash down on a single player, no matter who it is. Perhaps it is more prudent to diversify a little more, and that is why the values seem too high. The actual historical values you are used to account for that risk. I'm not sure how this would be estimated though. Degree of risk would be more based on personal preference I would guess.

 

Something else I've been thinking about is that if you have Ricky slated at say $100, but the final bid on him is only $70, then that is an extra $30 that should/will get distributed elsewhere later.

 

I'm not sure how to account for that. Draft Buddy does track whether the actual bid is above or below your max bid for each player, so you can see the trend and adjust accordingly during your draft.

 

Anyway, there are my initial thoughts. Does anyone have any more comments on calculating the auction values?

 

[ 06-30-2003: Message edited by: Mike MacGregor ]

[ICETIME25 0]

Mike,

 

I think that your baseline assumptions are sound.

 

Having said that, each league is unique. In the last three years of conducting an auction, I think that I've built enough history with a core group of 9 owners (of 12 total) to put some predictability into the auction. I've got a spreadsheet that I've sorted by overall auction bid, as well as by position. Each view provides important information regarding the value of each round (i.e. first 12 players, next 12 players, etc.), as well as the tiers within position (i.e. top 3 QB's, next 4 QB's, etc).

 

I think your model could be enhanced by the following method. For the number of players on your roster (i.e. 15 in my league), allow the percent of total cash pool that will probably get used during that round. Then you could use relative value of players within the round to allocate among the players.

 

The most important aspect of the auction (IMHO) is to understand a players value relative to those in the same scoring bracket.

 

Having said all of that, I need to think about the position aspect as well. While in our league's scoring Rich Gannon finished 4th overall, QB's never get the type of value a top 10 RB would get. Maybe that's an mistake by our ownership group. On the otherhand, there's a tremendous depth at QB and the differentiation may not warrant that value.

 

In the end, I consider both round and position in calculating my auction values by starting with round and adjusting for position based on league history.

 

Hope this helps.... :cool:

[ericr77 0]

Thanks for the reply Mike. In my league we start 2QB 2RB 2WR 1RB/WR 1TE 1K 1D with 6 bench spots. QB's get 6 pts. for a TD. This put tremendous value on the QB spot.

 

I think it would be valuable to allow for people to enter in allocations for what they intend to spend per position for their team. This way the compiler could calculate auction values personally tailored for you.

For example I might allocated $120 to RB's, $50 to QB's, $20 to WB, $1 K, 1$ TE, $7 D.

The compiler could then tell me how much I could afford to spend per player to fill my roster. This would be my preference based on past history and my personal preferences.

 

Eric

[Mike MacGregor 20]

Are you sure that would give the best result Eric? After a number of requests, I actually did a baseball version of the Compiler this past spring and it was set up to insert a % of cap dollars to each position.

 

I don't think the results were that great. The main problem being that each person in the league will spend a different % per position, plus we are getting away from calculating/comparing value across positions by fixing the amount spent on each position.

 

Really, we should be able to assess value of the whole group of players across positions, and allocate dollars appropriately. If we think certain positions are overweighted (i.e. Team D) or underweighted (i.e. QB in your case), then adjusting the baselines is perhaps the best method to correct the problem.

 

My initial thoughts anyway. I need to think about this some more...

[ericr77 0]

Yeah, I understand what you are saying. I guess all I could think of, it would be good to have a min and max so you know how much to bid up to not let a player go for too cheap and then also have a max to not over pay. Then the middle ground is basically left up to your preference.

I think adding a min. or a median price would be helpful.

 

Do you know of any website or software that let you do e-bay style auctions for free agents through the season. We use NFL.com and I have to manually manage the whole process.

 

thanks,

Eric

[Mike MacGregor 20]

I don't know of any service that handles auctions really, in particular like eBay. What do you mean during the season? Do you do free agents through a bidding process too?

 

I am in 2 leagues that do FA bidding, but it is single blind bid submitted to a 3rd party.