Tuesday, April 12, 2011

The winner's curse in Yahoo display auctions

Preston McAfee is an unusual economist in that he has passed a market test: Yahoo is willing to pay him for his advice. His latest advice is on the design of auctions for graphical displays on web pages. Yahoo runs billions of these automated auctions every day, through their Right Media Exchange (RMX).  Advertisers can bid for certain kinds of customers, and can pay for impressions, clicks, or sales.

To choose the highest-valued bidder, Yahoo develops predictors of how many clicks and sales result from each impression. For example, if one click occurs for every ten impressions, an advertiser would have to bid more than 10 times as high for a click as for an impression in order to win the auction.

Yahoo was very proud of its predictors, but was puzzled that they systematically over-predicted the actual number of clicks or sales after the auctions closed. A well-trained economist would recognize this as an example of the winners' curse:

In a standard auction context, the winner’s curse states that the bidder who over-estimates the value of an item is more likely to win the bidding, and thus that the winner will typically be a bidder who over-estimated the value of the item, even if every bidder estimates in an unbiased fashion. The winner’s curse arises because the auction selects in a biased manner, favoring high estimates. In the advertising setting, however, it is not the bidders who are over-estimating the value. Instead, the auction will tend to favor the bidder whose click probability is overestimated, even if the click probability was estimated in an unbiased fashion.

As with the winner's curse, there is a simple fix--bid as if your estimate is the highest among all the bidders. This requires shading your predictors downwards, based on the variance of the prediction.

The paper also details two other features that Yahoo uses to make their auctions more efficient: randomized bidding, used when a less-informed bidder bids against a more-informed one; and sometimes the auction is not won by the highest bidder, due to the value of learning. When a bidder has a novel or unusual use for the display, sometimes Yahoo lets the novel user win so that Yahoo can learn more about the value of the use. If Yahoo learns that the novel use of the display is more valuable than they thought, then they can earn enough in the future to more than compensate them for giving up some revenue earlier on.


  1. Many hundreds of Internet auctions you can find at http://1-auctions.com/, with rating of the best the best auctions by thousands of visitors.

  2. It was a awe-inspiring post and it has a significant meaning and thanks for sharing the information.Would love to read your next post too......
    Share Tips

  3. This is one of the good articles you can find in the net explaining everything in detail regarding the topic. I thank you for taking your time sharing your thoughts and ideas to a lot of readers out there.

  4. According to NYC Department of Consumer Affairs, the city’s administrative code requires that auctioneers must have a license to operate an auction house within the New York City Counties. All auctions must publish auctions in their licensed name “in one or more newspapers” http://www.nyc.gov/html/dca/html/licenses/036-071.shtml prior to conducting the auction. Jewelry auctions such as “diamonds, precious stones or other jewelry must include a signed document containing a description of the article sold and the representation made in regard thereto at the time of sale” http://www.nyc.gov/html/dca/html/licenses/036-071.shtml.
    Auction regulations and license requirements differs state to state as well as within localities. For example, New York State does not require a license but localities such as NYC within the state do. States like Nevada may require licenses at the state level instead of localities such as Las Vegas. The United States Uniform Commercial Code is the basis for Auction regulations used by state and local localities and transactions are considered a bidding contract between buyers and sellers when consummated. Some auctions houses such as Heritage Auctions and eBay have policies designed around buyer’s remorse. In the case of Heritage Auctions, buyers may have 3 business days upon receipt to return a won item but this policy includes fees up to 20% of the auction value. In the comparison to eBay, eBay guarantees at least a 14 day return policy between buyers and sellers without penalty based on performance metrics but requires the buyer to pay return shipping costs which may/can include restocking fees depending on information provided by the seller. eBay uses metrics to determine the level of performance by both buyers and sellers to determine the level of protected returns and loss responsibilities.

  5. In his blog “The winners curse in Yahoo display auctions,” Froeb explains how Yahoo runs billions of automated auctions online daily to sell customers, impressions, clicks and sales. Yahoo uses predictive analysis to estimate the highest value bid, but consistently over-predicted the clicks or sales. Froeb calls this the winner’s curse. In his text Managerial Economics, Froeb explains it this way: “Imagine that you are bidding on a suitcase of cash. You are more likely to win the auction if you overestimate the amount of money in the suitcase than if you underestimate it. You are also more likely to lose money.” (2016, pg. 237) So the winner in this kind of auction is actually the loser in some ways, because they overestimated the value above all other bidders. This stands to reason, then, that all the other bidders bid far enough below the highest bid that the mass of bidding is inconsistent with the highest bid. This explains Yahoo’s dilemma. They thought the mass of bidders would be more in line with the high bid and were confused when that wasn’t the case. Froeb says “to avoid the winners curse, you bid as if everyone else thinks the value is less than your estimate.” You also, “bid less aggressively as the number of bidders increases.” (pg. 237) Yahoo’s predictions would be more on target if they took this into consideration.
    Froeb, L., & McCann, B. (2016). Managerial economics: A problem solving approach (4th ed.). Mason, OH: South-Western Cengage Learning.

  6. thanks for sharing best topics, keep such update and share best ideas over to all and me so i very happy with this.WhatsApp status is an awesome method to convey what needs be.
    Whatsapp Quotes