High Frequency Trading and Online Advertising

The advertising world is being impacted by the adoption of technology at it’s core.  I speculate that the majority of all ad impressions will be “served” by some sort of technology within the next decade.  This speculation includes print, television, radio and OOH.  We’re seeing this start to play out within the online advertising space.

I’m a believer that history repeats itself and that we can learn from correlations to similar industries.  I was reading the Technology Review this morning and came across this article on Wall Street’s high frequency traders.

Are the likes of MediaMath, Turn, Invite Media, and Dataxu on the way to becoming high frequency trading platforms?

  • http://knowabout.it/ falicon

    This white board TOTALLY makes sense to me. Completely clear. I especially like the details on the “Feed Handlers” :-)

    • http://www.darrenherman.com dherman76

      I figured you were the architect behind it!

  • http://blog.botfu.com Kevin Marshall

    This white board TOTALLY makes sense to me. Completely clear. I especially like the details on the “Feed Handlers” :-)

  • http://twitter.com/shaunabe Shaun Abrahamson

    desperately trying to zoom in on the whiteboard :)

  • http://twitter.com/shaunabe Shaun Abrahamson

    desperately trying to zoom in on the whiteboard :)

  • http://danreich.com Dan Reich

    If what you say is true, “the majority of all ad impressions will be “served” by some sort of technology within the next decade” AND “that history repeats itself” (which i agree with) then won’t we experience a sort of ‘adverting collapse’ much like the recent credit crisis?

    Check this article if you haven’t already: http://blogs.hbr.org/haque/2009/11/facebooks_scam_ads_and_the_loo.html

    If the bulk of the market works off of a central system, and someone exploits a flaw in the system, won’t it essentially implode and correct itself?

    Good post.

    • http://www.darrenherman.com dherman76

      Dan, great comment. Theoretically, any market could collapse.

      What is technologically beautiful about the financial markets are the systems on which facilitate the trades, the exchanges, and the plumbing. That’s going to permeate the advertising world. There could be a collapse, but if that happens, there will be a group of brands who perform well, just like the financial collapse. It’s up to the brand to have a good trader/agency to keep their head above water should that happen.

  • http://danreich.com danreich

    If what you say is true, “the majority of all ad impressions will be “served” by some sort of technology within the next decade” AND “that history repeats itself” (which i agree with) then won't we experience a sort of 'adverting collapse' much like the recent credit crisis?

    Check this article if you haven't already: http://blogs.hbr.org/haque/2009/11/facebooks_sc

    If the bulk of the market works off of a central system, and someone exploits a flaw in the system, won't it essentially implode and correct itself?

    Good post.

  • http://www.darrenherman.com dherman76

    I figured you were the architect behind it!

  • http://www.darrenherman.com dherman76

    Dan, great comment. Theoretically, any market could collapse.

    What is technologically beautiful about the financial markets are the systems on which facilitate the trades, the exchanges, and the plumbing. That's going to permeate the advertising world. There could be a collapse, but if that happens, there will be a group of brands who perform well, just like the financial collapse. It's up to the brand to have a good trader/agency to keep their head above water should that happen.

  • http://mikepratt.tv/ Michael J. Pratt

    Darren – I agre with you that these two markets share many similarities and that algos will have an increased role in the media and ad markets. A few things might serve to prevent total similarity though. As HF traders, we generally don’t care what the instrument we are trading at the moment actually does. We care that it’s fungible and trades a lot. I’m not as sure about the impressions market but I don’t think that the same thing applies. In equities, you only have ~ 5000 instruments and of these, really only half are even considered by our finds for trading, the rest “trade by appointment” as we say. Many HF funds only trade in < 1000 stocks. Those instruments don't change much. They trade every day and a share in IBM today is the same as a share 6 months from now.

    To my eye, impressions and digital media are more like the bond market. GM will have 50 bonds with relatively small tranches, say $100mm ea. Now these bonds are very similar but not exactly the same. In fact, they are arbed between them. By themselves, they are hard to trade in the HF realm because they are too small. I think the same thing will apply to the ad market.

    Of course, the marketplace for ads could actually create a set of generic ad "instruments" that become heavily traded on some sort of exchange and heavily used by the ad consuming market. That would totally lend itself to the application of algos to exploit the mis-pricing that invariably accompanies markets.

    Point of correction the article you cited: "Flash Orders" (which uneducated outcry successfully caused a repeal of) did not give HF traders a chance to front run. The article quoted: "the "flash trade," in which exchanges alert designated traders toincoming orders. Critics call it a variation of front-running, an old (and illegal) practice that involved traders buying and selling in advance of large orders." The irony here is that big shops like fidelity were the ones who sent "flash orders to the exchanges, pinging the markets for liquidity. ANYONE who wanted could see these flashes. Most people didn't pay attention because you have to write software to be able to even react. Guys like us would simply listen for the flashes from these big, dumb players and act accordingly. Then Chuck Shumer comes along and cries foul on their behalf and paints us the bad guy.

    Great piece. Thanks for writing it.

    Mike

    • http://www.darrenherman.com dherman76

      Awesome commentary. I’m interested in:

      ANYONE who wanted could see these flashes. Most people didn’t pay attention because you have to write software to be able to even react. Guys like us would simply listen for the flashes from these big, dumb players and act accordingly. Then Chuck Shumer comes along and cries foul on their behalf and paints us the bad guy.

      Can this exist again today or in the future?

      • http://mikepratt.tv/ Michael J. Pratt

        re your question: Yes, it can exist. here’s all you need. 1. capability is provided e.g. flash order 2. large player uses said capability in such a way that small player can take advantage of him. It doesn’t make the small player evil. I liken it to the “Dr., it hurts when i do ‘this’ ” syndrome. Well, don’t do ‘this’ !!

        Back in the day, big mutual funds used to show big orders to the market place then wonder why the stock would always move away from them. So, as Darwin would have predicted, they stopped showing their entire order (read: intentions) to the marketplace and, wow, the stock stopped moving away from them.

        Enter flash orders and it’s like… “SEN Shumer, when we show our flash orders, the stocks move away from us. It’s just not fair” Go figure.

        • http://greghills.com greghills

          Great post Darren.

          Thanks for your commentary Michael. I’m thinking about what the first exchange-traded ad instrument might be….maybe an exchange traded fund like the Comscore 500, which would be cost of a homepage impression on the top 500 websites. Or maybe an ad ETF for specific “sectors” like Travel and News.

          I think that one important precursor for online advertising instruments will be the adoption of meaningful general indices, like daily CPMs for a Comscore 500. Right now online advertising price indices are limited since they generally rely on quarter-long averages of prices and also rely on a non-random sample of sale prices for select survey participants. Once advertising exchanges have the same transparent bidding landscape as certain financial exchanges, with upticks and downticks, we’ll be much closer to trading advertising instruments.

    • http://www.highfrequencytradingprofit.com/ High Frequency Trading

      Well put Michael,  when HFT is used for fungible arbitrage it helps to create a much more efficient market place for all.  Its funny that people take advantage of algos that were writtent to take advantage of humans in this round robin evolution in trading.  The losers often cry foul.  Sometimes they are right, sometimes they are just slow.

  • http://mikepratt.tv Michael J. Pratt

    Darren – I agre with you that these two markets share many similarities and that algos will have an increased role in the media and ad markets. A few things might serve to prevent total similarity though. As HF traders, we generally don't care what the instrument we are trading at the moment actually does. We care that it's fungible and trades a lot. I'm not as sure about the impressions market but I don't think that the same thing applies. In equities, you only have ~ 5000 instruments and of these, really only half are even considered by our finds for trading, the rest “trade by appointment” as we say. Many HF funds only trade in < 1000 stocks. Those instruments don't change much. They trade every day and a share in IBM today is the same as a share 6 months from now.

    To my eye, impressions and digital media are more like the bond market. GM will have 50 bonds with relatively small tranches, say $100mm ea. Now these bonds are very similar but not exactly the same. In fact, they are arbed between them. By themselves, they are hard to trade in the HF realm because they are too small. I think the same thing will apply to the ad market.

    Of course, the marketplace for ads could actually create a set of generic ad “instruments” that become heavily traded on some sort of exchange and heavily used by the ad consuming market. That would totally lend itself to the application of algos to exploit the mis-pricing that invariably accompanies markets.

    Point of correction the article you cited: “Flash Orders” (which uneducated outcry successfully caused a repeal of) did not give HF traders a chance to front run. The article quoted: “the “flash trade,” in which exchanges alert designated traders toincoming orders. Critics call it a variation of front-running, an old (and illegal) practice that involved traders buying and selling in advance of large orders.” The irony here is that big shops like fidelity were the ones who sent “flash orders to the exchanges, pinging the markets for liquidity. ANYONE who wanted could see these flashes. Most people didn't pay attention because you have to write software to be able to even react. Guys like us would simply listen for the flashes from these big, dumb players and act accordingly. Then Chuck Shumer comes along and cries foul on their behalf and paints us the bad guy.

    Great piece. Thanks for writing it.

    Mike

  • http://www.darrenherman.com dherman76

    Awesome commentary. I'm interested in:

    ANYONE who wanted could see these flashes. Most people didn't pay attention because you have to write software to be able to even react. Guys like us would simply listen for the flashes from these big, dumb players and act accordingly. Then Chuck Shumer comes along and cries foul on their behalf and paints us the bad guy.

    Can this exist again today or in the future?

  • http://mikepratt.tv Michael J. Pratt

    re your question: Yes, it can exist. here's all you need. 1. capability is provided e.g. flash order 2. large player uses said capability in such a way that small player can take advantage of him. It doesn't make the small player evil. I liken it to the “Dr., it hurts when i do 'this' ” syndrome. Well, don't do 'this' !!

    Back in the day, big mutual funds used to show big orders to the market place then wonder why the stock would always move away from them. So, as Darwin would have predicted, they stopped showing their entire order (read: intentions) to the marketplace and, wow, the stock stopped moving away from them.

    Enter flash orders and it's like… “SEN Shumer, when we show our flash orders, the stocks move away from us. It's just not fair” Go figure.

  • http://greghills.com greghills

    Great post Darren.

    Thanks for your commentary Michael. I'm thinking about what the first exchange-traded ad instrument might be….maybe an exchange traded fund like the Comscore 500, which would be cost of a homepage impression on the top 500 websites. Or maybe an ad ETF for specific “sectors” like Travel and News.

    I think that one important precursor for online advertising instruments will be the adoption of meaningful general indices, like daily CPMs for a Comscore 500. Right now online advertising price indices are limited since they generally rely on quarter-long averages of prices and also rely on a non-random sample of sale prices for select survey participants. Once advertising exchanges have the same transparent bidding landscape as certain financial exchanges, with upticks and downticks, we'll be much closer to trading advertising instruments.

  • http://www.indydisplays.com trade show booths

    Fantastic post for trading advertising. This is the kind of info I need…concrete suggestions to increase business. You don’t know what a treasure you are. I love your blog.

  • http://www.highfrequencytradingprofit.com/ High Frequency Trading

    Well put Michael,  when HFT is used for fungible arbitrage it helps to create a much more efficient market place for all.  Its funny that people take advantage of algos that were writtent to take advantage of humans in this round robin evolution in trading.  The losers often cry foul.  Sometimes they are right, sometimes they are just slow.