Why the BetFair model is partially obsolete

No Gravatar

I like BetFair and the BetFair people very much. I was the only blogger to talk up the BetFair starting price system and the BetFair brand-new bet-matching logic. But the other face of the coin is that 2 aspects of their model are rotten to the core.

BetFair was created in 1999 and started off in 2000. Since that time, 2 major things arrived on the world scene. Number one, we have seen the emergence of the prediction market approach. Number two, the Web has taken our lives, and Google has become the dominant Internet search engine. Here are how these 2 major trends are affecting BetFair negatively.

  1. Decimal Odds (a.k.a. Digital Odds). – The prediction market approach means that we attack the public with the news and their associated probabilistic predictions, expressed in percentages, where high prices mean high probabilities of happening. BetFair, at the contrary, approach the public with a betting universe and an arcane vocabulary (&#8221-backing&#8221- and &#8220-laying&#8221-) where low prices mean high probabilities of happening. That is totally counter intuitive.
  2. Non-Indexable Prediction Market Webpages. – Like it or not, Google is now the world&#8217-s #1 media. We &#8220-google&#8221- anything, first thing in the morning. None of the BetFair prediction market webpages can be indexed by Google and the other Internet search engines. That means that BetFair is missing out, in my estimation, on hundreds of thousands of Google visitors each year. Those Google visitors will favor other prediction exchanges (e.g., HubDub) whose prediction market webpages are indexed naturally by the Internet search engines.

The British, who drive on the wrong side of the road, don&#8217-t have the 2 most important keys of the future.

VP conditional probabilities

No Gravatar

BetFair is running markets on both who will be the next vice president and who will be nominated by the two parties.

As we&#8217-ve discussed before in other contexts, one can divide two probabilities like these to obtain a conditional probability: e.g., if the Democrats put X on the ticket, they will win the general election Y% of the time (where Y = odds of X becoming VP/odds of X being nominated).

These markets are thin, so the conditional probabilities should be taken with a grain of salt. But they are interesting nonetheless:

The pattern I see here is that conditional probabilities are higher for fresh faces (Webb, Sebelius- and arguably Bayh and Richardson despite their longer tenure) than for the old guard (Clinton, Nunn, Biden).

Of course, these should be viewed as correlations, not necessarily causal effects. For example, two possible explanations are: 1) putting a fresh face on the ticket helps Obama, either because there is less baggage or less of a contrast in national-politics resume length, or 2) Obama will only pick an old guard candidate in the state of the world in which he needs to shore up a weakness (i.e., picking Clinton to end a civil war, or Nunn to add foreign policy experience).
On the GOP side:

Huckabee has the highest conditional probability, and Pawlenty and Jindal are noticeably lower. Interpreting this one is harder: it depends on what aspect of Huckabee one thinks the market is expecting to be appealing (religion, likeability, Southernness, selective economic populism).

Technical note: the bids and asks reported above are actual quotes scrapped this AM- the mids are (bid+ask)/2, rescaled to add to 100 across all candidates.

I try to only follow electoral races in highly digested form -that is, thru the lens of the political prediction markets.

No Gravatar

Excellent formulation by Mike Linksvayer.

Tells a lot about the usage of the prediction market probablities, and how we should market them to people. BetFair, InTrade and TradeSports, are you listening?

I check prices at Intrade most days, which gives me a more accurate and much more concise status update than any amount of time spent reading or watching commentary.

BetFair Digital Odds = BetFair Probabilities

No Gravatar

Odds that Hillary Clinton gets the 2008 Democratic nomination = 1.56 (digital odds taken at 9:15 AM EST)

To get the implied probability expressed in percentage:

  • Take the number &#8220-1&#8243–
  • Divided it by the digital odds (here &#8220-1.56&#8243-)-
  • Then multiply the result by 100-
  • 64.1% = ( 1 / 1.56 ) x 100

BetFair-generated implied probability is not far away from InTrade&#8217-s 62.1%.

Monty Python and the Holy Grail

Psstt&#8230- This present post was prompted by Niall O&#8217-Connor, who puts all his faith in the BetFair instant &#8220-over-round&#8221- &#8212-which indeed doesn&#8217-t add up to the virgin and perfect &#8220-100%&#8221- that Niall is seeking (like the Monthy Python were seeking the Holy Grail). Good luck for your quest, Niall.


The French Taunter:

Your mother was a hamster and your father smelt of elderberries!


External Resource: Interpreting Prediction Market Prices as Probabilities – (PDF file) – by Justin Wolfers and Eric Zitzewitz


NEXT: Implied Probability of an Outcome &#8211-BetFair Edition


Fundamentals of Prediction Markets: Probabilities, Prediction Timescale, and Absolute & Relative Accuracy

No Gravatar

Jed Christiansen outputs the best explainer on prediction markets I&#8217-ve seen in years. Go read it.

– Fundamentals of Prediction Markets
– Different types of Prediction Markets
– Problem #1 – Understanding Probabilities
– Problem #2 – Prediction timescale
– Problem #3 – Assessing accuracy
– Problem #4 – Compared to what?
– Summary – How have the political prediction markets really performed?

Assessing Probabilistic Predictions 101

No Gravatar

Lance Fortnow:

[…] Notice that when we have a surprise victory in a primary, like Clinton in New Hampshire, much of the talk revolves on why the pundits, polls and prediction markets all &#8220-failed.&#8221- Meanwhile in sports when we see a surprise victory, like the New York Giants over Dallas and then again in Green Bay, the focus is on what the Giants did right and the Cowboys and Packers did wrong. Sports fans understand probabilities much better than political junkies—upsets happen occasionally, just as they should.

Previously: Defining Probability in Prediction Markets – by Panos Ipeirotis – 2008

[…] Interestingly enough, such failed predictions are absolutely necessary if we want to take the concept of prediction markets seriously. If the frontrunner in a prediction market was always the winner, then the markets would have been a seriously flawed mechanism. […]

Previously: Can prediction markets be right too often? – by David Pennock – 2006

[…] But this begs another question: didn’t TradeSports call too many states correctly? […] The bottom line is we need more data across many elections to truly test TradeSports’s accuracy and calibration. […] The truth is, I probably just got lucky, and it’s nearly impossible to say whether TradeSports underestimated or overestimated much of anything based on a single election. Such is part of the difficulty of evaluating probabilistic forecasts. […]

Previously: Evaluating probabilistic predictions – by David Pennock – 2006

[…] Their critiques reflect a clear misunderstanding of the nature of probabilistic predictions, as many others have pointed out. Their misunderstanding is perhaps not so surprising. Evaluating probabilistic predictions is a subtle and complex endeavor, and in fact there is no absolute right way to do it. This fact may pose a barrier for the average person to understand and trust (probabilistic) prediction market forecasts. […] In other words, for a predictor to be considered good it must pass the calibration test, but at the same time some very poor or useless predictors may also pass the calibration test. Often a stronger test is needed to truly evaluate the accuracy of probabilistic predictions. […]

InTrade is no psychic -but what if that bit of truth is systematically said BEFORE, as opposed to AFTER.

No Gravatar

David Leonhart in his New York Times blog, last week:

The political prediction markets just went through their version of the dot-com bubble. […]

Intrade’s odds have had a very good forecasting record over the last few years, having correctly called every Senate race in 2006, every state in the 2004 presidential election and all but one state in the 2004 Senate races. The odds also correctly called New Hampshire for John McCain this week and now make him the favorite for the Republican nomination- he is given a 38 percent chance, while Rudolph W. Giuliani is given a 29 percent chance.

Intrade’s executives, as well as the academic researchers who study the site, are careful to point out that its contracts provide only odds, not certainties. An outcome that’s given a 20 percent chance of happening should happen 20 percent of the time — not never. […]

The question I asked yesterday was: What would happen if that warning label were to be sticked on InTrade before each election, as opposed to after each predictive debacle? My bet is that, if you suppress the mention of InTrade&#8217-s magical touch, the Irish real-money prediction markets will be far less appealing to people. They want magic. All of the sudden, InTrade is not a psychic anymore, but simply a forecasting tool of convenience for busy people who don&#8217-t want to check the polls in details. This issue is crucial if we want to be able to define what is the &#8220-prediction market approach&#8221- &#8212-as opposed to the &#8220-betting exchange approach&#8221-.

Give me one reason why the political analysts should follow the US primaries thru the prism of the InTrade prediction markets instead of thru the polls. [My question is still unanswered, you will notice. Which shows to you the embarrassment of the prediction market luminaries (or so they think they are).]

Once the true nature of the prediction markets appears more clearly, it becomes evident that they are not tools for the experts, but tools for the ignorants, rather. Which is great, provided that this is said clearly from the start.