Other Sports

Some answers

Posted By: John May
Date: 4 Jun 02, 4:37 pm

In Response To: Re: UK Spread Betting Poisson Distribution (John Butterworth)

(Note: I refer to index betting rather than spread betting, since in the US that is a completely different form of wagering)

Mathsboy has stated that he does not think that they do conform to Poisson because of the nature of how teams play when a goal is scored, he mentions that is something he got from you.

My opinion on this comes Jacques Black's Spread Betting To Win, pg 113,

"Is the distribution of goals through a game random?

The empirical answer is that they are not. In 1995, I analyzed the times of goals scored in 500 games in the Enlgish and Scottish premier leagues. This analysis indicated that approximately 40% of the goals were score in the first half, and 60% in the second half......although there was a clear discontinuity at half-time, a second conclusion of my analysis was that, within any half, goals appear to be randomly distributed throughout time."

If it is true that football does not conform to Poisson, because of the nature of how teams play when a goal is scored, then how can the number of Quarterback Sacks conform to Poisson? (Example on page 161 of Sharp Sports Betting) If a Quarterback is getting sacked then surely the team offence would have to adjust and play differently to avoid that situation? Finally is there a way to work out a betting advantage directly from the spread without having to check other fixed odds terms.

I don't know enough about American football to be able to comment on the Quarterback Sacks issue. However, regarding soccer, the issue of tactical changes is one of three plausible causes Black suggests for the discontinuity between halves, the others being changes in the context of the game and physical tiredness among players. I'm sure you could think of others-psychology would seem important also. Without a more thorough data analysis the precise cause/s remain a source of mystery.

Finally is there a way to work out a betting advantage directly from the spread without having to check other fixed odds terms.

Yes, you just do the handicapping yourself. Measure the average goal distribution in the recent performances of the protagonists, adjust for factors such as home/away performance, team formation, key injuries etc. Mess about with the numbers till you get a forumula which closely predicts the scoreline of prior games and test it against a future data sample. Once you have a successful predictive model you can look for discrepancies between your formula and the posted lines of the index firms.

But you knew all that right? I'm not sure if I'm getting at what you are after.

I am now going to ask what might be a stupid question.

In financial trading I can use volatility and other historical price movements to my advantage. As an example, low volatility means expect high volatility and with high volatility expect low volatility. One way of doing this is to use Bollinger bands which are constructed using 2 x the Standard Deviation of an n day moving average of a closing price. By tracking the width of the bands over a period of time you can see periods of low and high volatility. This method not only helps with the timing but also exposes optimism and pessimism from the public. If a sports spread bet firm moves the spread dynamically before an event and even during the course of an event has anyone thought of charting the mid price of the spread and looking for optimism / pessimism or oversold / overbought conditions that you could use to your advantage?

Far from being a stupid question this is probably the most complex question ever asked on this site. Let me make this clear: you are way out on your own on this one. This type of mathematically soophisticated approach just does not get discussed generally in betting circles, at least in the UK.
I have actually thought about it but that is it. Data mining for proof of what you are asking is actually very easy to obtain, for this reason: betfair.com, a p2p betting exchange, offer index betting. A chart is created in real time on every event and wager, tracking the pricing fluctuations. Naturally the type of biases you are looking for would stick out like a sore thumb.

Finally could you comment on using stop losses on sports spread bets? I suppose that they move the spread wider if you opt for this.

If you mean limited risk spread betting accounts, that is right, they raise the spread to compensate. Maybe some kind of exploitable non-linearity exists here for the astute bettor.

Green Baize Vampire's Archive

Messages In This Thread

UK Spread Betting and Poisson Distribution -- John Butterworth -- 3 Jun 02, 12:12 pm

Interesting question -- John May -- 3 Jun 02, 12:52 pm

Re: UK Spread Betting Poisson Distribution -- John Butterworth -- 4 Jun 02, 4:21 am

volatility -- ricky -- 4 Jun 02, 5:15 am

Re: Curious -- John Butterworth -- 4 Jun 02, 5:41 am

Some answers -- John May -- 4 Jun 02, 4:37 pm

You need to test whether total goals is distributed as a Poisson -- Math Boy -- 3 Jun 02, 3:37 pm

Underlying distribution -- John May -- 4 Jun 02, 4:25 am

Was this written by John May? -- Math Boy -- 4 Jun 02, 7:54 am

A guess -- Rusty Starfish -- 4 Jun 02, 8:03 am

hockey totals -- ricky -- 4 Jun 02, 5:17 am

soccer vs hockey -- Math Boy -- 4 Jun 02, 7:59 am

Other Sports is maintained by Pi Yee Press