Hi
If you can get past the presenters monotone voice; the first four youtube videos here are a real insight into how an academic and scholar approaches modelling:-
http://www.betfairprotrader.co.uk/search/label/video
Although he applied it to a little known sport; there's no reason why it can't be applied to more mainstream sports too.
The interesting thing, is to see at what point he considered he had a system that would make him money. I think many people would considered they had such a system long before he did.
He goes just that bit further than the crowd in order to arrive at a successful model.
Regards
Peter
PS The above blog is a good read too. The blog owner has a Phd and you can almost hear his mind ticking over as he writes his posts.
Modelling Sports
I first modelled football back in the early eighties and found that modelling does give you a keen view on a market. But the problem with most models is that they show you the market is perfect, or damn near it. So modelling and actually doing are two entirely different things.
I've wheeled in a few Phd's over the years to work on specific problems. I often know the answer and have the data but my maths isn't good enough to join the two together occasionally. What I am looking for is an equation with which I can mess around with the variables. In football I have managed to trace back the likelyhood of a goal right back to individual players.
But the amount of data you need to use it to forecast is immense and when you do this you end up finding that the market is already there. It's amazing how effcient the market can be, even by accident. Because of that the people I wheel in just don't believe it's posssible to make money on gambling markets. So, despite much evidence to the contraty, they don't even try.
I've read this blog a few times and its interesting from an academic view point, but there are quite a few errors in terms of his reading of the markets. I would quite fancy teaming up with a Phd to work on some specific problems but you would be surprised how hard it is to find someone that is really interested in solving them.
I've wheeled in a few Phd's over the years to work on specific problems. I often know the answer and have the data but my maths isn't good enough to join the two together occasionally. What I am looking for is an equation with which I can mess around with the variables. In football I have managed to trace back the likelyhood of a goal right back to individual players.
But the amount of data you need to use it to forecast is immense and when you do this you end up finding that the market is already there. It's amazing how effcient the market can be, even by accident. Because of that the people I wheel in just don't believe it's posssible to make money on gambling markets. So, despite much evidence to the contraty, they don't even try.
I've read this blog a few times and its interesting from an academic view point, but there are quite a few errors in terms of his reading of the markets. I would quite fancy teaming up with a Phd to work on some specific problems but you would be surprised how hard it is to find someone that is really interested in solving them.
The technique is a standard modelling technique. The tricky part - and this is the bit he says he won't go into the details of - is given match data, how do you come up with an accurate estimate of the skill of a particular player. This really is the crux of this problem, and it is also the area where there is the least published information. I say skill, but I also mean match conditions, psychological factors and other less easy to estimate factors. This is possible to incorporate into a mathematical model like his - its just difficult.
Really interesting, thanks for posting that. I do wonder how many students/lectures at my university are involved in betting...
Anyway this made me laugh..
Anyway this made me laugh..
Ah, ah, I know there are charts, moving averages and what have you, in your chosen trading software but it all lags the here and now. Trading on data, so many ticks ago is worse than trading on any prediction of the current tick.
Money enters the back and lay sides of a spread randomly and for no particular reason. There is nothing to predict.
Maybe now you might begin to realise where the profit from trading comes from; licensing trading software!
But there in shows the point I was trying to make above. If you look at the market with academic eyes it shows you one thing, but I could quite happily trade from the touch prices and charts.LinusP wrote:Really interesting, thanks for posting that. I do wonder how many students/lectures at my university are involved in betting...
Anyway this made me laugh..
Ah, ah, I know there are charts, moving averages and what have you, in your chosen trading software but it all lags the here and now. Trading on data, so many ticks ago is worse than trading on any prediction of the current tick.
Money enters the back and lay sides of a spread randomly and for no particular reason. There is nothing to predict.
Maybe now you might begin to realise where the profit from trading comes from; licensing trading software!
I have also earnt much more from doing it than selling things to do it, but I realise the opposite is true for most people on the sell side of the market.