The previous post described levels of gamblers. This post aims to illustrate the difference in success between level 3 betting and level 4 betting.
In truth we will be using level 5 betting to illustrate the point, because the probabilities calculated are done via a model. The sample is the 760 Premier League matches from the 2009/10 and 2010/11 seasons.
Level 3: Betting on the event with the highest probability of occurring.
Matches Played: 760
Bets Void: 6
Bets Won: 369
Bets Lost: 385
Total Profit/(Loss): (£335.09)
Note: Void Bets refer to outcomes which had the highest probability of selection e.g. 40% chance of Draw, 40% chance of Home Win, 20% chance of Away Win. These matches were void. £10 Stakes were used.
Level 4: Betting on the event with the highest expectation.
Matches Played: 760
Bets Void: 15
Bets Won: 306
Bets Lost: 439
Total Profit/(Loss): £935.39
Note: Void Bets refer to matches where there is no positive value in all three outcomes. £10 Stakes were used.
Things to note here are that the second method produces a significant profit, even though the percentage of bets won in level 4 is less than in level 3. By seeking value instead of seeking the most likely result we are able to make more money!
Thursday 9 June 2011
Tuesday 7 June 2011
Levels
In betting there are levels of gamblers.
Level 1 is a gambler who simply bets at random. They have no thought process when selecting a bet, they simply choose at random. For example, lucky dip in the lottery, picking a horse out of a hat, or at very best based on their favourite name. Now this plays to the house clearly, because in the long term the odds favour them. The profit is made by non-frequent gamblers who “strike it lucky”.
Level 2 is a gambler who bets on partisan. Very often a level 2 gambler will make bigger losses than a level 1 gambler. By partisan betting, I mean that the gambler selects their favourite player, team, horse etc to win. For example, football fans who bet on their team to win, or their rival team to lose. The phenomenon of reverse-partisan betting also comes under this umbrella, where the gambler selects against what they want to happen so that they hedge their utility, i.e. if my team wins I am happy, if my team doesn’t win, I make some money so I am also happy. In general, only a lucky few make money betting this way.
Level 3 gamblers are those who bet on probability, they decide what they believe is most likely to happen and select this outcome. At first sight this doesn’t seem like a bad way to go, but there is often a disregard for the price being offered. There is a large amount of bias in this method too, with level 2 partisan thinking coming into play when these on the spot probabilities are calculated. Often the gambler will not even think of the selection in terms of probability. For example, “Chelsea vs Blackpool, Chelsea will definitely win” compared to “Chelsea vs Blackpool, Chelsea would probably win 90 times out of 100”. Either way, this method rarely goes against the favourite and this plays right into the bookies hands.
Level 4 gamblers are those who bet on value, they decide what the probability of an event is and then look at the prices being offered to see if the price is understated or overstated. For example, I think the Haye vs Klitschko fight is 50/50, therefore I would look at evens as the base price and bet on the fighter priced above evens, in this case David Haye. In general, this is as sophisticated as your gambler gets. They won’t calculate the probabilities any differently to a level 3 gambler, the only difference is that they won’t select the most likely event, rather they will take into account the prices being offered and go for the best value. There are types of level 4 gamblers, some will look to back the most likely outcome and only back it if the price is good enough, if it isn’t good enough then they will not bet at all, or strangely they will place the selection in an accumulator to get a perceived higher value. This flawed logic will often lead to a string of poor value bets rather than one good value bet.
Level 5 gamblers actively create models in an attempt to accurately assess probability. They then use these probabilities to judge the value of a particular bet and select based on the highest expectation. The largest difficulty with this method is the ability to incorporate all the variables into a model. This is in practice impossible without introducing bias. For example how do you assess the impact of a Kolo Toure drug ban on Man City’s probability of winning a football match. The result is a series of crude models which constantly adapt to absorb past data and weight them accurately. This means that the models are a representation of the past rather than an accurate forecasting method. Nevertheless through experience of using models you are able to identify trends and evolve the model to accurately represent probabilities.
There is a level 6 gambler, but there is an argument of whether this level of gambler counts as a gambler at all. They seek arbitrage. They will identify opportunities where they can make risk free profit, these opportunities are not as rare as you might think but are obviously rarer than the level 5 good value bets. So the amount of profit will be limited compared to the possibilities of level 5 gambling but this is at the benefit of risk free profits.
Level 1 is a gambler who simply bets at random. They have no thought process when selecting a bet, they simply choose at random. For example, lucky dip in the lottery, picking a horse out of a hat, or at very best based on their favourite name. Now this plays to the house clearly, because in the long term the odds favour them. The profit is made by non-frequent gamblers who “strike it lucky”.
Level 2 is a gambler who bets on partisan. Very often a level 2 gambler will make bigger losses than a level 1 gambler. By partisan betting, I mean that the gambler selects their favourite player, team, horse etc to win. For example, football fans who bet on their team to win, or their rival team to lose. The phenomenon of reverse-partisan betting also comes under this umbrella, where the gambler selects against what they want to happen so that they hedge their utility, i.e. if my team wins I am happy, if my team doesn’t win, I make some money so I am also happy. In general, only a lucky few make money betting this way.
Level 3 gamblers are those who bet on probability, they decide what they believe is most likely to happen and select this outcome. At first sight this doesn’t seem like a bad way to go, but there is often a disregard for the price being offered. There is a large amount of bias in this method too, with level 2 partisan thinking coming into play when these on the spot probabilities are calculated. Often the gambler will not even think of the selection in terms of probability. For example, “Chelsea vs Blackpool, Chelsea will definitely win” compared to “Chelsea vs Blackpool, Chelsea would probably win 90 times out of 100”. Either way, this method rarely goes against the favourite and this plays right into the bookies hands.
Level 4 gamblers are those who bet on value, they decide what the probability of an event is and then look at the prices being offered to see if the price is understated or overstated. For example, I think the Haye vs Klitschko fight is 50/50, therefore I would look at evens as the base price and bet on the fighter priced above evens, in this case David Haye. In general, this is as sophisticated as your gambler gets. They won’t calculate the probabilities any differently to a level 3 gambler, the only difference is that they won’t select the most likely event, rather they will take into account the prices being offered and go for the best value. There are types of level 4 gamblers, some will look to back the most likely outcome and only back it if the price is good enough, if it isn’t good enough then they will not bet at all, or strangely they will place the selection in an accumulator to get a perceived higher value. This flawed logic will often lead to a string of poor value bets rather than one good value bet.
Level 5 gamblers actively create models in an attempt to accurately assess probability. They then use these probabilities to judge the value of a particular bet and select based on the highest expectation. The largest difficulty with this method is the ability to incorporate all the variables into a model. This is in practice impossible without introducing bias. For example how do you assess the impact of a Kolo Toure drug ban on Man City’s probability of winning a football match. The result is a series of crude models which constantly adapt to absorb past data and weight them accurately. This means that the models are a representation of the past rather than an accurate forecasting method. Nevertheless through experience of using models you are able to identify trends and evolve the model to accurately represent probabilities.
There is a level 6 gambler, but there is an argument of whether this level of gambler counts as a gambler at all. They seek arbitrage. They will identify opportunities where they can make risk free profit, these opportunities are not as rare as you might think but are obviously rarer than the level 5 good value bets. So the amount of profit will be limited compared to the possibilities of level 5 gambling but this is at the benefit of risk free profits.
Wednesday 1 June 2011
What If?
I often get asked in various forums, What If....?
In this case I shall be exploring, What if I backed the outsider in each Premiership game?
Well if you put £10 on every outsider in the 760 games from the 09/10 and 10/11 seasons:
Total Matches: 760
Bets Void: 15
Bets Won: 144
Bets Lost: 601
Total Profit/(Loss): (£704.17)
The void bets refer to matches where there is no outsider, i.e. the odds for the draw are the same as the odds for the outside winner.
So a big fat loss.
However, let's split this up a little, 09/10, 380 games:
Total Matches: 380
Bets Void: 8
Bets Won: 67
Bets Lost: 305
Total Profit/(Loss): (£647)
Which leaves the loss for the 10/11 season at £57.17.
It may appear clear then that backing the outsider is fruitless.
To find the fruit we must take a look at the home, draw, away splits:
Away Outsider:
£907.50 loss in 09/10
£110.67 loss in 10/11
£1018.17 loss in total.
Draw Outsider:
£49.50 loss in 09/10
£7.00 loss in 10/11
£56.50 loss in total.
Home Outsider:
£310.00 profit in 09/10
£60.50 profit in 10/11
£370.50 profit in total.
So if you look beyond the headline numbers, there is evidence that backing the home outsider is profitable!
In this case I shall be exploring, What if I backed the outsider in each Premiership game?
Well if you put £10 on every outsider in the 760 games from the 09/10 and 10/11 seasons:
Total Matches: 760
Bets Void: 15
Bets Won: 144
Bets Lost: 601
Total Profit/(Loss): (£704.17)
The void bets refer to matches where there is no outsider, i.e. the odds for the draw are the same as the odds for the outside winner.
So a big fat loss.
However, let's split this up a little, 09/10, 380 games:
Total Matches: 380
Bets Void: 8
Bets Won: 67
Bets Lost: 305
Total Profit/(Loss): (£647)
Which leaves the loss for the 10/11 season at £57.17.
It may appear clear then that backing the outsider is fruitless.
To find the fruit we must take a look at the home, draw, away splits:
Away Outsider:
£907.50 loss in 09/10
£110.67 loss in 10/11
£1018.17 loss in total.
Draw Outsider:
£49.50 loss in 09/10
£7.00 loss in 10/11
£56.50 loss in total.
Home Outsider:
£310.00 profit in 09/10
£60.50 profit in 10/11
£370.50 profit in total.
So if you look beyond the headline numbers, there is evidence that backing the home outsider is profitable!
Subscribe to:
Posts (Atom)