August 12, 2015

How to Bet: My 0.02 Units on Staking by @CrazedAlchemist

How To Bet: My 0.02 Units on Staking

It would be impossible for me to write at any great length on the subject of gambling without talking about staking. Whilst I like to find bets from a very structured and statistical viewpoint, I am far more flexible in my approach to staking. This does not mean that I am not very consistent once I have decided on a staking strategy, but more that I am aware that staking is in part, always about how comfortable you are with the bets you’re placing, not just with the size but staking more or less on certain types of bet at different odds levels. You want to optimise your staking but you want to optimise it for your own long-term profitability and not necessarily the theoretical optimum, which will not necessarily take in to account the psychology of placing hundreds of bets over a prolonged period and the need to not become disheartened part way through the process.  Some will say that if you need to stake sub optimally then maybe you do not have the right mind set for serious betting. I disagree with this point, we do not ask F1 drivers to drive without helmets just to save weight. I will always exchange a bit of ROI for a touch of sanity providing I am still making an amount of money that makes the effort worthwhile.

There are five things that we want from any staking plan: consistency in approach, bank preservation built in, consistently increases stake sizes as your bank grows, the stakes vary in relation to the odds and finally stakes vary in relation to the amount of value the bet has. Consistency means that we avoid any emotion contaminating the staking process, we don’t want to be staking more on bets we have become emotionally attached to. Built in bank preservation is helpful because the calculation of betting banks with any accuracy is a difficult thing so if our staking has some built in protection it can help if we have miscalculated. We want to make sure our stakes grow as our betting bank grows so that we can accelerate the process. We want to vary stakes according to the odds because we want to be staking more on things the more likely they are to happen in order to optimise our return on our capital. Finally we fairly obviously want to stake more on bets the greater the value the bet is. It is for all these reasons that the Kelly staking method is the optimum because it incorporates all these factors. Before I just end the article there though, let’s look at the staking methods that might be used prior to arriving at Kelly and why Kelly is not for everyone.

First we have level staking. I still use level staking for every new model or strategy while I am in the testing phase. Any betting strategy that can’t make a profit from level staking will not be able to make profit using any other staking system in the long term. While paper trading is OK, I prefer to bet very small amounts as it also helps to actually go through the motions of placing however many bets your new system generates and it always matters more when there is skin in the game, make yourself care more about the success or failure of new systems without doing any serious financial damage. If you have any doubts about your ability to find value bets or maintain discipline when betting then I would still always recommend small level stakes bets while you get used to the ups and downs and process of betting regularly. So level stakes is both consistent and easy to apply but falls short in all other areas.

Next we have betting a particular percentage of our betting bank e.g. 1%. By betting a percentage we can ensure that no matter how small our bank gets it will not disappear altogether. It will also increase our stakes as our bank increases. I don’t necessarily agree with updating the stake size after every single bet and I would still try and keep staking to round numbers where possible so that bookmakers do not necessarily become curious about your £47.33 stakes. Over the years when I have done this I used to increase my stake sizes after a 10% increase in betting bank and then only decrease the stakes if my bank fell back to the previous level attained. This increase is completely arbitrary and I have heard of people only adjusting when the bank doubles or increases by 50% etc. Again I would do what feels right rather than trying to fathom the optimal settings for an already suboptimal strategy.

Now instead of betting a percentage of our betting bank we could instead bet to win a fixed percentage of our bank say 2%. Now we have the benefits of the previous options along with staking more on the more likely outcomes and less on the less likely ones. This will smooth out our profit line and I find it particularly psychologically appealing in a betting system where there will be a wide range of odds bet on. An example would be tennis where it would not be uncommon for me to bet on prices ranging from 1.20 – 5.00 and I know that the market is efficient enough for the prices to be a reasonable reflection of the probability of an outcome happening but still leaving me with some room for an edge. It also has the added bonus of taking advantage of a favourite long shot bias if there is one. If the bias exists then you will be staking more heavily when the odds are statistically more in your favour even before you factor in your edge. This is particularly helpful if you are betting a relatively new system that you are unsure of just how big the live betting edge will be relative to the back testing established edge. Of course should you have found that your system has no edge when betting short odds on favourites and all your profits are coming from long shots this will do you no favours. Hopefully such a pattern will emerge before you go live with your betting of the new strategy.

Now if you want to introduce a value element without going down the Kelly route there are no hard and fast rules. You could create value bands so say if you make a bet 3 – 6% value you bet to win 0.5% and if it falls in the 6 – 9% value range you bet to win 1% and onwards. Already you can see the issue that might arise should your model not generate predictions that are at least mildly in line with bookmaker’s odds. If your model can throw out numbers suggesting say a draw in a football match is 60% likely to happen and there is no suspicious element to the match then you are going to be having a very large bet. This exaggerated example highlights the issues we can encounter when we come to Kelly. Alternately you might bet to win a fraction of the value you make the bet. So say you make a bet 5% value and you use a fifth fraction for your system you would bet to win 1% of your bank. In this scenario again the bigger the value the bigger the bet but the fraction you use has not been calculated to optimise the staking but more to generate a stake size you are comfortable with or indeed that allows you to place all the bets you may have happening at the same time.

There are hundreds of Kelly calculators online and there has also been a fair bit of work done on Kelly betting when you need to place multiple bets simultaneously and Kelly staking suggests staking more than your whole bank when you add them all together. Very few betting websites advocate using full Kelly but a fraction of the stakes advised, which is largely to make it feel more comfortable for the punter coming from level stakes betting and because it is recognised that it is hard to build a model you can have complete conviction with, regarding its predicted probabilities. Factoring in some doubt, even though Kelly will not let you go bust, helps to ease what would be a sometimes nerve wracking journey with full Kelly. With models I use that have proven themselves over a large sample of bets, I like to use a restricted full Kelly. So I may cap the maximum stake at 3% of my bank for example rather than using a fraction of the suggested amount. This is a nod to the fact that if a model makes something huge value, then there is every chance you may have missed something significant that models are not that good at catching. My longest standing models have actually been through all the above staking stages at some point and I would not steer anyone away from any of them. It is hard enough to find an edge and stay the course of betting it relentlessly without stressing yourself further by trying to conform to how the expert betting community expect you to stake if you do not feel comfortable with it. I would advise that if you are not happy quantifying the approximate edge from one bet to another but are happy that the system has merit overall that you steer clear of Kelly and try to adopt some of the other aspects in to your staking instead. I want you to be in a position where you can say your edge finally ran out rather than you ran out of money or your nerves got so bad you had to stop.

A final note on betting bank calculation. I am not really an authority on this but I can pass on the method that I use to calculate a level stakes betting bank. I look at the average odds range I will be betting at and for the example we shall say it’s around 2.0. I then ask myself what the minimum edge I would require before betting would be so let’s say 3%, which means we are betting when we think the odds should be closer to 1.89. In this scenario full Kelly suggests 6% of your bank.  A quarter Kelly is probably the most often recommended fractional Kelly, which would in turn suggest a 1.5% stake. Consistently betting these types of bets would mean betting 1.5% over and over again and that could be argued to be akin to betting 1 unit of a 66 unit bank. Call it 70 units between friends and you have an adequate yet pretty arbitrary bank calculation fudge. Using the same method but betting on golfers around 67.0 average price that you make 34.0 on average and you get a betting bank around 270 units. This of course falls apart slightly, as of course each way betting on golf is actually two bets not one and at different odds, so you can see my method unravels fast. This method at least gives you a starting point as long as you try to underestimate your edge and have a rough idea of the average odds. Hopefully another will be able to provide greater insight to the subject than I can.

By @CrazedAlchemist.

Join Eastbridge Sports Betting Brokerage today!

Leave a Comment

Your email address will not be published. Required fields are marked *