FTS Pod & Blog 24th November – Money Management 3, Kelly.



FTS Pod & Blog 24th November – Money Management 3, Kelly.


If this is the first episode from the series you are reading or listening to, please go back to the beginning and work through them. I try and cover as many topics as possible to help you on your betting journey and get them in an order that makes some sense. To stop the ridiculous notion of people trying to dive in and win at this game and similarly to stop people making it far more complicated than it needs to be, both can be equally damaging.

Please also leave comments below the written version and reviews on your podcast provider if you enjoy this series. Feel free to leave positive or negative feedback via email, Twitter or the FTS Telegram group.

We have been looking at staking and discussed issues around fixed odds betting systems staking and a simple way to have more money on bets that win more often hence smoothing the ride.
With fixed odds betting, we discussed the variance and losing streaks which cannot be avoided no matter what anyone tells you. With a fixed odds betting system, we must/should place every bet and accept that variance as part of the ride. If we miss bets, we are not being true to the system.
When we are trading, we can treat staking slightly differently. Provided we have discipline. We are in total control of when we enter a trade, how we enter it, how much we have on it and when we exit the trade. We have complete control over every element of execution. We do not, however, control the events required for the markets to necessarily do what we want them to – e.g. goals in football.

Usually, we will not need as big a bank in points when trading as we can control the flow of our money better. According to our method and any edge we have identified, we can pick our spots based on the money available, money in the markets and the size of that edge.

The issue with that statement is the “size of that edge”. In sports betting, it is difficult to identify the exact size of any edge at the time of bet placement. We may be able to work it out retrospectively based on known outcomes and prices taken, but we are working in a market where we have market odds. These are still perceived, and hopefully, if doing things correctly, our own odds and these again will be perceived.

We may have a model that has produced odds for us of 1.8. The odds available on the Betting Exchange for that same outcome maybe 2.0. We would want to place that bet and keep placing it. We are not saying we will win; we are betting/trading on the fact that the odds we can bet at offer value on the odds we have generated. If your model is accurate, you will repeat this play over hundreds, and thousands of bets come out in front.

But of course, you will have situations where we generate odds of 1.66 against available odds of 2.0 and 1.8 against available odds of 2.0. Should we treat both of those situations the same?

The answer, of course, is NO.

If the odds are available in both instances are 2.0, the market is giving that outcome a 50% chance of happening.

100/2.0 = 50%

In our first example, we generate odds on the same outcome representing around a 60% chance of the same outcome happening.

100/1.66 = 60.24%

In the second example, our odds generated represent only a 55% chance of the outcome happening
100/1.8 = 55.55%

– it makes no sense to treat both the same.

Numerous staking plans have been devised in gambling to maximise returns when these disparities in our perceived odds of winning vary from the odds available.

John Larry Kelly Jr devised a popular method for staking in 1956. Kelly was looking to maximise long-term growth rates of investments, and this staking formula was taken up by sports gamblers across all sports as well as those who were investing in stocks and shares.

The premise of Kelly is that the bigger your edge, the more you have on. It takes your current betting bank, looks at the odds available, the edge compared to your odds and determines your optimal stake.

Sounds great; however, if you follow Kelly to the letter, there is every chance you will go bust at some stage. My personal view of it is that it is too aggressive, and we will still have a series of losers somewhere. Aggressive staking will see bank depletion; of course, it relies on us knowing our true edge. We don’t. Quite often, the advised stake is a significant portion of your bank.

The Kelly formula is as follows.

Stake = ((Decimal Odds x % Chance Win) – 1) / (Decimal Odds – 1) * 100
Stake = Optimal size of the stake
Decimal Odds = Odds offered by the bookie
% Chance Win = Estimated probability of the bet winning

Let us assume we have a £1000 bank and use our examples above.

Example 1 – Odds available to bet at 2.0, odds we generate 1.66 – 60.24% – (I will round the percentages for ease of calculations)
((2.0 x 0.60)-1/(2.0-1) x 100 = 20

Kelly would recommend we bet 20% of our bank on this bet where we have odds of 1.66 and odds of 2.0 available.

Example 2 – Odds available to bet at 2.0, odds we generate 1.8 – 55.55% – (I will round the percentages for ease of calculations)
((2.0 x 0.55)-1/(2.0-1) x 100 = 10

Kelly would recommend we bet 10% of our bank on this bet where we have odds of 1.80 and odds of 2.0 available.

In example 1, we would be staking £200 (£1000 x 20%) on our selection, and in Example 2, we would be staking £100 (£1000 x 10%).

We have already seen that when betting at odds of 2.0, we could experience losing streaks of 8,10,12 bets, risking 20% of our bank on individual bets. Therefore, it is rather silly.
However, the premise of staking more when we have a bigger edge is one I stand by. When we have a fixed odd system working it out bet by bet can be laborious and sometimes impractical if we have 15-20 bets over a weekend, for example, having to monitor every price. Hence, I operate as detailed in the previous episode.

When trading, the way to trade everything is numerically driven, so I am sat there to execute the trades. I am generating numbers and have a market price I can choose to operate against.

I can use a Kelly-style operation, and as I said on opening, I can decide when I want to enter any trades.

I use my model to generate a price for over 1.5 goals, for example, pre-match.

I will plan to enter that trade after 20 minutes of play – providing nothing has happened, no goals, red cards etc. if I have identified value pre-game, that value will remain in play if no significant event has occurred. My models can now identify by the minute, so they have become a bit more sophisticated as time has passed.

Let’s assume my trading bank is £10,000, and I use “100 points” as a guide. My base stake is £100 per trade.

What I do from here is calculate my edge and run it through Kelly (I include commission rates, so it reduces the edge slightly from the numbers we have calculated in the examples above)

I then apply an increase to my base stake of either a ½ Kelly or ¼ Kelly, dependent on the performance of my chosen trading method and the strike rate. The riskier the method I use, ¼ Kelly, the safer the method I use, ½ Kelly.

As I win, the bank grows, the base stake increases, Kelly increases the stakes accordingly, and the whole thing starts to grow. Once you gather pace and the bank grows, it increases exponentially. My Excel screenshot is below – b = commission rate, p = my odds as an estimated probability of winning, q = my odds probability of losing

In our 1.66 example, Keelyn would advise me to stake 19.67% of my bank. If my base stake was £100 what I would stake is £109.84 instead of £100 based on Kelly edge. If my model were odds of 1.8, then the stake would be £105.10 – lower edge, lower stake. Over 1000’s of bets compounding as you go this works so well.

I may drip that stake in, so if I was entering my 1.5 trade at 20 minutes, I might place bets through to 25 minutes, break up my 109.84 stakes into 3 chunks and place the bets. I trade on the goals. As I profit, my bank grows, the base stake increases, Kelly fractional increases to the base stake increases, and we rinse and repeat.
Of course, all the above relies on me being disciplined in execution, only taking trades with an edge. Notice the “Ignore if Negative” – we do not trade if we have zero or negative edge, unfortunately, something most people do daily.

So, no matter how attractive the staking plan is, if you act like an idiot, you won’t come out in front.

Next time we will look at execution and compounding effects.

If you have any comments or questions, please leave them below this post or email them to support@ftsincome.co.uk.

One Response

  1. Hi Ian.
    Another great edition.
    Is fixed odds better suited to 50% increase in stakes when bank increases 50%?
    Can this be done in trading also? Or better dividing bank after every trade?
    And would you not decrease stakes as in fixed odds?

    Many thanks


Related Posts

Welcome to the next episode of the Pod Blog to improve your approach to football trading....

First, to be on the Right Mental Pathway, you need to look at your WHY.  What...

In the world of sports betting, bettors employ various methods to increase their chances of profitability...


GET OUR FREE 6 Part Email Course

Enter your details below to get instant access to the first email

"*" indicates required fields

This field is for validation purposes and should be left unchanged.