Judging the Risk of a Football Betting Portfolio

Understanding how risk can be controlled when betting is sometimes quite a challenge. Learn today how risk can be reduced & controlled when betting on football. Risk management of a portfolio should be an essential part of your betting strategy.

Please note that this article addresses system betting (= a predetermined selection of bets to bet on the same criteria over a longer period of time).

System betting simply means that you are not judging the risk of any individual bet prior to placing it (this would be ‘value betting’). System bettors identify an overall ‘edge’ or ‘advantage’ that is likely to persist over a longer period of time and then follow-through from the start to the finish of a campaign. Of course, there is ‘value’ involved (a ‘mathematical advantage’), otherwise it wouldn’t work.

The examples in this article are our actual picks for the 2017-18 season that were identified using the HDAFU Tables.

What Exactly is a Portfolio?

A portfolio is a package of bets where extensive analysis has determined the choices (picks). Diversifying the portfolio is an essential part of betting strategy with the aim of reducing the risks of losing.

Please do not confuse the term ‘portfolio’ with ‘best timing of placing bets’. A portfolio is planned well in advance of a weekend (or round, or even season) and determines the assortment of bets that are to be placed later.

A portfolio of bets is therefore a varied group of individual bets, not just one particular match on a particular day.

To judge the success of any particular betting portfolio it is necessary to evaluate the performance of all of its systems together as one, not just the individual group members (bets in just one league or system).

What is Diversification?

Diversification is a technique that reduces risk by allocating bets among various leagues, bet types, and other categories such as times or seasons. The rationale behind this technique contends that a portfolio constructed of different kinds of bets will, on average, pose a lower risk than any individual bet (system) found within the portfolio.

Diversification strives to smooth out unsystematic or anomalous risk events (outcome of matches/leagues) in a portfolio so that the positive performance of some bets (winnings) neutralizes the negative performance of others (losers).

The questions that probably arise right now are: How big should my portfolio be? How much should I diversify?

I will answer these two questions at the end of the article but for the time being let us first look at our portfolio for the 2017-18 Winter League season. I will explain step-by-step how I came up with the choices and what my thinking was behind them. This will probably already answer many of your questions.

Our Portfolio for the 2017-18 Season

For the sake of brevity, I am showing only the first 10 European leagues we used in alphabetical order: Austria, Belgium, Denmark, England, France, Germany, Greece, Italy, Netherlands and Poland.

The overall results you will see would have been similar whichever 10 leagues we chose to illustrate with.

For betting, it doesn’t matter which leagues you choose so long as you have easy access to all of them with the bookmakers available to you. A well-balanced portfolio normally delivers the results as expected.

Here is our portfolio for 2017-18:

Image 1: Portfolio of bets for the 2017-18 winter league season

As you can see it is a conservative portfolio with a good number of low risk and medium risk systems.

As a side note, this article isn’t about how the individual systems (within a portfolio) are picked. However, for the 53 bets in the EPL, you will find a detailed explanation in this article: Finding a System Using the HO/AO Quotient

For the other nine leagues you will just have to trust that the systems were picked in a similar manner to the EPL example.

Probability / Expected Hit Rate / Risk

The terms probability, expectation and hit rate are all closely related, and express more or less the same thing. The main differences are that before a game starts (or a whole system of bets/portfolio is played) the terms ‘probability’, ‘expectation’ and ‘prediction’ are used but, once results are known, these terms are supplanted by actual ‘hit rate’.

The observed distribution of the past (as displayed in the HDAFU Tables) becomes the probability for the future (= expected hit rate).

Referring to image 1 above, there was an expected hit rate of 80% in the Greek Super League. This meant that from every 10 bets placed, on average, 8 were likely to win. Therefore, the longest losing streak when placing 50 bets in a row was 2 (see article: The Science of Calculating Winning and Losing Streaks). This individual system within the portfolio was evaluated as being ‘low risk’.

On the other hand, the expected hit rate of 31.58% in Germany meant that from every 10 bets placed, on average, 7 were likely to lose. The longest losing streak expected when placing all 57 bets in a row was 11. This is an example of low probabilities, or ‘high risk’ classification.

Losing Streak Observed vs. Maximal Losing Streak Expected

The ‘longest losing streak expected’ is not necessarily the ‘longest losing streak observed’.

For example, in the EPL 2012-17 the observed longest losing streak was 6 in a row and this happened 3 times during the previous 5 seasons.

Image 2: EPL HO/AO group 0.603 to 1.396 – Backing the Draw 1st Half

Nevertheless, to judge the risk for the future you have to allow for the worst-case scenario even if it hasn’t happened for a very long time – one day it will happen, believe me!

It is always safer to prepare for the worst rather than relying on luck.

At the end of this article (image 6) you will see that 2 of the 10 systems (Austria, EPL) reached their expected longest losing streaks, and one system even outstripped its calculated (predicted) expectations (Denmark).

Calculating the Expected Maximum Longest Losing Streak

Here’s a screenshot from the EPL 2012-17 HDAFU Table representing the HO/AO quotient cluster group (0.603 to 1.396) for backing the draw. The expected total number of bets and expected hit rate for the forthcoming season are circled in red.

Image 3: EPL 2012-17 HO/AO group 0.603 to 1.396 – Backing the Draw 1st Half

To calculate the longest expected losing streak use the following formula:

Formula longest losing streak

n = number of trials (i.e. total number of bets expected)
ln = natural logarithm
P = probability / expected hit rate (for losing streaks: expected losing rate)
| .. | = absolute value or ‘modulus’

Here’s the formula in action for the EPL example:

The winning probability of 39.62% means of course that the losing probability is its inverse:

100% minus 39.62% = 60.38%

To calculate the longest losing streak use in Excel the following formula:


Read more about the above formula, its use and interpretation in our article: The Science of Calculating Winning and Losing Streaks

The result of our calculations meant that for the 2017-18 EPL season, when backing ‘Draw in the first half’ within an HO/AO Quotient between 0.603 and 1.396, the expected maximum losing streak was 8 losses in a row.

Furthermore, sorting the 2012-17 data falling within this HO/AO Quotient group into rounds revealed that there weren’t more than 4 available bets in any round of matches; sometimes there was only one, and occasionally there were none.

With a losing streak of 8 bets possibly spanning a period of 5 to 6 weeks (= 1.5 months), this system was not likely to be easy on the nerves.

To guard against nerves becoming mass panic, you will need to ensure that your portfolio does not contain too many systems with lower probabilities. Although not very likely, it is quite possible that they may all experience their longest losing streaks at the same time (2017-18 being a case in point, where the season began badly across the board).

Rules of Thumb to Judge Risk

A high yield ALWAYS indicates high risk!

High Yield (over 30%) = High Risk (normally)

…Until it doesn’t!

I have included in the example Netherlands Eredivisie (Backing the Favourite 2nd Half) because the analysis came out with a 75% probability of winning. However, the expected yield of 32.23% was exceptionally high for this probability group and there were only 15 bets expected.

Normally, high probability of winning and high yield do not go together. Therefore, it may have been just an observed anomaly because of the small number of matches on average each year in this group.

Nevertheless, because only 15 bets were expected (of over 500 in the whole portfolio) and with an expected high hit rate, the overall judgment for this particular pick was ‘medium’ although the high yield would have normally marked this system as ‘high risk’.

Moral 1: There are no rules without exceptions.

Next rule:

Low Probability (expected hit rate under 35%) = High Risk

In Germany, the expected hit rate was 31.58% (high risk), whilst expected yield was 29.38% (medium risk)

The combined expectations of hit rate and yield here would perhaps have labelled this system as ‘high risk’ but downgraded to ‘medium risk’ as expected yield was under 30%.

Calculating the overall Probability / Expected Hit Rate / Risk of the whole portfolio

Portfolio compilation and decision-making involved the balancing act of achieving as high an expected hit rate as possible, together with an acceptable yield.

  • The overall expected hit rate of the portfolio should be in the region of ± 50% (of course, if you can get it higher, the easier it will be on your nerves)
  • An acceptable yield is anything in the region of ± 20%

To calculate the overall expected hit rate of the whole portfolio, list your systems like I have done in image 1 (Portfolio of bets for the 2017-18 winter league season)

  1. First you multiply the number of expected bets with the expected hit rate for each system, e.g.

    Austria: 72 times 31.94% = 22.9968
    Belgium: 30 times 86.67% = 26.001
    Denmark: 36 times 41.67% = 15.0012
    …and so on

  2. Then add them all up:

    22.9968 plus 26.001 plus 15.0012 plus etc., etc. .. = 243.2526

  3. Now divide the result by the total number of bets expected:

    243.2526 divided by 505 = 0.4816883 ≫≫ 48.17%

To calculate the overall expected Yield the same calculation applies.

The portfolio example in this article had 505 expected bets over the period of a whole season, with an expected Hit Rate of 48.17%, and an expected Yield of 22.22%.

It was therefore a ‘medium risk’ portfolio.

‘Medium risk’ means that you must be prepared for bumpy rides – substantial ups and downs in your betting bank. But with a sensible staking plan, this shouldn’t mean bankruptcy before the situation picks up again.

Three Scenarios of the same Medium Risk Portfolio with 500 Bets

Despite a ‘high’ hit rate of 48.17% (which isn’t as ‘high’ as you may think) you may experience a betting campaign like this one if you don’t get off to a good start:

Image 4: Profit/ Loss Scenario I for an expected hit rate of 48.17% and 22.22% yield

However, you could get a good start and a very smooth curve over the whole season.

Please note, these screen captures relate to the same portfolio of bets. Nothing has changed: 48.17% probability of winning; 22.22 % expected yield.

Image 5: Profit/ Loss Scenario II for an expected hit rate of 48.17% and 22.22% yield

Scenario 3 has an extremely good start (up to bet number 80) and then gives way to a bumpy ride to the end of the season:

Image 6: Profit/ Loss Scenario III for an expected hit rate of 48.17% and 22.22% yield

Have a think about these profit/loss curve simulations.

The first (unlucky) scenario even has an average hit rate of 52%, much higher than the second and third. However, the second and third are much smoother on the nerves and both have a better financial result at the end.

What I’m trying to say is that you must always remember that you are gambling. Although you are using sound statistics for compiling your portfolio, never lose sight of the fact you are still gambling!

The two very simple rules to negotiate a season with a profit at the end are:

  • If you have a sound portfolio do not give up too quickly.
  • If you have at the beginning a very good start don’t become over-excited and start increasing your stakes more than originally planned.

Always remember, gambling has a lot similarities with rolling a dice; you may have to wait for ages until the sixes start popping up or you have them all at the beginning and then for a long time none.

The Final Performance of the Portfolio

At the end of the season, the portfolio produced the expected result (the overall hit rate was more or less as predicted) although the leagues themselves within the portfolio performed erratically with peaks and troughs during their seasons.

Image 7: Final performance of the 2017-18 winter league portfolio

All three ‘high risk’ systems produced, as expected, very long losing streaks: Austria, EPL, Poland.

Denmark produced a longer losing streak than expected. It started on the 26/8/2017 and continued until 29/10/17 – more than 9 rounds of games! Nevertheless, they recovered to finish with a positive return, although far lower than expected.

I will write more about the final performance of that portfolio in another article where you will also be able to download the whole monitoring spreadsheet including staking and ante post odds movements.

How big should my portfolio be? How much should I diversify?

Betting, in general, is a very unreliable venture and to be able to ‘control’ risk (reduce on losing streaks) a good number of high probability bets is required.

However, the big challenge with football betting is that especially the lower odds (up to 2.5) are often reduced because market demand forces bookmakers to reduce prices for favourites and increase the prices for the opposite bets (underdogs and occasionally draws).

You will therefore find plenty of draw and underdog strategies in the HDAFU tables but unfortunately, these two bet types come normally with very low probabilities of winning (= low hit rate expectation/high losing rate expectation).

And, if you have too many low probability bets in your portfolio and experience several weeks in a row where the favourites are mostly winning, your betting bank may become distressed beyond your comfort levels.

As I have shown you in the example, to get a balanced portfolio of 500 bets requires approximately 10 leagues. If one league doesn’t play according to statistics (e.g. Denmark) another league will hopefully make up for it (e.g. Germany).

Of course, the larger your portfolio, the better.

The more ‘low risk’ you can identify the better.
Reduce on ‘high risk’ as much as you can.

Image 8: Relationship between a systems expected hit rate and its yield

Pay attention that you don’t have too many systems with just 10 to 15 bets in your portfolio as they are much more volatile than systems with 50 bets or more.

However, in the example portfolio, there were only 5 systems with 50 or more bets. It’s not always easy to stick to every rule.

The Rule of Thumb is: 50 is the magic number!

  • At least 50% ‘medium risk’ strategies.
  • At least 50% of the systems within the portfolio with at least 50 bets (for the whole season/year).
  • An overall expected hit rate of the portfolio of around 50%.
  • A minimum of 500 bets in the portfolio = an average of 50 bets in 10 different systems

If you can achieve more than 50 – the magic number – in any of the above equations, then your portfolio should have an even stronger chance of succeeding.

Good luck with your betting!

I hope this article has answered the questions we receive about judging the risk of a system identified with the HDAFU Tables, and will help you to construct a well-balanced portfolio. However, if you are still not clear, then please feel free to ask any questions via the comment section below.

>>> buy your hdafu tables <<<

Last Update: 8 November 2018

Categories:1x2 Betting Betting Guidance Betting Systems Case Studies

17 Responses to “Judging the Risk of a Football Betting Portfolio”