How to Predict a Draw using Historical Football Data?

In this post I am going to look at some methods to help predict a draw with the help of historical football data and football stats. This involves looking at the historical stats themselves and any associated characteristics between such games ending in a draw.

predicting a draw with football data main image

It is important to note that I am not trying to make future predictions here per se, instead, I will be looking for interesting trends and stats which could be developed further and in-turn help make existing football trading strategies (such as lay the draw) even more successful for those who deploy them.

All of the thoughts and methods discussed here are my own and I will be making use of my own (private) football stats database. The actual data I am using will be from the last 5 full seasons (2015/2016 to 2019/2020) of the following mainstream European leagues:

  • English Premier League
  • English Championship
  • Spanish La Liga
  • Spanish La Liga 2
  • German Bundesliga
  • German 2. Bundesliga
  • French Ligue 1
  • French Ligue 2
  • Italian Serie A
  • Italian Serie B
  • Austrian Tipco Bundesliga
  • Danish Superliga
  • Netherlands Eredivisie

How Common are Draws in Football

To start off let’s have a look at how common draws actually are across these 13 leagues and over the last 5 full seasons. To calculate this, as I usually do, I will make use of my SQL Server football database for an accurate answer:

sql for finding draws in football data

As can be seen in the calculations above, for this collection of leagues and seasons we see that 26.73% of all sampled fixtures ended with a draw. But what about how these draws are distributed, next, let’s see which scorelines see the most draws:

most common draw scorelines chart

As can be seen above, 1-1 is by far the most popular draw scoreline with approximately 47% of all sampled games ending this way. After 1-1 we see 0-0 as the second most popular scoreline with approx. 29% of games and 2-2 third with just over 19%. In fact, 0-0, 1-1 and 2-2 combined make up over 95% of all draws seen in this sample of 23,210 football fixtures across the mainstream European leagues!

What characteristics do Fixtures Ending in a Draw Share?

Whilst knowing these stats is very interesting, they don’t necessarily help us in predicting a draw (although they could help in trading “lay the draw” as they show approx 75% of the time we shouldn't see a game finishing in a draw).

In this section I want to look behind these stats a little bit further and see what characteristics the teams and leagues might share and which ones might be more associative of an upcoming draw.

What Leagues see the Most Draws

For a long time now I have been aware that different leagues themselves can each have different characteristics when it comes to scoring goals, this naturally extends to the number of draws seen per league as I wrote about in this blog post about the best leagues for draws some time ago.

range of league draw percentages from old article

As can be seen above (note the key table for the abbreviations), there is a quite a big range in the number of draws seen per league, in fact, I would wager that a 9% difference is bigger than many might expect such a difference to be when looking at the different leagues themselves. Having said this, I wrote this above article a long time ago so let’s have a look based on the current sample of fixtures used for this article.

range of draw percentages table

Interestingly, when looking at the current set of league and seasons above, we see two very interesting things appear in the data.

Firstly, we see that (as in my older article) the range of games ending in a draw can vary dramatically, form a low of 22.46% with the Dutch Eredivisie all the way up to 31.61% with the Italian Serie B league, a difference of over 9% across all games per league!

Secondly, and this is something which appears more than most people think, is that we always tend to see more draws in a given countries second leagues versus the main top-level leagues (e.g. Serie B, 2. Bundesliga, English Championship etc. versus the Premier League, Serie A, Bundesliga and so on).

Therefore, given this disparity is so consistent over the 23210 games in this sample, as well as the older work I have done on the same subject, it is very fair to say that when backing the draw (e.g. value betting) it makes sense to try and focus on the lower / second leagues. In direct contrast to this last statement, when looking for good games for a “lay the draw” strategy it makes sense to focus more on any countries main / top tier leagues!

What Teams see the Most Draws

Now we have looked at which leagues see the most draws we can start looking at the teams themselves, here we can ask if some teams have more of a tendency to draw (or not to draw) than others and to what extent might any differences be.

In the image below we can see that, as the data stands, we see the lowest team draw rates close to 10% of the time and the top teams see draws at around 50% of the time!

number of draws (as a percentage) per league

It is also important to note that when we are looking at averages for this whole data set (over 23k games) we can make sound statistical conclusions due to the large number of fixtures combined. We should, however, in this example, take note that some teams have played only a few games in this sample set (for whatever reason) and as such might be skewing our average.

number of draws (as a percentage) per league (50% example)

For example, if you look at the top drawing teams (as per image above) you will see team ID 4088 (SC Wiener Neustadt) have only recorded 2 games in this sample and since one ended in a draw they have an unusually high draw rate of 50%. To try and counter this I will re-run the query, only this time excluding any teams with less than 50 fixtures played as per this data set.

number of draws (as a percentage) per league (min 50 fixtures)

Now, as per the above table, we now only see teams represented who have played at least 50 games during this sample period, this should make for a more realistic representation of the percentage ranges we could expect to see.

Whilst still looking at the above image, it becomes clear that some of the top teams for draws are topping out in the late 30’s and early 40’s percentage wise when it comes to the number of draws seen. In contrast to this, some of the lowest drawing teams are in the early teens percentage wise. This represents quite a big range (31.41% from lowest to highest) and also helps to prove that as well as different leagues having more or less tendencies to produce a draw so do the teams within those leagues as well!

How Does League Position Affect Draws

Next, I want to look at how the league position of a team going into any given game might affect the likelihood of a draw occurring. Before getting into this section too deeply I need to point out I don't have league positions recorded for all teams in this full sample set, this means the below data is for 7327 games out of the full sample of 23,210 fixtures (still, however, enough for a meaningful result).

top and bottom 5 draws based on teams league standing

As can be seen above, when looking at league position we start to see a much smaller range of percentage values from the league positions with the most draws (close to 30%) and those with the lease (closer to 20%). What does stand out for me at least is that we see 3 of the top 5 league positions (1, 2 and 4) amongst the positions with the lowest draws whereas (with the exception of position 6) the league positions with the highest draws are in the teens and twenties, much higher on average!

If we move on and look at a table of the full range of draw rates to league positions we see, with the exception of the highest positions having a slightly lower draw rate, that broadly speaking the trend across all positions is very flat:

predicting draws based upon a teams league position - table and graph

As per the above graph (even with the noted exception of the very top few positions) there is a very flat trend here, enough to suggest that the league position of a team is not really an area of interest when assessing the likelihood of a draw (potentially providing opportunities for laying the draw on well priced favourites in leagues with less tenancies to produce a draw?).

How Can we Apply this to Football Trading

Now we have seen some stats regarding angles as to where we might start seeking draws, we can start looking at how this knowledge can be applied to football trading. I wrote a detailed article last year on football trading stargates and, if your haven't already, I would advise reading this before continuing on here.

As I mentioned at the very beginning of this post, the aim when predicting draws isn't to predict the future in itself (e.g. a game ending 3-2), instead, we need to take this understanding of trends and use it to help find an edge when applying a trading strategy such as those discussed in the aforementioned article. Trading strategies are essentially mechanisms in their own right and to make then deliver long-term profits we need to be making good decisions and applying them to the correct games (for example, focusing more on higher league games and lower league position teams when deploying a lay-the-draw strategy as per the data reviewed earlier on).

With this understanding of the need to apply a trading strategy to the correct games, we can move on to the issue of finding value. As a trader you will inevitably become aware that in certain games the odds on a draw will differ and, despite what you might want to think (conventional thinking), lower odds on the draw don't necessarily mean the game will finish in a draw the majority of the time (nor the other way around).

I am conscious this post is starting to get very long now so I wont go into odds in this post (maybe in the future I will do another post on odds concerning draws), but the key point I want to make, when starting to apply your draw based trading strategies is to seek good value odds (and, ideally be a little bit greedy before entering your trade).

Here’s the thing, whilst it doesn't feel very good scrapping a trade because you didn't get the opportunity to enter at the price you wanted, it is better to scrap 9 trades and get 1 winner than get all 10 trades on and end up loosing loose 8 of them (boiled down this becomes a profit vs a loss, the number of games isn’t really relevant on any given trading day, its the profit figure we look at).

Conclusion

With over 23,210 sampled games across the mainstream European leagues it is clear we continue to see trends when it comes to draws. With the understanding that you will see around a quarter of games finishing in a draw continent wide (and, therefore 75% not finishing in a draw) this should be helpful in validating lay the draw strategies as a viable trading mechanism.

My main point when trading the draw is to take this understanding and with the stats discussed in the first part of the post work hard to understand where draws might be more likely. Be sure to also think about combing this trait of seeing draws with the need (or greed) required to find the best prices to help ensure profits – never be afraid to scratch a trade if the price isn’t right, simply move onto a game more suited to your strategy instead and focus on honing in on such games.

And Finally…

I hope you have found this post interesting, I like to piece together such data and break away from the hum-drum (and quite frankly boring) football stats you littered across the web. Digging deeper into stats such as those mentioned above can also help in understanding teams and leagues much better in terms of their performance, in-turn, this knowledge can then used to help develop successful football trading strategies (such as those discussed in my football trading strategies article).

If you are interested in building up you own, professional standard, football stats database and in-turn learning the skills needed to query it and develop such stats for yourself, please check out my detailed eBook course: Supercharge Your Football Betting Stats!

Likewise, if you are looking for more advanced football data, including advanced stats and minute-by-minute in-game events data, please be sure to check out my new Pro-Data Football Stats service.

Free Football Data Downloads

Download Free Football Data CSV Files from betonjames.com

Leave a Comment

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

Scroll to Top