Welcome to The Premier League True Form Table

Analysing form is a tricky business and for a long time, football in general and the Premier League has taken to simply looking at the last few results and branding one side or another the "Form Team" because they've won 4 of the last 5 matches.

Consider a scenario where a top-tier team from the previous season, ranked within the top four, narrowly secures victories with scores like 2-1 against a team from the bottom six, or clinches a last-minute 1-0 win against a team facing relegation. In such cases, questions arise: Shouldn't this team have secured more convincing wins? Perhaps with margins of 3-0 or 4-1? This pattern of close victories might indicate underlying issues, such as an underperforming attack or a vulnerable defence, despite their high ranking. These questions are answered by True Form in a way that simply looking at a pattern of WWDWD can't.

The True form Table addresses this by working in the following way:

Teams are ordered into a range of "tiers", according to their league finishing position the season before. The newly promoted sides are always in the bottom tier, and the top 4 teams are tier 1, the rest of the table are arranged into various other tiers in between. 

There is no bias towards any team who we think didn't perform as well as they could have last season. The tier each team is assigned is based purely on their finishing position from the season before without any adjustments.

Using historical data from thousands of Premier League matches, we can see, for example how well a side like Manchester City should perform against another team such as Crystal Palace. We know how many points they should come away with, how many goals they should score and how many they should concede. This is the baseline number and we track this for 7 matches. As a new match is added, the oldest one drops away and is disregarded for True Form. 

7 matches is quite a long time, perhaps over 2 months, so older results get assigned a lower degree of relevance than newer ones

Using this baseline, we can see who had the hardest run of games, as their baseline figure for league points expected, goals scored and goals conceded will be low for a hard run of games and higher for teams who have had a theoretically easy run.

Comparing the baseline with the teams actual performance allows true form to be revealed.

Questions and Answers

How do you account for historical anomalies like a newly promoted side finishing 7th next season skewing the historical data?

It's true, this kind of thing does happen, but then so does things like sides who finished 8th the season before getting relegated the next season. This can't be weeded out entirely and rather then trying to achieve smoothed perfection by excluding outliers, this kind of oddness is what makes football fun and gives us some more interest in the True Form Model. You throw enough data at a model and these oddities are still there, but get diluted out, just like the 9-0 wins and such, and have a tiny impact on the final table. Besides, when your team does get promoted and goes on a crazy run up the table, at least True Form will recognize your achievement with more than just WWDWW that says you're worse than the "Big 6 Side" that just went WWWWW.

How exactly does the True Form Algorithm work?

We'd really like to share our methods with you, but we need to keep them under wraps to avoid others copying them. It's quite a complex mix of math, but it's basically as we described earlier with a few special adjustments applied that we can't share. We're sorry, but that's as much as we can say. 

Is this service free? The table and the Score Predictor?

At this time, everything on this site is provided free for personal consumption and use by individuals, If you're a business and would like to make use of our data, analytics and prediction suite, then please do get in contact regarding your requirements and we can take the discussion about licensing further.

I checked your numbers and you said Arsenal conceded 8 in the last 7 games, but actually they conceded 9, your data is flawed!!

Our algorithm assigns less relevance to older matches, particularly those that occurred months ago or were followed by a cup or international break. We also round the data presented to you to avoid excessive complexity. For instance, the impact of a significant win or loss might be substantially reduced in our calculations if it happened 6 or 7 games ago.

It's possible that, even after this adjustment, the figures may appear misleading once rounded. However, rest assured that the algorithm accurately processes the correct numbers with precise decimal places. This approach also applies to goals scored and points earned.

We prioritize more recent games as they are more reliable indicators of current performance. While we do use the correct and detailed data, we present goals and points as whole numbers for ease of reading. Consider the difference between seeing expected goals as either 9 or 8.8496 – the rounded figure is simply easier to read.

This rounding method can also explain why teams with similar results might be ranked differently. The actual difference in their performance might be small but significant, becoming less apparent after rounding.

What's the Score Predictor? How does that work?

The score predictor calculates the expected outcomes for both the home and away teams in any given match pairing. It assesses the expected goals that each team is likely to score and concede. The system then adjusts these figures by applying a multiplier or divisor. This adjustment is based on each team's actual recent performance, specifically whether they have been scoring or conceding more or fewer goals than expected. Based on these calculations, the predictor then provides a final result. 

It's crucial to keep in mind that football isn't purely mathematical, so applying some common sense to the predicted outcomes can be useful. Take, for example, a match where Brighton, playing at home, is expected to score 3 goals against a newly promoted team like Luton. However, if Brighton's attack has been misfiring, scoring just 66% of their expected goals, their projected goals scored drop to 2. Additionally, if Luton's defence has been particularly mean, conceding only half the goals expected of them, Brighton’s expected goal tally is further reduced to just 1. 

It's not perfect, as football isn't perfectly mathematical or it would be boring, but it's an interesting guide and seems to hit a surprising amount of correct predictions.