This article expands on the work done in our Talisman Theory article. Though I’ll mention this a fair bit, if you haven’t read it I recommend going back to it before you start on the below.
We are indebted to Mitchell Stirling (@MitchellSt) for working on this with us – thank you so much!
“Last time on Talisman Theory”
So, we didn’t quite expect to be here again.
In the first article on Talisman Theory, we looked through the data for the top scoring player at each club through several steps, and decided that removing appearance was a better gauge of whom a team’s ‘Talisman’ could be. The final outcome is below:
What became clear was that Talisman theory is most applicable for your 3rd or 4th midfielder, and that identifying this guy early was the key to getting the maximum output from him.
However, what we were missing was emergence data, which is to say we didn’t have the capacity to go through the FPL data and identify just when this Talisman is likely to come through.
Luckily, this is where Mitchell kindly stepped in…
How to find a Talisman
The last article introduced the concept and identified a few players who scored a disproportionate amount of a team’s FPL points (excluding apps and clean sheet points).
At the start of the season anyone who gets a few braces and assists for a mid-table or lower team will fall into this category.
However, how can we be more confident that someone who has 30-40% of a team’s non- app / clean sheet points is a sustainable asset?
To do this, let’s look at some of the heroes of the previous article and a few people who looked good for a few weeks in Autumn last year.
Any player playing for a non-big six side is, with some exceptions like Jamie Vardy, likely to score fewer points in those games (vs the top 6). So, it isn’t that much of a concern if they blank or their scores are lower when the going gets tough.
To examine this, we effectively created a “Talisman detector” (in reality, a six week moving average).
We did through looking at how many points a player scored over a six-week period from last season so that, for every Gameweek from 6-38, we can see how well they’ve been doing at different points in time, and chart their progress throughout the campaign.
To understand performance relative to fixture strength, we’ve looked at the fixtures over that period and used final league position to assess fixture difficulty.
(Of course, we won’t have final league positions for the 2018/9 until May so you’ll have to use current position in the league (which after 10 GWs looks remarkably like 38 GW for most teams https://www.11v11.com/league-tables/premier-league/01-november-2017/ ) or look at a spread betting index for an up to date view on this. Better yet, use FPLs own “FDR” data which takes home and away fixture difficulty into greater consideration)
Looking at the data
Looking back over last season, we can see that a true talisman, as you’d expect given their usual status as a third or fourth midfielder in our squads, score fewer points per game when they have more difficult match-ups. However, when the difficulty of the fixtures eases up, they start to score more again. Let’s look at some examples where that does happen, in Riyad Mahrez, Xherdan Shaqiri, Pascal Gross, Aaron Mooy, Marko Arnautovic and Jonas Lossl.
On the following charts, the Y-axis is the 6 week average number of points a player scored, which is why it starts at Gameweek 6.
The X-axis shows the Gameweek periods.
Where team data is shown, this still uses the Y-axis but shows the average league position of the opposition faced during the 6 week period – the lower the line to 1, the better the average finishing position of teams they were facing (i.e. 20th place isn’t good in the actual league, so a higher rating means easier fixtures).
There’s a lot going on in this busy chart, so we’ll unpack some of the individual players shortly.
The point that most of these players had sustained periods where they were averaging 4-5 points a game over a six-week period which equates to 150-190 points for the season.
This begins to work towards backing up the conclusions in the original Talisman Theory piece – identifying these guys at the right time is key. For example, buying in Arnautovic at the right time and catching those peaks would have clearly elicited gains for your team last season.
Let’s start to look at these players starting with Lossl of Huddersfield – a randomly selected example from the array of keepers whom stubbornly remained in the Talisman data (until we artificially removed them by discounting keepers/defenders!).
Whilst Lossl was remarkable consistent all season, as the fixtures got easier (average league position faced increased) the points tally increased as well. Here is the same graph with Mooy, the offensive talisman at Huddersfield with clean sheets/defender points removed, grafted on, in addition to Huddersfield fixture data.
Mooy tailed off towards the end of the season but had a 15 point haul against Watford in GW 18 that really boosts his numbers overall. The fact is that, if you got him at the right time and held him for the right period, you’d have benefitted from his spike in points return. This was sufficient for him to run out as the team’s Talisman, despite tailing off slightly
This proves somewhat a correlation between fixture difficulty and the need to purchase a team’s Talisman in that period. Perhaps, then, Talismen are easier to identify than we thought…
… well, no, not quite.
False Dawn Talismen
For some, there is no return to form after a spike in data which may deceive us into thinking they’re a Talisman.
Let’s name some of these to bring them front of mind – Richarlison, Eric-Maxim Choupo-MistakeMoting, Tammy Abraham, Jermain Defoe and Aaron Ramsey. For a variety of reasons, these guys came to the fore for their sides at certain points in the season but faded badly – a mixture of injury (Ramsey), form fading (Richarlison) and generally being a bit useless save for a 7 minute goalscoring appearance (Choupo-Mistake) – which put paid to their pretensions of being anointed Talisman.
Let’s take a look at this data over time:
So what makes these guys “false dawn” Talismen?
The key is understanding that they provided a short spike early doors (see around 6-9, or 13 in the cases of Ramsey) before fading into near-obscurity. There are many factors behind why this happens, from unsustainable conversion rates to external shocks changing the environment – for example, Marco Silva being distracted by Everton’s interest.
This isn’t to say that these guys are invalid picks during their “hot period”; it’s just that in the context of Talisman Theory these individuals don’t have the staying power for a variety of reasons to be worthy of Talismanic status.
It’s clear in the graph that the likes of Richarlison and Choupo-Mistake started well – but how can we see that was unsustainable?
Let’s compare the latter with his side’s actual Talisman, new Liverpool signing Shaqiri:
You can see how Stoke had three periods with a decent run of fixtures, peaking around GW 15, 20 and 29.
Shaqiri peaked in 15 and 29. Choupo-Moting, in contrast, peaked in Gameweek 9: the 15 points he scored against Man Utd the haul that accounts for a lot of that. After a brief increase around GW12, the story of the rest of the season is declining returns that didn’t pick up when fixtures became more favourable, until he eventually wasn’t even appearing as a sub.
In effect, though Shaqiri did dip below Choupo at one point, the reality was that he was always the better pick on average.
Mitchell thinks that between Halloween and Guy Fawkes Night (Gameweek 11) is a good point to be able to identify who may be emerging as a Talisman. By then, you should have a sense who will ebb and flow with the tide of harder and easier fixtures, plus have sufficient data to enable you to plan for the second half of the season by Christmas.
Why is this useful?
Identifying a Talisman, then, is a multi-step process.
We need to be conscious of the confines of Talisman Theory: with appearance and defensive points removed, barring a phenom like Mo Salah these individuals are likeliest to be a midfielder playing for a mid-tier side that sport a fairly competitive price. They’re therefore likeliest to be your third or fourth midfielder, using two of last season’s top three (Xherdan Shaqiri and Pascal Gross) as examples.
Understanding how fixtures and players interact, as in FPL generally, is key. But this is especially acute in the case of Talisman Theory.
This is illustrated by a mind-blowing finding of this research, which is this:
Looking at last season’s data, if you had set a space in your squad for a ~6-7.0m midfielder last season (coupled with an extreme run of fortune), you could have started with Groß in GW1-9 and moved through Shaqiri (10-15 and 24-31) and Arnautovic (16-23 and 32-38) to score 246 points for one slot by playing the fixtures for the season.
Incredibly, this means that you could have outscored Harry Kane (217) and Raheem Sterling (223) if you played that midfield slot perfectly – that’s more than any other player bar Mo Salah. This surely shows the value of identifying the Talisman at the right time, and underlines the importance of this kind of player to your FPL team.
Now this would be incredibly unlikely (near impossible) that you could do this perfectly or even close to perfect through the season.
However, given these assets are decent value at ~7.0m, and score 110-150 individually, I don’t think it’s unrealistic to be able to anticipate these runs of form and turn a revolving cast of mid-range Talismen as they emerge into 160-200 points. This will surely intensify the monitoring of key individuals in the hugely busy 6.5-7.5 midfield bracket. Given the power of the advantage in terms of spotting the Talisman early, this should be a key area of focus for FPL managers to look for trends and spot value.
Of course, this is looking a small sample size of one season – we have to consider that there’s a chance that the big six may not be as far ahead as last time (though the opposite could be true and they could be even further ahead). All of this will impact on the chances of successfully pulling this off, but finding assets who will return 150 odd points and then anticipating when they might score the bulk of these is a strategy that could pay off in 2018/9.
Perhaps a key learning here is that this shows that, where some big hitters are a set and forget, there’s huge merit in playing fast and loose with mid-tier Talismen based on the fixtures. The data we’ve looked at for this article supports that theory.
Thanks again to Mitchell Stirling for all his help with this article.