FPL Poker Tables GW5

Thankfully, it didn’t take long for my spreadsheet’s predictions to deliver on last week’s expectation that they will become progressively more accurate with each passing gameweek. The omens are good that after struggling to navigate the choppy waters of the first 4 gameweeks, my spreadsheet’s course is being corrected, and the weather is set fair for smoother sailing from here on!

My spreadsheet belatedly registered it’s first correct score forecast (ARS 2 SHU 1). Incidentally, the bookmakers have only marked one match up correctly on that front too. It was fitting that the breakthrough should happen in that game as I chose that particular matchup last week to explain the rationale behind favouring the method I use for score predictions, rather than simply going by the scorelines assigned the highest probability (1 – 1 in that instance). Only 1 correct forecast from the opening 38 fixtures, however, is way down on last season’s average of 1.7 per gameweek, and highlights just how unpredictable results have been so far.

For all that my spreadsheet’s predictions are not yet performing to last season’s standards, they continue to fare better than the outcomes reckoned by bookmakers to be the most probable. In GW4, the bookies were closer to the actual number of goals scored for only 2 teams. Namely, MUN & WOL, who they had to score once rather than twice, but even there, I did flag up last week that WOL to win one nil was actually the highest probability my spreadsheet had assigned to any scoreline all season, and MUN were reduced to ten men after 28 minutes.

Unsurprisingly then, my spreadsheet convincingly beat the bookies when it came to the mean absolute error (MAE) of number of goals predicted, averaging 1.25 versus their 1.45 (the lower the better). Scorelines of MUN 1 TOT 6 and AVL 7 LIV 2 were the main reason for these MAE values being so high. By way of comparison, my spreadsheet last season was often able to record a MAE of around 0.50, which equates to being only half a goal out (on average) for each team’s score prediction.

With a new gameweek on the horizon, however, we should be looking forwards not backwards, so here then are the score predictions for GW5:

Disappointingly, from the FPL manager’s point of view that more goals equals more points, there are 3 fewer teams deemed more likely to score twice than once this week, and in descending order of expected goals, they occupy the top 5 places in the table below:

WBA, LIV, MUN and LEI more likely to score once

Ordinarily, I’d have been bracing myself for howls of protest about EVE being ranked top for goals with a Merseyside derby next on the agenda for them, but I reckon I can relax now after AVL knocked seven goals past the Champions!

Azpilicueta (5) and Saiss (9) both featured in the shortlist of 18 players expected to exceed the 5 FPL points average (excluding bonus) threshold last week, but so did Alexander-Arnold unfortunately. The first named is likely the only playing defender in the GW5 shortlist below (James bench and PVA injury), but it seems reasonable to assume Chilwell would be on this list too had he more gametime with CHE under his belt. I should point out that recent transfers and players from newly promoted teams are still not yet eligible for inclusion.

In a new development, I’ve shown what penalty takers’ expected average points would decrease to if penalty kicks were removed from the equation (see below). Encouragingly for Kane owners, this week’s table topper doesn’t owe his score to any.

Only Kane has appeared in all five of these tables so far, and it was my spreadsheet’s long range forecast that he would do that has made him an ever-present in my team so far. Son, Salah, and Antonio are the only players to have featured in all but one.

Kane maintaining a 100% record in making the cut

Note that Salah is ranked higher than all of the Toffees’ attacking triumverate, despite EVE being predicted to outscore LIV, and the reason why gives a good insight into how my players points predictions are calculated. Salah’s share of expected goal involvement in LIV’s last 8 away games was 38% of expected goals, and 17% of expected assists. This compares favourably to Calvert-Lewin (37% / 2%) and Richarlison (27% / 22%) in EVE’s last 8 home games, especially when the extra point per goal for Salah over DCL is factored in.

Sadly, my captain picks this season have followed on from last season in terms of being unmitigated disasters! So, I will be looking to Kane, Antonio or KDB to bring the current sequence of FOUR blanks to an end.

In recognition of the fact that I have perhaps become overreliant on my spreadsheets as the sole source for my captaincy decisions, I have returned to the methods that saw me win my main money mini-league three years out of four.

Looking at My Stats Tables in the Fantasy Football Scout Members Area, it bodes well for would-be Kane captainers to see TOT leading the way in terms of Minutes Per Big Chance (see below).

Sods Law would decree, however, that when I looked at my Team Defence table to see who are worst for Minutes Per Big Chance Conceded (see below), the hope that WHU are among them was completely confounded. In an echo of the ‘irresistible force paradox‘ it turns out that the unstoppable force that is Kane will be meeting the immovable object that the WHU defence have improbably become!

A case can still be made for Kane though if we consider that of the 4 teams WHU have faced so far, NEW, ARS, and WOL have hardly been setting the stats for attacking football on fire lately! In fact, those three teams are in the bottom half of the table for other metrics in My Stats Tables: Shots – Inside Box; and, Shots On Target.

Scout is the only fantasy football service I’ve ever paid for over the past seven seasons, and I’ve always believed their membership subscriptions offer great value for money. Anyone considering joining can find out more here.

Pleasingly, the 2 teams highlighted last week as having unusually high probabilities of a clean sheet (WOL 67% and CHE 62%) lived up to the hype. The latter are the only team this week reckoned more likely than not to keep a clean sheet in GW5. Whilst the 4 teams highlighted in green (MCI, WBA, MUN and SHU) are deemed less likely to concede one goal than none, they are still more likely to concede than not, because of the added possibility of conceding two, three, four, etc.

I made reference last week to my spreadsheet’s clean sheet probability calculations historically being its strongest suit, so it is pleasing to report that they were on average 4% more accurate in GW4 than the bookies probabilities posted by @FPL_Salah (see below).

And this wasn’t because the bookmakers had a particularly bad week by the way. In fact, their mean absolute error of 34.25 was better than the MAE of 37 they averaged in the 11 weeks I monitored between GW7 and GW17 last season. As promised last week, I did take a retrospective look back at the first 2 gameweeks of the current season, and the bookies came out on top in neither, which means they have now fared better than my spreadsheets in ONLY 3 of the last 16 gameweeks compared.

Another thing I tested last season was the accuracy of my spreadsheet’s longer-term forecasts, and the results were very encouraging indeed. I compared the projections for the following 5 gameweeks with current gameweek only predictions provided by “the world’s most powerful predictive fantasy football algorithm”. And, despite that model having the informational advantage of being up-to-date prior to each of the following 5 gameweeks, my spreadsheet outperformed it in each category I ran correlation tests on (number of goals teams scored, correct score forecasts, and mean absolute error). In other words, even from 5 gameweeks earlier, my spreadsheet’s projections proved more accurate than up to the minute predictions provided elsewhere.

The possibility of a double up on WOL defence in GW6 was touted last week, and they continue to look as though they have good prospects for 3 clean sheets out of 4 thereafter (see below).

WOL clean sheet potential looking golden in GW6-9

My 6GW spreadsheets are now available for purchase at a cheap as chips cost of £2 each, or the heavily discounted price of £30 for a season ticket. Please feel free to DM me any questions you may have about acquiring your own copies, or click here to go directly to my PayPal link.

May the GW5 flop be with you!

Coley a.k.a. FPL P0ker PlAyer @barCOLEYna

FPL Poker Tables GW4

Good news everyone. I’ve hit upon a genius idea to make my spreadsheets more accurate. Simply add 1 to every team’s predicted score, to take into account the effect of the new handball rule!

I’ve previously cited grounds for treating my spreadsheet’s predictions with caution at the start of the season (the impact of new transfers; the suspect form of several teams post-lockdown; and, the unknown quantity that are newly promoted teams), but it became clear last weekend that I’d missed the most distorting effect of all. The new handball rule is hands down (or up, away from the body, and outside the body line) the single biggest factor making results and scorelines so unpredictable this season.

After three gameweeks, my spreadsheets are yet a register a single correct score forecast! Last season, they managed an average of 2 per week, which may not sound like many, but is actually decent. That said, they had the BHA 1 MUN 2 scoreline right, up until the 4th minute of additional time, and the CRY 1 EVE 1 prediction would have been two out of two if you chalk off the ridiculous penalty awarded to the visitors because Joel Ward had forgot to superglue his arms to his sides!

Who saw MCI 2 LEI 5 and WHU 4 WOL 0 coming? Not my spreadsheet, that’s for sure! FUL 0 AVL 3 and BUR 0 SOU 1 were the other results to diverge from my model’s predictions markedly.

All I have to do to cheer myself up whenever I am feeling disappointed by my spreadsheet’s limitations, however, is compare its predictions with those deemed most probable by bookmakers. Half of the predictions tallied, but the bookies were out by more in 4 of the other 5, and this was enough to make the mean absolute error (MAE) and mean square error (MSE) of their predictions considerably worse than my spreadsheet (see below).

correlation test of predicted and actual goals

By way of comparison, my spreadsheets averaged a MAE of 0.8 last season, which is actually very respectable. Note, the lower the MAE the better. For ease of understanding, a MAE of 1.0 would equate to being one goal out (on average) for each team’s score prediction.

In other good news, the expectation is that my spreadsheet will become progressively more accurate with each passing gameweek, so let us turn now to the predictions for GW4. Here are the scorelines suggested by putting together each team’s most likely number of goals (see below). The probability of each scoreline is shown alongside.

17.8% for WOL 2 FUL 0 is the highest probability my spreadsheet has given to any result this season so far. That said, WOL 1 FUL 0 has an even higher probability of 20.2%! Remember, the method used to arrive at these correct score forecasts doesn’t necessarily equate to the scorelines with the highest probability. In fact, only 4 of these do so. The highest probability for the other scorelines are as follows: EVE 1 BHA 1; LEE 0 MCI 2; SOU 1 WBA 1; ARS 1 SHU 1; and, AVL 0 LIV 1.

So, you’re probably wondering why I don’t just list the scorelines with the highest probability? The short answer is because the method I prefer yields better results. Last week, for example, going with the highest probability scorelines would have fared worse than the bookies, with a MAE of 1.42.

The long answer would invite you to take the example of ARS vs SHU this weekend to understand why I prefer the method I use. Whilst it is true that 1 – 1 has a higher probability than 2 – 1, it is also true (paradoxically) that ARS have a higher probability of scoring twice than they do just the once, and I deem that to be the deciding factor.

Further good news for GW4 is that there are 8 teams deemed more likely to score twice than once and, in descending order of expected goals, they are shown in the table below:

good news for most of the big hitters

As highlighted by my Season Review blog earlier this month, Palace are nowhere near as defensively resolute away from home, and Werner sellers may come to regret their impatience this weekend. Aubameyang sellers last week may likewise be nervously eyeing his forthcoming fixture vs SHU, who will be without O’Connell. Meanwhile, Jimenez buyers disappointed by his blank vs WHU last weekend can be hopeful of belated returns vs FUL, seemingly the league’s whipping boys. There is plenty of encouragement here too for owners of the usual suspects from SOU, EVE, MCI, MUN and LIV.

In further welcome news for Jimenez owners, WOL also feature in the top 5 projected highest scoring teams in 2 of the following 3 gameweeks (see below).

WOL sit in the top 5 for expected goals in 3 of the next 4 gameweeks

Courtesy of my spreadsheet’s player points prediction model, Rashford is my sole MUN attacker, and I was pleasantly surprised just now to discover that he’s the least owned of the Reds’ front four. Especially so, given that he’s at the summit of my spreadsheet’s GW4 expected FPL points table (see below). Some of you will remember that Fernandes was ranked highest of the quartet last week, and justifiably so as it happened.

Intriguingly, 2 defenders from CHE feature in the top 4 here, which is territory normally reserved for the marauding fullbacks of LIV.

4 defenders made the cut

Scepticism was expressed last week about Sterling not featuring in the GW3 table, and I anticipate similar protests this week about the notable absences here of Fernandes, DCL, Jimenez, Aubameyang and KDB, so for context they are provided in the best of the rest table shown below.

a pack of Wolves

Remember though, these are averages. In reality, players points tend to polarise between high and low. So, in the absence of clean sheet points, a midfielder scoring a goal and earning a bonus point one week (8), but blanking the next (2), would have averaged 5 points.

At the other end of the pitch, meanwhile, there are much stronger candidates for clean sheets this week than there were last, although the unusually high probabilities assigned to WOL and CHE (see below) fly in the face of the fact that they conceded 7 goals between them in GW3! The hope is that the 4 goals shipped vs WHU last week was just a bad day at the office for the former, whilst the chalk and cheese performances of the latter’s defence playing at home rather than away was something else picked up on by my Season Review blog.

Historically, my spreadsheet’s clean sheet probability calculations have been its strongest suit, so it was disappointing that none of the 4 teams highlighted in green last week (MUN, LIV, MCI and CRY) managed to keep a clean sheet.

I kept a record last season, up until lockdown, of how my model’s clean sheet probabilities fared against those implied by bookies odds, and made popular by @FPL_Salah, and on only 2 out of 12 occasions did the bookies do better than my spreadsheet (1 tie). Last week, the bookies were on average 0.52% more accurate though, so that makes it 3 out of 13 for them now. Time hasn’t permitted me to look back at how they fared over the first 2 gameweeks, so I will report back on those after the International Break.

Just when confidence in the WOL backline has taken a knock, they actually top my clean sheet probability projections for the next 6 gameweeks, so it will be interesting to see how they fare this weekend before possibly committing to a double up in GW6.

Interestingly, one of the promoted teams are currently riding high in this same table in 4th place, which has me sitting up and taking notice, and bodes well for me making savings in my team’s defence.

Another thing I tested last season was the accuracy of my spreadsheet’s longer-term forecasts, and the results were very encouraging indeed. I compared the projections for the following 5 gameweeks with current gameweek only predictions provided by “the world’s most powerful predictive fantasy football algorithm”. And, despite that model having the informational advantage of being up-to-date prior to each of the following 5 gameweeks, my spreadsheet outperformed it in each category I ran correlation tests on (number of goals teams scored, correct score forecasts, and mean absolute error). In other words, even from 5 gameweeks away, my spreadsheet’s projections proved more accurate than up to the minute predictions provided elsewhere.

I raise this here because it’s high time I let it be more widely known that my 6GW spreadsheets have been on sale for some time now, since Project ReStart in fact. They are at a cheap as chips cost of £2 each, or the heavily discounted price of £30 for a season ticket.

I have hitherto been unsure about broadcasting this development, and consequently only sold sheets to those who took the initiative to enquire about buying them. In light of the increasing number of take-ups I’ve had lately, however, and the positive feedback received, there seems no good reason for me to continue keeping to the shadows, so please feel free to DM me any questions you may have about purchasing your own copies, or click here to go directly to my PayPal link.

May the GW4 flop be with you!

Coley a.k.a. FPL P0ker PlAyer @barCOLEYna

P.s. Since my last blog I’ve had the honour of being signed up to the Scout Network. Scout is the only fantasy football service I’ve ever paid for over the past seven seasons, and I’ve always believed their membership subscriptions offer great value for money. Especially as their Members Area played a key role in me winning my main money mini-league three years out of four! I tend to assume that everyone who takes FPL seriously is already a member these days, but if you’re not and you’re considering joining, then you can find out more here.

FPL Poker Tables GW3

Honestly, as the goals flew in at the weekend at an unprecedented rate, I felt like declaring the weekly exercise of formulating predictions suspended until further notice. Until GW17 perhaps when I’d have sufficient new season data to trust in them more. The ridiculously high conversion rates achieved by goalscorers this season so far is making a mockery of xG based models, and I was all for binning mine off in a fit of pique!

Then today, I reminded myself of 2 things: a) regression to the mean is unlikely to be a redundant concept; and, b) my spreadsheet’s score predictions weren’t as bad as those given the highest probability of occurring by the bookmakers, who remain the undisputed benchmark for such things. My sheet may have underestimated the volume of goals that would be scored, but by nowhere near as much as the bookies did (see below).

Those who read my blog last week will recall my making a strong case for why my spreadsheet’s predicted CHE vs LIV scoreline should be changed from 2 – 0 to 1 -1, and the match was still poised at 0-0 at the halfway point before Christensen‘s sending off changed the course of that game.

The most pleasing results for me were in the ARS vs WHU and LEI vs BUR games, both predicted by my spreadsheet to finish 2 – 2.  I am sure most people expected an easy win for The Gunners (hence Aubameyang being the most captained player), and a tight, low scoring affair in the other game. Anyone who saw the first of these two games will know that WHU were unlucky to lose, and to have not scored at least 2 goals themselves. Whilst the high scoring (4 – 2) in the other game ensured my bet on Over 3.5 Goals at odds just shy of 3/1 was a winning one.

Looking at the tables above now, the only result the bookies were nearer to was the last one (WOL vs MCI), but even there, I did point out last week that my spreadsheet actually gave a higher probability to a one nil win for City.

So, for now at least, I will keep faith in the law of averages making a belated appearance sometime soon, and trust in my spreadsheet’s process, even though strictly speaking I can’t fully until around GW17. Turning our attention to GW3 then, here are the number of goals likeliest to be scored by each team.

The first thing to notice with these scores is the absence of any clean sheets, which if GW1 was anything to go by, will no doubt mean a glut of them. As with last week, we have an unusually high number of scorelines here (7), that are deemed paradoxically to have a higher probability than the ones shown above. Namely, BHA 0 MUN 1; CRY 1 EVE 0; WBA 1 CHE 1; BUR 2 SOU 1; TOT 1 NEW 1; MCI 2 LEI 0; and, LIV 1 ARS 0.

There’s a very surprising name at the top of this week’s predicted expected goals table. One of the things my Season Review blog highlighted was the strong attacking numbers BUR posted towards the end of last season, and with their defence in bad shape through key injuries to Mee (and Tarkowski?) at the moment, there’s cause for optimism that they’ll need goals in their upcoming fixture vs SOU. I am personally toying with the idea of bringing in Wood for a one week punt, prior to switching him to Calvert-Lewin.

Having said, there are no clear cut clean sheets chances this week, there are four teams (MUN, LIV, MCI and CRY) for whom conceding 0 goals has a higher probability than conceding 1. These are highlighted in green in the table below.

Now, I appreciate most of you are stanning for EVE assets right now, so please don’t @ me. There’s no denying Palace have been impressive so far, and they did totally nullify the MUN attack you have also been drooling over the prospect of doubling up on beforehand. That said, half of the data used to calculate EVE’s expected goals comes from their very poor post-lockdown form, and before the arrival of Rodriguez, Doucoure and Allan, who have had an immediate positive impact, so I wouldn’t judge you for not heeding my spreadsheet’s warning here.

And finally, the only players deemed likely to average at least 5 FPL points in GW3, not including bonus points, based on their share of their team’s respective xG+xA in the last 8 home or away games, whichever is relevant, are shown in the table below.

Please feel free to ask any questions about the data presented in any of the tables above.

May your arrows be green!

Coley (a.k.a. FPL P0ker PlAyer @barCOLEYna)

FPL Poker Tables GW2

Nothing makes bigger fools out of us than football results can! So much for the shortage of goals and scarcity of clean sheets forecast last week by my GW1 tables. In fact, there were more clean sheets than there normally are when we have a full complement of fixtures! And having predicted a near clean sweep of one-one draws, there were zero! This might sound like a remarkable feat of incompetence, but the probability of such an outcome implied by best bookies odds was 39%. And not because of how outlandish my spreadsheet’s predictions were, by the way. In fact, they corresponded remarkably closely with the probabilities assigned by the bookmakers.

Only one scoreline was deemed less likely by the bookmakers (LIV vs LEE)

In mitigation, the biggest divergences away from my spreadsheet predictions involved the 3 games featuring promoted clubs and, as I explained last week, trying to peg the respective attack and defence strength ratios of promoted teams before the season starts is a guessing game. I also highlighted the ‘known unknown‘ effect of new transfers and, prior to the disappointing performance of Havertz, it seemed as though every single one hit the ground running, and had maximum impact! Think Gabriel, Willian, Wilson, Hendrick, Lewis, Castagne, Allan, Doucoure, Rodriguez, etc.

Understandably, because of the kind of ‘informational disadvantage‘ touched on above, my spreadsheets are likely to be less accurate at the start of a new season. The trouble is that because 8 gameweeks of data for both home and away results represents the optimal data range, we would ideally wait until GW17 before using my spreadsheet’s predictions, but that’s halfway through the season! Instead, we must cling to the hope that the last 8 home and away results from the previous season have some meaningful bearing on the start of the next one. It’s fair to say though that any confidence in this proving so was dented by the opening round of results.

Anyhow, we move; onwards and upwards (not difficult!) to GW2. First off, here are the score predictions, with an added caveat that 7 of these 10 are not actually the scorelines with the highest probability!

Remember the paradox I highlighted last week with regards to WHU vs NEW? And how I explained why 2 – 1 was given as the likeliest score when in fact 1 – 1 had a higher probability? Well, there are not one, but SEVEN such instances this week: EVE 1 – 1 WBA; LEE 1 – 0 FUL; MUN 2 – 0 CRY; ARS 1 – 2 WHU; LEI 1 – 1 BUR; AVL 1 – 0 SHU; and, WOL 0 – 1 MCI. My expectation is that you will find all of these alternative scorelines more plausible, with the exception of the ARS result!

As alluded to last week, there are far stronger candidates for clean sheets this week than last, which pretty much guarantees a complete absence of them if last week is anything to go by! The one that will undoubtedly have people scoffing here is the 61% probability assigned to CHE vs a LIV team that just put 4 past the best defence in the Championship last season.

there are strong grounds for downgrading the P% of a CHE clean sheet to 40%

Please remember though that this is based on The Blues’ last 8 home games and The Reds’ last 8 away and, as discussed in my 2019-20 Season Review, the champions’ form did peter out in the run-in, whereas the fourth place finishers defensive record at home was actually very decent from an xGC point of view. Indeed, their best sequence of 8 home games was better than that of any other team. Shame about the Kepa.

In recognition of the fact that LIV had nothing to play for towards the end of last season, I checked how switching the data from last 8 games only to those for the season as a whole might effect the prediction, and it changed things considerably, reducing the clean sheet probability for the home team to 40%, and making 1 – 1 the likeliest scoreline.

The plea bargaining above should be taken into consideration, therefore, when seeing CHE topping the projected expected goals table below.

CHE would drop to below EVE on 1.6 if using season data rather than last 8

MUN would become the table toppers this week if LIV are given the dispensation discussed above, which is good news for all those planning to captain one of Fernandes, Martial or Rashford, with the last named preferred by my spreadsheet’s expected player points table (see below).

Note that this is not an argument for making Rashford your sole attacking asset from MUN, because next week Bruno will be the MUN player to own. The disparity arises because my spreadsheets distinguish between home and away form. As FPL managers, likely to be owning MUN assets for longer than a one week punt, we are generally interested in overall form, so for the record it is close between these players, and dominance will likely come down to who takes the most pens. Thanks to @AlbertLucue9 for asking the question that prompted this paragraph.

WHU were the big underachievers in last week’s predictions, so scepticism about their prominence here, and that of Antonio below, would be understandable. The other team to leap off the page here is BUR, who as discussed in my Season Review blog finished the season with their underlying attacking stats on a noticeable upwards trajectory. Whether or not they return to type at the start of a new season, with the emphasis back on defensive solidity, remains to be seen.

excludes players from promoted teams

I am in no hurry to repeat the mistake of captaining Antonio again, so I will be entrusting my armband this week to the child poverty champion in the hope that he can feed my team’s hunger for catch up points!

May the flop be with you!

Coley a.k.a. FPL P0ker PlAyer (@barCOLEYna)

FPL Poker Tables GW1

The start of a new season was complicated enough with both Manchester clubs blanking, and the uncertainty around players in our squads going in and out of quarantine against the backdrop of a global pandemic, but the fixture schedule has thrown us an additional curve ball if my spreadsheet predictions are anything to go by. 

Some gameweeks see fewer goals, but the impact on our scores is usually offset (partially at least) by more clean sheet points being available.  However, what my spreadsheet predicts for GW1, is the perfect storm of a shortage of goals and a scarcity of clean sheets.

correct score forecasts for GW1 wth corresponding probability

One-one score draws are deemed the most probable scoreline in all but 1 of the matches, with only CHE and WHU reckoned to be more likely to score more than once (see below). Surely that’s a typo? I must mean abb but 2, right? Wrong. The Hammers fall into a strange paradoxical position I’ve commented on previously. Namely, that whilst it’s true that WHU are more likely to score at least 2 goals than 1, they are less likely to score twice than once. Fried your brain yet? Then might I suggest some lighter reading on quantum entanglement!

predicted xG scores for GW1

In a similar vein, 5 teams are forecast to have a higher than 40% chance of a clean sheet in GW2, whereas only 4 teams are rated to have a better than a one in three chance (33%) in GW1.  Of those 4, only LIV cross the 36.7% threshold, which is significant because that is the point at which conceding zero goals becomes mathematically more likely than conceding one goal.  Remember though, the likelihood of conceding two, three, four goals, etc., also needs to be factored in, so a clean sheet only becomes the likeliest outcome of all when the assigned probability is higher than 50%.

calculated clean sheet probabilities for GW1

There are always caveats attached to my spreadsheets, and rightly so, for algorithms always have shortcomings, but there are additional ones to consider at the start of this season. Obviously, no algorithm can second guess the impact that COVID19 will have on team lineups and results. Likewise the impact of empty stadiums on players and teams.

The other ‘known unknown‘ at the start of any new season is the transformative effect new signings can have on teams’ performances, and one only need reflect back on the arrival of Fernandes in Manchester in January to know how big this can be.

More pertinent to my spreadsheet, however, is the fact that these predictions are based on each teams xG performances in their last 8 home or away games (whichever is relevant), 4 or 5 of which will have come post-lockdown. And, as I showed in my recent review of last season, some teams prospered during this period, while others faltered dramatically.

LIV belong in the latter category, and as they face a promoted team whose defensive credentials in the top flight are yet to be tested (and so can only be best guessed at), it is possible that the probability of a clean sheet for the defending champions is being underestimated by my sheet.

The attacking and defending strength ratios attributed to the promoted teams, based solely on matches against the other six sides in the Championship’s top seven last season, have been scaled back by 20% to reflect the tougher competition they will face in the Premier League.

Unavoidably, this is somewhat arbitrary, but not random (never random!). Rather it is based on some research I’ve read, but I accept that it’s an inexact science. The point being that if you believe the transition from EFL to PL is more than 20% harder, then the probability of LIV keeping a clean sheet increases accordingly. For what it’s worth, the probability of LIV keeping a clean sheet only increased by 3% to 42% when I substituted the figures for the season as a whole, rather than taking their last 8 games only.

That all said, it’s quite conceivable that LEE will be like rabbits caught in the headlights of an oncoming juggernaut, and an early goal will open the floodgates, and demoralise them. Personally though, I’m not banking on it.

Bring on the new season, bring on the green arrows, but most of all, bring on the coronavirus vaccine!

Coley a.k.a. FPL P0ker PlAyer

2019-20 Season Review of my spreadsheet’s rolling 8GW data

During the course of last season, I established that an 8 gameweek data was optimal for my spreadsheets.  From GW17 onwards, my spreadsheet’s better than bookies odds predictions were based on each team’s last 8 home or away games, whichever was relevant for the upcoming gameweek.

Recently, I found myself wondering to what extent teams’ attack and defence strength ratios fluctuated throughout the season.  And I wondered if looking more closely at how these values changed would reveal which teams showed the most improvement post lockdown.

So I coloured each team’s best to worst 8 game sequences on a scale of blue to red, and was pleased to discover it was easy to see where in the season each team’s best and worst xG form was from the start of the season to the finish (left to right).  In the end column I highlighted in blue the teams that performed to their best levels in terms of xG during the run in.  The table below is how it looked for xG expected to be scored at home vs an average defence.Rolling xG HDisregarding relegated BOU, this table suggests ARS BUR CHE LEI and SOU finished the season strongly in terms of attacking threat at home, but it is important to understand that the blue to red scale was applied to each team separately.  The table below is the same as above, but with the same scale applied to all teams relative to each other.  Now we see that we should not perhaps be getting too carried away with the attacking strength of ARS and SOU at home just yet.Relative xG H


xG H rankingsNext I ranked each team according to their highest, lowest and average (mean) ratios. The table to the right shows these rankings for xG expected at home.

As can be seen, CHE were a force to be reckoned with at home. Their best 8 gameweek attacking spell at Stamford Bridge was bettered only by MCI, whilst on the defensive side of things (see table below), the xG they conceded in their closing 8 home games ranked as the best 8GW sequence recorded by any team last season. Away from Stamford Bridge they ranked only 5th and 10th best respectively.

WOL ranked 4th best in terms of their xG ceiling at home, but drop to 8th place with regards to their xG floor. This does not account for why they finished behind TOT in the league, however, as they bettered Spurs in every category (highest/best, mean, lowest/worst) for xG scored and conceded both Home and Away.xGC H rankings

CRY were a top 8 side for keeping things tight at home (see left), but this might be a euphemistic way of describing their lack of attacking threat at Selhurst Park, which ranked 2nd worst across all 3 categories (see above).

Surprisingly, BUR‘s best attacking spell at home, during the second half of the season, ranked 6th highest in the division (see above), but appears to have come at the expense of the defensive side of their game. Certainly, they consistently conceded less xG in the first half of the season (see below).


Rolling xGC H - BURLEI‘s game appears better suited to playing away from home. The Foxes’ best 8 gameweek attacking spell on the road ranked 2nd best in the league, and the xG they conceded during their worst sequence of away games was comparably better than any other team managed.  On average, they had the 3rd best away attack and 4th best away defence (see below), whereas at home these only ranked 13th and 9th best.

xG & xGC A rankings

The overlapping between different teams’ highest, average and lowest ratios can be better visualised in the following charts.  First off, here’s the 8GW sequences of xG at home.  The number on the Y axis represents the number of xG that teams would be expected to score vs an average defence.Home ATTACK

And here’s the 8GW sequences of xG expected to be conceded at home vs an average attack.Home DEFENCE

Another piece of information conducting this exercise provided me with was the difference between the average number of xG teams would be expected to score and concede vs average opposition.

Strangely, only 7 teams had a positive average xG difference at Home. In descending order these were MCI LIV CHE MUN WOL SHU and EVE.  Counterintuitively, 9 teams had a positive average xG difference playing Away.  These were MCI LIV LEI WOL EVE MUN SOU CHE and ARS.

Of the teams above, only LIV and SHU were not amongst the teams highlighted as having strong finishes to the season in terms of attack and/or defence at home and/or away.  The inclusion of EVE in the ranks of strongest finishers (Away defence) is the biggest surprise here, but can probably be explained by the fact that half of their last 8 away games came before lockdown.  If EVE can recover their pre-lockdown form, however, their defenders might represent good value if FPL Towers base their price points on the poor form shown after the Restart.

Hopefully, the insights gained from this exercise, combined with an accurate assessment of the difficulty of fixtures faced by teams at the start of the 2020-21 season, can give us a flying start.

Good luck, everyone!

Coley aka FPL Poker Player @barCOLEYna



Using baseline BPS to identify backline bonus magnets

I had another FPL itch recently that I needed to scratch.  I had an idea for a more nuanced approach to using bonus points system scores to better identify from which teams the goalkeeper was the best pick for our FPL squad defences.

There will be few FPL managers who’d have argued against Pope being the best BUR option last season, for example, but what about other teams?  Are goalkeeper always the pick of the crop when it comes to clean sheet bonus points?

It won’t surprise you to know the answer to that last question is an emphatic No!  In fact, the results from my subsequent research showed wide variance in BPS points earned by goalkeepers last season when clean sheets were kept.

My idea was to compare the BPS points that defenders and goalkeepers earned whenever their team kept a clean sheet.  To my way of thinking, it is more meaningful to analyse how these players fared only when successfully doing so, because for them that’s generally when bonus points are up for grabs.  These tight-at-the-back games are also more likely to be a better predictor of future bonus points allocation than when we also include the noise from matches when teams are chasing the game.

As well as filtering for the games players kept clean sheets, I also removed the effects of goals, assists and yellow cards from their BPS scores to establish baseline averages.  There are obvious exceptions, but generally speaking it remains the case that goals and assists are rare events for most defenders, and clean sheets are much more predictable than attacking points are.  Again, some defenders are known card magnets, so this will still need to be factored in when using the results of this exercise to determine future transfer targets.

So, without further ado, here are the table toppers who banked a minimum of 5 clean sheets last season.  Firstly, the 20 players who average a baseline BPS of 27+ per clean sheet:


Top 20 baseline BPS

As you can see, only 4 goalkeepers (highlighted in lime green) feature in this top twenty, which is consistent with most defences’ ratio of defenders to keepers.  These were Lloris, Pope, Guaita and Leno.

The players with under 8 clean sheets are also highlighted (light red fill dark red text).  Small sample size notwithstanding, Cancelo already looked the best route into the MCI backline, even before 2 recent UCL ties suggested he is now Pep’s preferred choice for the left back position.

Gomez being best for baseline bps for the league champions came as a surprise to be honest, but I don’t see his slight edge over Alexander-Arnold here being enough to change many FPL managers plans!

High BPS scores are all well and good, but they do not in and of themselves count for anything unless accompanied by Bonus Points, so the next table shows the top 20 players for average bonus points collected per clean sheet:

Top 20 average bonus

This time there are 6 goalkeepers in the top twenty, with Henderson and Ryan both entering the frame.

I read somewhere recently that in terms of of BPS per 100 mins, Ryan was the best option from the BHA backline, but my method shows that Dunk is actually the standout pick.  He averaged over one and a half bonus points on the 9 occsions he was involved in a clean sheet.  Intriguingly, there has been speculation linking CHE with Dunk, a boyhood fan, and he could be a real bargain for FPL managers if BHA agree to sell.

Luiz being next best of the defenders will no doubt come as a shock to those who witnessed many of his lamentable performances for ARS last season, but we should be wary of our cognitive biases blinding us to his potential bonus point magnetism, which averaged 1.5 on 8 occasions last season.

Interestingly, EVE have more representation here than any other team with 4 defenders making a fifth of the top twenty.  This might merely be a reflection of how little cutting edge they have at the other end of the pitch.  Sidibe‘s prominence here probably owes much to him being fielded out of position as a right midfielder on a number of occasions, so Keane might have been the best of the bunch even if the former’s loan contract at EVE had not expired.

I imagine most managers would have expected Dubravka to be the clear choice when it came to NEW assets, but my method points to Fernandez being a better option.  Certainly as far as collecting bonus points is concerned.

Chilwell and Digne both average more than a bonus point per clean sheet, and for half a million less, the former could depose Azpilicueta as the primary beneficiary of bonus points when CHE do keep a clean sheet if sealing a move to Stamford Bridge.  Unless Dunk transfers there too of course!

Mina and Van Aanholt are the only other defenders in the 1BP+ members club but there are better (and cheaper) bonus point magnets in their respective teams.

Another thing to notice in the second table above is how there are no represantatives from the teams who finished in the top 3 places.  This highlights one aspect of owning defenders from top teams, which is that they will struggle for bonus points in the absence of attacking contributions, because these teams score more goals, and BPS generally favours goal scorers and assisters.

Remember though, the purpose of this exercise was to identify the teams where the goalkeepers were the best option.  Seemingly, the only teams where this was the case, from the point of view of bonus points at least, were TOT, BUR, CRY, SHU and SOU, in the form of Lloris, Pope, Guaita, Henderson and McCarthy, respectively.

There has been much debate about whether Pope still represents good value at 5.5m, but only Lloris, with a much smaller sample size (6 clean sheets vs 15), is ranked higher than him in the tables above.  This remains the case (see below) even if we factor starting price into the equation by using an average Bonus Points per £m metric instead:GK average bonus per £mPope does drop down to 7th though, with Ryan and McCarthy 1st and 2nd, when ranking keepers by baseline BPS per £m, but as stated previously, BPS are meaningless if not accompanied by bonus points.

The only defender who can better the 6bps per £m of Matt Ryan is another Ryan.  Namely, Ryan Fredericks of WHU (see beloe) who manages 6.1, so he is another name to keep in mind if fixtures fall kindly for the Hammers.

Top 10 DEF baseline BPS

Where it matters most though, Dunk is once again the star defender when it comes to average bonus points per clean sheet per £m starting price:

Top 10 DEF average bonus per £m

Keane, Fernandez and Mings are the other names to join Dunk and Fredericks in both these top tens, and might be worth keeping in mind when the fixtures are released.

Ultimately though, these metrics I’ve used are only useful if the players under the spotlight here are again involved in clean sheets this season.  Furthermore, there are other considerations than just the ones focused on here when it comes to selecting our initial squad, such as team structure and balance.

I am glad I scratched my FPL itch though, because I feel like I have greater clarity now about which goalkeepers and defenders to include in transfer considerations throughout the first half of the upcoming season.

Hopefully, you’ve found the contents of this article interesting food for thought too, and whether it informs your squad selection or not, I wish you all the very best for the season ahead.


Coley (aka FPL P0ker PlAyer)







What FPL and Poker have in common


It always makes me smile when FPL managers express the hope that others will have a good gameweek.  It reminds me of the way poker players wish each other good luck” when strictly speaking that’s the last thing they should want.  In reality, they usually don’t care much whether or not your luck is good, just as long as it doesn’t affect how much they win.  Similarly, when FPL managers say “may your arrows be green”, they usually don’t mind if they are, just so long as theirs are greener!

But that’s okay because the world is a better place for such niceties.  Less friendly are the passive aggressive congratulations given to players who have clearly ‘lucked out’.  So in poker, aggrieved players often say nice hand or well played to their arch-nemeses with barely suppressed sarcasm.  The same dynamic is also seen sometimes in replies to FPL tweets when screenshots of lucky autosub outcomes are posted.

One of the most obvious similarities between the two worlds is with the abundance of ‘bad beat’ stories.  In poker, these are told by victims of perceived misfortune.  In FPL the same kind of hardluck stories are seen all the time regarding either the big points scored by recently transferred out players, or the big points missed on a coin-flip decision.  Deep down we all know it’s wrong to inflict such self-pity onto our peers, but it’s a compulsion we find tough to resist.

The bête-noire of poker etiquette, however, is the ‘slowroll’, where someone leads you to believe that you have won a hand, but then turns over a very strong hand.  The FPL counterparts of these offenders are the grandstanders who, having amassed an impressive gameweek score, tweet to ask everyone how many points they’ve scored before then revealing their own better one.


MatrixThe correct answer to most questions posed by poker novices is “It depends.”  The same is true in FPL.  Most “Should I transfer in Joe Bloggs for John Doe?” type questions would benefit from a similar precautionary principle being applied.

For example:

Do you have more pressing problems elsewhere?

Where are you in your Mini-League(s)?

How does your squad compare to your ML rivals?

How will future transfer plans be affected?

These are all valid questions to ask before giving yes or no answers.  In poker, this is like being asked how to play a specific hand in early position in a tournament, where the comparable “It depends” type answer might include the following preliminary questions:

Are you nearing ‘the bubble‘, i.e., the prizemoney?

How many chips do you have in relation to the other players at your table?

Will you have enough chips left to make a move later if losing the hand, e.g., stealing the blinds from passive players to your left?


One hallmark of a good poker player is discipline when it comes to bankroll management, and most good FPL managers exercise budgetary discipline also.  A bad one makes kneejerk transfers to bring in premium priced players with little regard for no longer being able to afford to replace or upgrade any of their other out-of-form squad members.

Clearly, there are aspects of both games that are beyond our control.  In poker it’s the cards we are dealt, and in FPL it’s events like injuries and suspensions.  All we can do is try to play optimally in the areas within our control.

One such area is the weekly free transfer.  A FPL ‘shark‘ knows these should always be used to increase ‘expected value‘, primarily in terms of points, but also with regard to team value.  The ‘fish‘, however, is more likely to make kneejerk moves, that leave them wishing they’d hung fire, and/or maverick moves that usually share the same fate as those made by poker players guilty of ‘fancy play syndrome‘.


calculated riskThere is a tendency in the FPL community to describe any transfer not conforming to template as maverick, but there is a distinction to be made between maverick moves and calculated risks.


Few things in poker are more maddening than being ‘rivered‘ when ‘all in‘ as a massive favourite to win a hand, only for the final face-up card dealt to be the one and only card that gives your opponent a better hand.

poker skill and luck

The same frustration is felt in fantasy football when head-to-head or mini-league rivals hit the FPL equivolent of such ‘one outers‘ by having an autosub goalkeeper (Gomes in 2015-16) make two penalty saves in the same gameweek say, or an autosub outfield player score a brace on a Monday night.

Bad players of poker and FPL will frequently ‘suck out‘ on us, and be rewarded for poor game management, but the good news is that being rewarded for playing badly only encourages them to go on playing badly!

All poker players and FPL managers will experience ‘downswings‘.  What seperates the wheat from the chaff is how well they cope mentally with an inevitable aspect of the game.  In poker, most players are prone to going ‘on tilt‘ when they experience bad luck.  Such players are as a result more likely to lose money because they’ll be making decisions based on anger and frustration.  In FPL the same can be true if playing in cash prize mini-leagues.


Last season, for example, I was so filled with self-loathing about my failure to confirm activation of my Bench Boost chip in GW34, thereby costing me 43 points, that I self-destructively wasted my Triple Captain chip the next gameweek by gambling on serial cameo appearer Coutinho (1 point!).

Admittedly, I was comfortably winning my main money mini-league at the time, but my fog of despair was so dense that a few weeks passed by before I realised I’d sabotaged any prospect of cashing out in any of my other money leagues.  I’d been knocking on the door of the top one thousand club for several weeks, and those 43 points would have seen me well and truly kick the door down!


Moneymaker The way FPL has evolved over the past few years reminds me of the sea change that happened in the poker world after 2003.  There was a sudden growth in interest in poker after the $2.5 million first prize World Series of Poker Main Event was won that year by Chris Moneymaker, a 27-year-old accountant and amateur poker player from Tennessee, who had won his seat into the event through a $86 satellite tournament in a PokerStars online poker card room.  The ‘Moneymaker Effect‘ would give rise to a new breed of internet poker players who honed their skills online before facing off against old-school competitors.

Doyle v Durrr

The poker player stereotypes that had been around for an age were transformed over the course of a decade. The old-school competitors who believed in poker being an art, not a science, based on feel, instincts and reads, were quickly overrun. A new breed of math nerd, wonks using a mountain of sortable data from the millions of hands played online began to dominate the game. Math whizzes changed the game using probability theory to their advantage.

Similarly, I think the generation of fantasy football managers, who believed that football knowledge and a ‘good eye’ were the only requirements for success, are in the process of being overthrown by algorithm builders, probability theorists and statisticians.

Vanessa SelbstConsider yourself warned if you’re one of those who underestimates the relevance of maths in FPL. That geeky person who invited you to join his or her money mini-league is probably hustling you!

BlessedI think there might have been a typo in Matthew 5:5; perhaps it should have read:


Blessed are the Math gEEKs!

Good luck at the tables y’all!


FPL Poker Player @barCOLEYna

If Carlsberg did FPL spreadsheets….

They would probably be like mine.

Some spreadsheets simply track each team’s fixtures in gameweek order.  Some incorporate fixture difficulty ratings, with the aim of identifying favourable/unfavourable sequences to assist with managers’ transfer decisions. Although, as Richard Kenny @InfernoSix showed, FDR is a very blunt instrument when it comes to assessing fixture difficulty.  And some spreadsheets even assign predicted FPL scores to players based on a wide variety of algorithms, including RMT.  Mine do many of these things and more, but probably better.

Cheat Sheet

How I Got Here

I first came across the concept of season tickers during my first proper FPL preseason.  I immediately liked the concept, but as my second season progressed I grew dissatisfied  with what I perceived to be the arbitrary nature of team rankings, and distrustful of how often they’d be updated to reflect new realities.  Having finished 17,394th in my second season, I believed there was plenty of scope for improvement, and so I set out to learn what I could do with my desktop excel program, which I’d never had any use for previously.

I went into my third season determined to fly solo and rely solely on my own formulations.  With no formal background in maths whatsoever, I intuited a method of calculating the offensive and defensive strengths of each team relative to all the others.  As my Excel know-how grew I had several Eureka! moments that led to my second blog (December 2015) proclaiming my first prototype of a season ticker spreadsheet based on what I would later come to understand as a crude form of xG, though I’d never heard of that concept at that point.   My spreadsheets were constantly being refined throughout that season, but I nonetheless achieved a new personal best finish of 12,599th.

Unfortunately for my average overall rank though, I cut corners the following season by substituting my homegrown variety of xG with the much less labour intensive, and more readily available, Shots On Target data.  This was considered by many FPL managers at that time to be the most important metric of all for predicting attacking returns.  See One Stat To Rule Them All for example.  My complacency and laziness proved disastrous, and I did well to finish inside the top one hundred and fifty thousand (148,327) having been still outside the top two million going into Gameweek 12.

When I first met David Wardale @DavvaWavva during the summer of 2017 to be interviewed for his excellent book about fantasy football – Wasting Your Wildcard – I was feeling very bullish about my fifth season prospects.  This wasn’t the same blind optimism most players of FPL experience before the first bunting of red flags or first quiver of red arrows.  No, my optimism was based entirely on my discovery of a reliable source of xG data.  I’d done tonnes of research into the many different models used to calculate xG, and had settled on a source I trusted to take my spreadsheets to the next level.

In the course of that research into xG models, I’d been very pleased to learn that my homegrown method was almost identical to that now used by many professional sports bettors.  This was also when I learned of a concept called poisson distribution, which came to revolutionise my understanding of most likely correct scores based on my spreadsheet calculations, but more on that fishy sounding concept later.

Where I Am

When deciding where to ‘invest’ their money professional sports bettors use websites like FiveThirtyEight and ClubElo.  These were reckoned to be amongst the very best last season by alex b @fussbALEXperte who, apart from writing excellent articles about football and psychology, also measures the quality of football predictions.  After taking a close look at the methods and principles used by FiveThirtyEight and ClubElo I was pleased to see that they are essentially the same as mine.  Coincidentally, the latter was also credited by the aforementioned Richard Kenny as offering a more reliable barometer of teams’ attacking and defending strengths than FDR.

As far as I am aware though, these sites focus on upcoming matches only, which makes perfect sense given that most of the bettors they cater for have little interest in betting on the result of league matches several weeks away.  After all, there are a multitude of variables that can change teams’ future prospects in the meantime, e.g., injuries, morale, suspensions, transfers, etc.

Catering for FPL managers is a different ball game, however, as they must plan ahead to be successful.  They only have two wildcards per season and one free transfer per week, so my spreadsheets are just as focused on long-term projections about teams’ forthcoming fixture runs as they are for more immediate short-term predictions.

What I Do

After each gameweek I carefully enter the expected goals scored and conceded values reported by my preferred source, for all of the matches in the latest round of fixtures, into pre-prepared cells on my spreadsheets.  Consistently applied adjustments to these values are made in certain circumstances, e.g., red cards, unfairly disallowed goals, missed penalty kicks resulting in follow up shots, etc.

These are then systematically weighted to give every team individual values for strength in attack and defence, for both home and away.

The rationale for distinguishing between home and away form is simply that many teams approach away games differently to when playing at home, often changing their formations and personnel in the process.

The sophisticated part comes next, when my algorithms model what my subsequent spreadsheets will look like if, and admittedly it is a big if, my spreadsheets’ current predictions correlate precisely with actual outcomes.  Clearly, this is never going to happen, but I do find these dynamic xG projections to be more reliable than static ones that assume the status quo will remain relevant.  In effect, my sheets anticipate the cascading effects of predicted new data being added, and redundant data being subtracted.

xG GW25-30

The analogy I use is modern weather forecasts, and how they are based on computer simulations that evolve the state of the atmosphere forward in time using an understanding of physics and fluid dynamics.  They attempt to predict what the weather will be in the future, not what it is now.

Extrapolating from the values my spreadsheet assigns to every team allows me to do many things, including sorting teams by the predicted number of expected goals that will be scored and/or conceded over any number of future gameweeks for any range of gameweeks desired.  It also enables me to produce weekly correct score predictions.

An important distinction to make here is that what my spreadsheets predict is ‘expected goals‘, not actual goals.  As can be observed in the match stats shown during Match Of The Day post-match interviews with club managers, these often don’t correspond with each other.

Before coming to the vexed question of what the point of generating theoretical goals is if they can differ markedly from goals scored in reality, I should add that over the course of a season there’s usually very little to separate the total number of goals scored in the Premier League from those expected by the xG model I use.  Last season, for instance, total expected goals for all teams only exceeded actual goals by 46, which works out as an average of under 0.1 of a goal per match per team.

Why xG? 

Have you ever watched a game of football and seen the team with the biggest chances to score goals lose?  Then there’s your answer.  Before I started playing FPL, unfortunate events like calamitous errors by individual players, poor refereeing decisions, and unlucky deflections, were all just grist to the opinion mill.  It was only when I started recording match results onto my spreadsheets, and saw the distorting effect such simplistic data had on my team rating equations, that I came to realise all goals ought not to be accorded equal significance.

After 7 matches last season, Crystal Palace had yet to score any points or goals.  By the standard measure of actual goals they were destined to drop down into the Championship.  Clearly, they were a team who couldn’t score goals, and on that basis, wouldn’t score the goals needed to avoid relegation.  The Expected Goals model, however, told a different story.  One in which they were cast as merely the unlucky victim of variance who had experienced the unfairest results up until that point.  And the villain of the piece was our old friend ‘standard deviation‘.

Even before I came across xG on my fantasy football travels, my homegrown variety had me start Leicester’s title-winning season with Vardy and Albrighton in my FPL squad (alas not Mahrez), at a time when they were very unfashionable picks.  My DIY xG spreadsheet had identified Leicester as a vastly underestimated team judging from performances during their ‘great escape’ the season before.  I remember well my increasing exasperation with radio and television pundits alike midway through that season as they all declared relegation for the Foxes a foregone conclusion.

These, and countless other examples besides, have led me to confidently conclude that xG gives a more accurate picture of the attacking and defending strengths and weaknesses of teams than do the Goals For and Goals Against columns in a league table.

Predictably enough though, there are still large swathes of the FPL community yet to embrace this revolutionary way of understanding football matches.  My twitter feed is still littered with uninformed commentary and misguided sarcasm, from some of the most followed twitter accounts, at the expense of the xG approach.

Flat Earth in space with sun and moonTo my mind though, these snipers and swipers are akin to ‘flat earthers’ denying the earth is round.  I expect Tony Bloom and Matthew Benham wouldn’t have any sympathy for these conspiracy theorists either, given that they were able to buy football clubs (Brighton Hove Albion and FC Midjylland & Brentford) with the proceeds from their xG-based sports betting operations.

What’s The Plan? 

I’ve been refining and perfecting my spreadsheets for several years now, and they’ve helped me to sub-twenty thousand finishes in three out of the last four years, the last of which was a decent 4,733rd.

What I’ve enjoyed most about letting my xG spreadsheets govern my decisions is that they often promote going against the grain of template teams, against the flow of groupthink, and against the tide of crowd wisdom.  And yet, the maverick moves I’ve made have generally kept me ahead of the curve.

The time and effort I put into my spreadsheets has sometimes been hinted at in screenshots I’ve shared on social media posts, but the time has now come for me to make them available to other FPL managers.

The elusive nature of form in football, and the sensitivity of my algorithms, means that fluctuations in my player and team ratings are inevitable, but if my spreadsheets perform well, these should be gradual rather than volatile.

My spreadsheets will not eradicate difficult decisions regarding captaincy and transfers, however, and should be used alongside managers’ judgement and knowledge, not instead of them.  Getting the most from my sheets will depend greatly on synthesising them with managers’ instincts.  The onus will still be on managers to make adjustments and allowances for events like key injuries (and returns from injury), suspensions, transfers, etc., the significance of which cannot be immediately captured by my spreadsheets.

Reality check 

The most common scoreline in the Premier League last season was one-one, which happened just under twelve percent of the time (11.84%).  The next most common score was one-nil (11.58%).  And then there were just as many nil-nil draws as there were two-one wins (8.42%), meaning that in just under a third of all matches (31.84%) neither team scored more than a single goal.


Using a poisson distribution applet I was able to calculate that the most probable that a 1-1 scoreline could ever be is 13.53%.  That’s longer than 6/1 in fractions, but shorter than 13/2.  Accepting any odds of 6/1 or less (11/2, 5/1, 9/2, etc.) for a 1-1 score draw, therefore, would mark you out as a ‘mug punter‘.

The implications for successfully predicting scorelines are considerable.  Even if we find 10 matches on a football coupon that all had the highest probability of a 1-1 scoreline possible, the chances of getting at least 3 out of the 10 correct will never exceed 29.75%.

In other words, don’t be calling my spreadsheets out if they only get one or two score forecasts correct each gameweek, because in  reality, achieving 3 out of 10 with any more regularity than once every three gameweeks is against the odds.

As for correctly predicting all the results of the free-to-play SkySports Super 6, you will never have higher than a 0.00061% chance of winning the jackpot.  That’s a one-in-one hundred and sixty two thousand, seven hundred and fifty two chance (162,751-1), in the best case scenario.  Little wonder then that I’ve never won it!

If you were in any doubt about the random nature of much of what happens on a football pitch, then these startling odds should help you better understand the enormity of the task faced by those trying to provide accurate predictions.

If you understand predicting scorelines is difficult, then you will realise that forecasting who will be doing the scoring and assisting can be an even more unpredictable business.  Like weather presenters assuming the weather tomorrow will correspond with that of yesterday, much of what passes for FPL punditry too often presents evidence of what has happened in recent gameweeks as incontrovertible proof of what will happen in future gameweeks.  In my experience though, such simplistic thinking in FPL rarely work out how we expect it to.

Health warning 

Finally, I should warn you about the dangers of dependency.  Use of spreadsheets can become seriously addictive.  Never binge drink on them as they can really go to your head.  And they might give you dutch courage to make maverick moves that leave you with a really bad overall rank hangover.

Please drink responsibly. 


Coley aka FPL Poker Player @barCOLEYna





Challengers come and go, but my spreadsheets remain undefeated!

It’s five weeks since my last blog If Carlsberg Did FPL Spreadsheets…. in which I announced my pioneering spreadsheets, and invited interested FPL managers to test-fly them in the following 4 week period (GW5-8).  I have said all along my sheets would need the first 8 gameweeks to be fully up to speed, so now seems a good time to bring you all up to date with how things have gone.


Thanks to my test-pilots, anomalies were identified, teething problem fixed, and tips received that ensured smoother functionality of sortable table columns, as well as a number of other breakthroughs and improvements made.

My initial plan to supply my test-pilots with player ratings, however, was abandoned after our first mission.  My belief that trying to predict individual player points is a fools errands was only deepened by what unfolded in GW5.  I am firmly of the view that averaging out aggregate player xG/xA values, short-term or long, fails to recognise that player points tend to polarise between high and low scores.  So I returned to my original position that the best approach is to advertise teams with the best combined form and fixtures, and leave it to individual managers to determine the best fit transfers for their teams thereafter.

That said, I have looked back at the player table I provided my test-pilots with prior to GW5 and, with the benefit of hindsight, they look pretty damn good!  xFPL P DEF FWDIf we ignore Bryan and Davies, who only had one start between them, 6 of only 7 other defenders highlighted (Doherty, Laporte, Trippier, Monreal, Robertson and Alonso) were all amongst the top two dozen highest scoring defenders (1st, 4th, 6th, 14th, 21st and 24th respectively) during this period.

Obviously, the standout pick here was Doherty, who had averaged only 2 points per game at that point remember.  Bringing him in on my GW6 wildcard was one of my better moves, and coincided with my ranking improving from over 233K to under 53K in the space of 2 weeks.

Something else I did during the trial period was to compare and contrast my spreadsheet’s predictions with those of a well-known predictive fantasy football algorithm, marketed as the ‘world’s most powerful.  Small sample size notwithstanding, my algorithm outperformed the paywalled one in all but one respect.

My spreadsheet correctly forecast more team goals, more match scorelines, more accurately the number of goals scored, and had a significantly better correlation between predicted and actual goals as measured by Mean Absolute Error.  Odd then that it predicted fewer correct results.

The main selling point of my spreadsheets, however, is the way they are able to anticipate patterns and trends before they’ve even emerged by projecting forwards the long-term implications of short-term predictions.


One of the things I’ve enjoyed most about tweeting screenshots of my spreadsheets is the way that they are like a red rag to a bull for some members of our FPL community who cannot help but contest some of the more leftfield predictions that my sheets make.

Going against the grain

I should confess here that my somewhat contrarian nature is well suited to my sheets.  I like that they regularly highlight maverick moves, because that is where most differential value is to be found.  I do understand though that most FPL managers don’t have the requisite nerve for going against the crowd quite as often as I do.  Maybe that’s the poker player in me.

By the way, it does amuse me when my sheets are objected to for not being in line with received wisdom and/or bookies odds.  After all, what would be the point of going to all the trouble of generating predictions if they merely mirrored other widely available resources?

Mainly though, my enjoyment of the challenges to my spreadsheets stems from the fact that they usually backfire on the doubters and knockers.  For example, before GW5 my sheets were challenged for singling out CPL as the clean sheet banker of that gameweek.

AWB 62% CS - Crop

Admittedly, an entry error subsequently came to light that scaled down their probability odds from 62% to 55%.  Even so, in bookmakers terms, that’s about a 9/11 chance, so an odds-on favourite, whereas odds were available to back them at 7/4 against, which equates to a probability of around 36.5%.

MCI and WLV were the only other teams to keep a clean sheet in GW5, which could have been highly significant with so many managers having benched Palace defenders after their failure to deliver on an expected clean sheet the previous week.AWB vs Ralls - Crop

Courtesy of my sheets though, I’d benched Wan-Bissaka the week before, and swerved his zero points return, and now my sheets had me moving in the opposite direction to the majority once again.

Unfortunately for those of us who purposely started Wan-Bissaka in GW5, our advantage was negated by the unexpected absence of Mendy that weekend.  This meant most owners were spared the indignity of having a player with 9 points sitting on their bench.

Screenshot_20180929-142240_WhatsApp - Crop

Before GW7 I was asked my opinion about transferring in Arnautovic before WHU faced up against MUN.  Whilst understanding the reluctance to take a hit to do so, I felt obliged to share my sheet’s forecast that the Hammers would score twice.  In fact, they scored three times – stupid sheet – and predictably enough, Arnie was amongst the goals, because if ever the FPL Gods have an opportunity to punish managers exercising patience and showing restraint, then sure enough they’ll take it.

TOT 1 goal thoughtsMore recently, I was asked (politely) by one of my test-pilots to account for my sheet’s prediction that TOT would struggle to score more than once vs CDF in a week when so many FPL managers were taking hits to bring in and captain Kane.  This enquiry came in response to the following excerpt in the GW8 preview I emailed to all of those in receipt of my spreadsheet tables.

leaked GW8 document - Crop 2

Kane shyThis was another big call by my spreadsheet, and one I heeded by transferring in Lacazette and leave 3m in the bank, rather than bring in the most transferred in and most captained player of GW8, purely because my sheet rated the chances of goals to be higher for ARS than TOT.

The move would have been a master stroke, but for FPL’s uncanny habit of bringing us down to earth.  The player I sold to bring Lacazette (12 pts) in was Wilson (14 pts).  D’oh!

MCI 44% CS - CropIt was the final game of the last weekend, however, that arguably represented the biggest test yet for my spreadsheets, as its forecast of a MCI clean sheet (projected before GW6) was met with scepticism and barely concealed ridicule in several quarters.

MCI CS Anfield record




MCI 44% CS - Crop 2

Like Peter denying knowledge of Jesus after the Last Supper, I was guilty of making excuses for my spreadsheet’s prediction.

In fact, by the time GW7 results were taken into account, my spreadsheets position on the probability of a clean sheet for the champions at Anfield had hardened from 44% to 56.5%.

MCI greater than 50% CS - Crop

Going into GW8 though, I began to repent my earlier disowning of my sheets, and posted a rationale for having faith in their prediction.

leaked GW8 document - Crop

Not only did MCI not concede, they rarely looked like doing so.  Furthermore, my sheet’s 0-1 forecast was only prevented by a Mahrez penalty kick miss that is probably still in orbit around the Moon!

As I’ve said many times before, the thing with stats is that they are only true until they are not.  They tell us what has happened in the past, not what will happen in the future. From my perspective, the LIV vs MCI result illustrates this point perfectly.

In the interests of balance, I ought to acknowledge one public challenger has yet to fall by the wayside.  Prior to GW6, the prominence given by my sheets to BOU defence for their next 6 fixtures run was publicly disparaged by one manager.

BOU 2nd best GW6-11

Although misinterpretating my Clean Sheet Probability table because teams with the same number of predicted shutouts were shown in alphabetical order (CHE were actually rated as 2nd best), the shock 4-0 defeat suffered by the Cherries at Turf Moor did little to contradict such scepticism.

We are only halfway through the 6 week period in question, however, and the jury is still out as to whether the cynicism was fully justified.  It is worth noting that only six teams (MCI, ARS, TOT, CHE, LIV and WLV) have more clean sheets than the one BOU gained last time out vs WAT over those first 3 gameweeks, and I have said all along that my spreadsheets will not be firing on all cylinders until they have 8 gameweeks of data.

That success at Vicarage Road notwithstanding, Bournemouth’s defence ranking for the next rolling block of 6 fixtures (GW9-14) has fallen to 15th, so my sheets may not remain unbowed for much longer.


I have been back in touch with the 5 managers who missed out on my last intake of test-pilots, and am pleased to welcome them all to my squadron.  Having gained experience in distributing reports, reviews and spreadsheets to a large crew of people over the past month, I now feel able to expand my operation further, and repeat the exercise all over again for the next 4 gameweeks (GW9-GW12) leading up to the next international break.

If you are interested in trying out my spreadsheets, and willing to give constructive feedback, please RT and DM me to register your interest.

In the meantime, it would be very much appreciated if the naysayers could continue throwing down gauntlets, as my spreadsheet predictions seem to thrive on them!

And remember:  Flying isn’t dangerous.  Crashing is what’s dangerous!  To your overall ranking that is.


All FPL accounts and tweets in this blog – even those based on real people – are entirely fictional.  All FPL account tweets are recreated – poorly.

No feelings were intended to be harmed in the making of this blog.