UK Power rankings

[Wayne J]

My views on the current IFPA ranking system, which rewards more play and doesn’t punish poor play, are well known.

Many people have said if you don’t like it run your own ranking system – so here it is.

The IPFA has a Power 100 system. The way this works is it looks at all of the players ranked in the top 250, and compares their head to head records to give a “win percentage”. I think that this is a much better indicator of a player’s strength.

The way it works is that if the top 250 played in a single comp and a player finished 1st they would have be recorded as wins 249 (as they finished higher than 249 others competing) – ties 0 – losses 0. Giving a percentage of 100%. If a player finishes 125th they would have a record of wins 125 – ties 0 – losses 124 giving a percentage of 50.2%. The ranking system compares ALL competitions where 2 or more players in the top250 have competed. Ties are ignored in the calculations.

Obviously if a comp only consists of 2 players their percentages for THAT competition would be 100% and 0%, but their wins and losses would only increase by 1, not 249 as in the previous example. This gives a natural weighting to competitions with more players in the top 250 as the final percentage is based on TOTAL number of wins and losses, not individual tournament percentages.

With a LOT of help from Paul @Wizcat and access to the IFPA records for ALL of the registered tournaments (no matter where they were held) direct from their database - I have utilised the same system for the UK players.

Due to the fact that it takes so long to run through all of the data, there have had to have been a couple of conditions put on which players are being compared.

To qualify for the ranking, the players must be ranked in the top 150 in the UK and they must have competed in 9 tournaments in the last 3 years. This ‘qualifies ‘ 94 players included in the report. It’s worth noting that of all of the UK players, only 47 have played 20 comps or more.

The reasons behind this are to ensure that the report does not take forever to run, comparing and searching for results of players who may have only competed in the UK League or at a single event; and that statistical anomalies are filtered out from the offset.



The full final ranking table is attached, but the top 20 stands at:
Craig Pullen 85.64%
Andrew Foster 84.46%
William Dutton 77.33%
Matt Vince 76.98%
Wayne Johns 74.61%
Martin Ayub 74.46%
Rich Mallett 73.17%
Nick Marshall 71.28%
Greg Mott 70.03%
Garry Speight 66.67%
Andrew Shillabeer 65.59%
Peter Blakemore 65.08%
David Mainwaring 63.65%
Phil Dixon 62.75%
Chris Poyntz 62.52%
Clive Bush 61.11%
Tim Porter 60.68%
Kate R-Jackson 60.32%
Andrew Wilson 59.88%
Ivan Miles 58.23%

Phenomenal records for Craig @roadshow16 and Andy @PUP

Feel free to make of it what you want; it’s certainly not going to change anyone’s life.

Personally, from looking at the players I know and have played against the most, I feel that this is a much more accurate measure (backed up by the data) of each players comparative strength compared to the current WPPR system of ranking.

 

[Wizcat]

I'll try and get the code up on Github over the weekend, if any fellow developers want to have a looksee. Or if any foreigners want to run a similar list for their country ( @Martymainframe ?) and want the tool, just send a PM with your email address this way.

 

[roadshow16]

Nice @Wayne J and @Wizcat

Thanks for this, interesting data that I’m sure took a long time to go through and some interesting results.

I would like to see where Martyn R and Lukasz R would feature in an all time list, don’t know if that’s possible or not.

An immediate reflections on this list/power ranking. Andrew Shillabeer outranks Peter Blakemore on this list based on these results. And yet, in their individual H2H rankings, Peter beats Andrew 37-6. If it was close, i’d say, fair enough. But that is a completely conclusive stat. Peter is clearly a stronger player and has proven it many many times. And yet based on these rankings, Andrew is higher. I understand why that has happened, but can’t agree with the order based on that finding

 

[Wayne J]

Nice @Wayne J and @Wizcat

I would like to see where Martyn R and Lukasz R would feature in an all time list, don’t know if that’s possible or not.

Martyn has only 1 event in the last 3 years (UK League in 2015), and is currently ranked 260th in the UK and 10270th in the world, hence him not included. The report only considers results over the last 3 years, so will change almost daily.

Tim R was ranked in 5th place on the initial run which included everyone. However I felt that this wasn't a suitable report as it drew up too many anomalies with people playing too few comps. (3 people appeared in the top 10, and 6 in the top 20, who had played too few comps to make a single decent comp push them way up the list.) There are a few other notable examples of players not appearing anywhere on the list due to lack of participation in the last 3 years. Dom L, Simon P, Steffan P, Andy J, Darren B, Chris E are just a few. These are people who mainly play just in the leagues.

This time the report was ran with players who are classed as being in the UK from the IFPA website, hence why Louise Wagensonner appears on it. I think the next time there is value to add those players who, although foreign nationals, reside or mainly compete in the UK. This would include Lukasz R , Thomas and Oscar O, Sverker N, Jason P & probably others I can't think of right now and remove those who don't reside or compete in the UK.

An immediate reflections on this list/power ranking. Andrew Shillabeer outranks Peter Blakemore on this list based on these results. And yet, in their individual H2H rankings, Peter beats Andrew 37-6. If it was close, i’d say, fair enough. But that is a completely conclusive stat. Peter is clearly a stronger player and has proven it many many times. And yet based on these rankings, Andrew is higher. I understand why that has happened, but can’t agree with the order based on that finding

There are a few, and only a few, anomalies like this.
I sense checked the vast majority of players on the list and in general the results hold up. Where discrepancies like Andrew and Peter arise is when players play in a more enclosed group - SWL, Tilt & Flipout for example and then have better than expected results in the UK League or other big competitions.

Looking specifically at Peter vs Andrew. Over the last 3 years the head to head record is 8-1-1, with 5 of those results coming from leagues at SWL. The only results for the league I can find online show that Andrew only attended 1 of the 3 meetings (when he finished 4th= to Peter's 1st=), so had no chance whatsoever of finishing above Peter over the 3 meet season. For the 4 'on the day events' (2 of which were held at SWL) their record is 2-1-1, much less of a discrepancy.

 

[Wizcat]

I don't know Andrew Shillabeer, so can't comment too much, but just for reference here is an unfiltered list with no-one removed
(Head to head results of top 150 UK players by IFPA rank over last three years)
https://docs.google.com/spreadsheets/d/1k9Isegw3n_V5kFxk_7_f2qQvH-VMLqe4JCfNQxwE3XI/edit?usp=sharing

In this one Andrew is one place below Peter, and the difference is the head to head results against players that are otherwise excluded. I'm still of the opinion that an unfiltered list might be a better option as otherwise we're filtering out some really great players
(Though I do appreciate there are some quirks in the list such as Jon M who looks to have only played the league, and now I look I see he is actually considered unranked on IFPA. Same with Frederico. Maybe I can do something to exclude unranked players...)

 

[roadshow16]

As in discussion with Josh when the POWER100 came about, I suggested there being a certain number of H2H’s that a player had to hit in order to make it on to the list. In the end Josh also went with the unfiltered version, but maybe that could work for the UK?

From that list I can’t understand why Tim R has been removed and Gary S is in? How did you draw the line?

 

[Wayne J]

9 comps in previous 3 years was the cut off I used (as in original post ?).
I did it all manually, checking every individual players profile on the website. So there is the potential I miscounted one person's tournaments, but not in this case.

The difference between the UK Power and the Power 100 is that the Power 100 ranks against the 'top 250' players who have all played well over 20 comps
The same can't be said in the UK where there are only 47 players who have reached 20. The average no. of comps played per person in the UK top 150 is less than 12.5.
There just isn't the same volume of data to make it more accurate.

I could run it for the 47 players who have all played 20 comps or more, but am certain that would exclude a good number of players.

 

[Wayne J]

The thing to take notice of, isn't so much the actual ranking position, but the percentages.

Statistically speaking, with the amount of data examined 65.59% & 65.08 are effectively the same.

The new list also changes the position of Phil D, who climbs above both of them.


The problem with trying to change the parameters based on 1 or 2 players you think are misplaced is that you can tinker with it until you get the results YOU think.
It's something I mentioned to Paul before I even produced the initial published report.
I could set parameters which could improve any number of individuals positions.
The problem is then, that it is me manipulating the data to produce the results I think should be produced. Not using reasoned and thought out parameters BEFORE running the report, and then analysing the results.

(Can you tell I studied statistics at Uni?)

 

[johnwhitfield]

The bottom of the list is interesting. Becca makes the cut off so there are a few people below a 10 year old child. CJ is only a couple of places above her despite being a great player. Not sure why he would be so far down the list as Id personally rank him as a top 20 UK player or even higher

 

[Wayne J]

The bottom of the list is interesting. Becca makes the cut off so there are a few people below a 10 year old child. CJ is only a couple of places above her despite being a great player. Not sure why he would be so far down the list as Id personally rank him as a top 20 UK player or even higher

Again, it's down to the UK League being by far the largest comp so it has more weight than a number of other comps.
Last year CJ finished 231st, the only UK league he's been in recently (without looking - I'm guessing he only competed in 1 of the 6 meets, and did poorly, meaning he finished near the bottom of the L & SE league, which when multiplied across the 6 regions results in the poor finish).
I'm sure his position will vastly improve when the results for this year's League are submitted). This means just for that comp his No. of losses will far outweigh the number of wins. You've got to enter a lot of smaller comps (like Flipout League) to start turning that around and increase the win %ge, or wait until that result 'expires' after the 3 year cut off

 

[ianw]

Sorry guys, drink beer and play pinball

 

[Nickoss II]

Yay, love stats! I'm heading the leaderboard for most losses in the UK! @thelab you've got some catching up to do!

 

[Deleted member 2463]

TLDR;

another scoring system with equally as many flaws as IFPA

 

[JMP]

Liking this cos. my local competitor @lukewells isn't on it

 

[Wayne J]

I can't find a valid argument to complain about this so I'll just make something up.

?

 

[Deleted member 2463]

?

you provided all the evidence required in your followup posts!

All of these systems make a lot of assumptions based on players who may not have actually played each other in any actual game; but took part in a competition together each with wildly different rules and scoring, in my view thats just another flawed view of the same flawed data and the movements of who are good players versus who isn't hasn't changed a great deal. If you want to know who is the best player, then you need to ensure that they all play together in a consistent competition form.

but Wayne -

If it makes you happy inside - then thats all that counts

Whats great is that my position at the bottom of the table is constant throughout!

Regards,
Neil.

 

[astyy]

To give me some chance, I propose a ranking system based on alphabetical christian names, one caveat @aaronhicksuk is banned

 

[Wayne J]

you provided all the evidence required in your followup posts!

All of these systems make a lot of assumptions based on players who may not have actually played each other in any actual game; but took part in a competition together each with wildly different rules and scoring, in my view thats just another flawed view of the same flawed data and the movements of who are good players versus who isn't hasn't changed a great deal. If you want to know who is the best player, then you need to ensure that they all play together in a consistent competition form.

but Wayne -

If it makes you happy inside - then thats all that counts

Whats great is that my position at the bottom of the table is constant throughout!

Regards,
Neil.

So what you're saying is that both ranking systems are flawed and pointless?
I may well agree.

 

[Deleted member 2463]

So what you're saying is that both ranking systems are flawed and pointless?
I may well agree.

Not completely pointless but flawed for sure. And many ranking systems are the same.

I think they are a reasonable indication of improvement over time.

As a point in time guide I think they do a decent job but nobody remembers a great player because of their ranking - they remember them for what they win.


Sent from my iPhone using Tapatalk Pro

 

[Wayne J]

Not completely pointless but flawed for sure. And many ranking systems are the same.

I think they are a reasonable indication of improvement over time.

As a point in time guide I think they do a decent job but nobody remembers a great player because of their ranking - they remember them for what they win.


Sent from my iPhone using Tapatalk Pro

My point exactly. No system can ever reliably be used to rank the best player.
Who was the best Pele, Maradonna, Messi, Ronadlo? There's arguments for all of them - except Messi after recent performances ?
If there was betting on pinball, I know which system the bookies would be paying more attention to, and when was the last time you saw a poor bookie.

My objection is, and always has been, using the IFPA rankings for seeding pre-tournament, and invites to the major tournaments based on the ranking, even though I have benefitted from pre-registration in the past.

 

[domlouis]

My penny worth: there is no way I am in the top 25 players in the UK as shown in the unfiltered list. So that unfiltered list is wrong. As Wayne originally worked it out you need a large number of results for the stats to start to make sense. So it does need filtering somehow.
Bottom line for me: I don't care where I rank, I just want to play!

 

[Replay]

Not completely pointless but flawed for sure. And many ranking systems are the same.

I think they are a reasonable indication of improvement over time.

As a point in time guide I think they do a decent job but nobody remembers a great player because of their ranking - they remember them for what they win.


Sent from my iPhone using Tapatalk Pro

True.

Golf/Tennis - no one remembers what ranking a player achieved.

It's how many titles, specifically major titles.

Would be quite cool if pinball was able to have similar 4 majors, with one being held in Europe/UK.

 

[Matt Vince]

Seeing as how results from leagues are also relying upon attendance, I would be interested to see this list without the uk, london, swl Tilt & Flip Out leagues (are there any more?)

 

[Wayne J]

Seeing as how results from leagues are also relying upon attendance, I would be interested to see this list without the uk, london, swl Tilt & Flip Out leagues (are there any more?)

I'm sure it's possible, as the Power 100 on the IFPA site has been created with and without leagues.
@Wizcat would have to investigate

However I think then that the Tilt highscore & Flipout Highscore comps ran alongside the leagues would then make even more of a difference, I don't think they can be filtered out.

 

[Wizcat]

Yep, the latest version I'm working on can conditionally exclude league events. Also (conditionally) only adds players with more than 5 lifetime events, which should remove some of the quirks I hope. Will try and send you an update later Wayne

 

[James]

Tell you who will be the best at Daventry though...

The person who walks away with the trophy. FACT.*

*FACT is subject to intepretation, items presented as factual may in fact not be fact, but present in a factual sense. No purchase neccessary.

 

[DRD]

You are being too modest Dom. I remember you winning the Midlands league meeting at my place a couple of years ago. Winning any league meeting with 25 plus players over 8 or more machines is no fluke. One awful game is enough to cost you a victory in this format

My penny worth: there is no way I am in the top 25 players in the UK as shown in the unfiltered list. So that unfiltered list is wrong. As Wayne originally worked it out you need a large number of results for the stats to start to make sense. So it does need filtering somehow.
Bottom line for me: I don't care where I rank, I just want to play!

 

[Matt Vince]

I'm sure it's possible, as the Power 100 on the IFPA site has been created with and without leagues.
@Wizcat would have to investigate

However I think then that the Tilt highscore & Flipout Highscore comps ran alongside the leagues would then make even more of a difference, I don't think they can be filtered out.

Perhaps any comp under say a TGP of 20% should be excluded? If a comp requires less than 4 games to win, then there is an argument to say that they should be excluded from the rankings anyway (at least in a custom rankings system).

 

[yellowreggae]

I agree that this is a very useful way to look at how consistantly good UK players are. However that's not how we are ranked and therefore many players I suspect play according to how the system works. If consistancy isn't rewarded then that sometimes may affect a player's decisions during a tournament or league. For example, a top player's lowest event score might be 7. So knowing that if they don't get at least that many it won't improve their ranking they might either not take the tournament seriously from the start or perhaps stop trying once they realise they can't make the finals for instance. Maybe you qualified and eased back. At no point are you thinking about your win loss ratio...

Maybe some players play well even when it doesn't matter and maybe some players turn it on when it matters.

Maybe. The point is, the rules are layed out and we play those rules. You can't change the rules post play and then appoint new winners.

That said I do think this is a very good way to see who the most consistantly good players are and its obvious to anyone who's played with them a few times that Craig and Andy deserve recognition. Thanks to Wayne and Paul for compiling it.

 

[Matt Vince]

What everyone should do is head to Flip Out this weekend and join in one of the comps on Saturday or Sunday! (shameless plug - http://flipoutlondon.pinballnews.com/tournaments.html)