"The number of the end is the median (average) Standard Scratch Score (SSS) for each parkrun." - why would the median be preferred over the mean? Something about down-weighting outliers or de-skewing?

@@M3WDD Yeah both. Also it allows me to present a number to 0.1 exactly the same as the SSS scores given to each parkrun event or race. I compute the SSS to an "exact" number but store it and use it to the nearest 0.1 so there is some element of being unsatisfactory in computing a mean to many decimal places when using the rounded SSS figures. Ties are separated by the mean. Mostly the mean and median are very similar. For instance the mean at Victoria Dock is 0.959 and Great Yarmouth North Beach 9.814 compared to 0.8 and 10.0. Both would also "win" by that criteria although Storeys Field is only 0.003 higher. Victoria Dock also has the lowest 25% "median" at 0.5 and ties with the 2nd and 3rd for the 75% figure at 1.2 so those three are almost always "quite" fast.

This is very interesting and certainly corresponds well with the parkruns I’ve tried. Has Power of ten/Runbritain rankings ever released a dataset with all SSS scores by parkrun and date - it would be interesting to correlate the difficulty level with other factors (e.g. weather, perhaps over time)!

Thanks for dropping by and thanks for doing Great Yarmouth North Beach to sort of "prove" I am not talking nonsense - well at least some of the time :) Also thought better stop you making a beeline for Berkeley Green given it has now dropped down somewhat. Yeah I think of the ones I have done Dulwich, Poole (albeit on a very crowded Xmas Day) and perhaps Worthing (albeit with one dead turn) felt most like a 5K road race course. Each one though think need to pick your day. When I did Dulwich as well as being not very fit I did find that drag up the past start/finish zapping and it was a bit windy that day. Worthing is a classic prom one - could be very very quick on a still summer morning but a slog in a gale and Poole is pretty flat Tarmac (mainly), no dead turns, and always has a decent field so plenty of people to chase. As mentioned I picked the worse day to do Victoria Dock as was 40 mph winds. I see from other comments the course is currently just on the north side. It looked very good weather for your run there - warm sunny summer morning and I always find that helps if only getting out of bed first and foremost ! I guess another factor is that many 5K road races (say like the ones you have done of late in Battersea Park) tend to attract runners who are firstly quite good and then are trying 100%. The algorithm "sort of" adjusts for that but it is hard to weed out those trying say 90-95% which would be more common in parkrun. I wonder if that Great Yarmouth one had more runners and say they were in XC spikes not Invincibles then likely to at least go sub 20!

Thanks a lot! The only ones I have done there is Lanhydrock & Eden Project which are the opposites ends of slow & fast respectively. Have you done the Torbay one?

Thanks and good to know. I even had the More or Less team on the BBC ask me about Millom as they did a feature on the then hardest (or slowest depending how you view it) parkruns and sent a reporter to go and run it in its XC days of course. It now seems relatively quick. I just off for a track 5000 which is 12.5 laps so in a way 8 laps does not seem so bad.

Thanks. Yeah that one I am sure is tough. I have done Hove Prom and even though it was windy and recall having quite a few dead turns it did seem fast. Probably one of those that would be fast still on a calmer day - bit like Worthing down the road a bit as it were.

Yeah I think there is always a bit of a difference between easy/hard courses and slow/fast ones as an "easy" course should just allow you to go faster with the same effort. Hilly and/or muddy courses can be mentally tough though especially if feel are grinding to a near halt at times in the "hardest" bits.

@@TimGrose Yes Tim.I suppose everyone (including myself)goes through a bad patch during a race.When doing a parkrun I think to myself:well,I've done a few 10ks.Therefore this parkrun will not last for ever.Obviously only half the distance👍

Ah that's good to hear! I will hope to try it at some point.

Oak Hill ? Not done that one but I see is Tarmac laps.

@@TimGrose yes is fully tarmac apart from the last 3 meters where it finishes on the grass

A bit surprised that Springburn is ahead of Drumpellier, I have ran at both, all my PB's have been at Drumpellier.

I think Lanhydrock is one of the few times ever in a "race type effort" that trying to run seemed little different to had broken into a walk when was going up a hill towards the end! The differences between these courses on average I think are basically smaller than how you feel on the day coupled with the weather etc. For instance, my fastest time in recent years was at Redcar which is just outside the Top 100 but it was a windless, fairly warm summer morning and I felt good and it was "competitive". Also one of my first tries in the NEXT% which helps too! Hard to get any better conditions. When I did Victoria Dock there were 40 mph gusts and running into the wind was "interesting"! So was somewhat slower there that day. Is there an element of knowing the course too at Eden Project? "Unseen" I found it hard to know how much was worth pushing the downhill as the ups and downs after it were zapping if had set off too quick which I think I did. It wasn't great weather either. Bit drizzly and cold. I ended up with similar time to what I did at Victoria Dock.

Curious who else is ranking parkruns this way? I did this one about 3 years ago and was certainly the "slowest" one (relative to then fitness) I have done although I haven't done any of the ones I have in the 5 slowest. If it is now even tougher that is tough!!!

@@TimGrose It was one of the Marshall's that told me. They effectively reversed it, so you start just down from the home, two laps around respryn woods, then it's all the way to the top of the woods, then down and back round to finish. Only the fittest can run it all the way, I can only manage half and I'm fairly competent.

Yeah sounds like one that could be one of the very fastest on the "right" day but much less so otherwise. These are "average" conditions so need to pick your day if want to go fast.

I did that one in March 2018 when wasn't very fit and remember being pleased to just break 20. Has the course changed as seem to remember loads of dead turns ? And yeah it was windy!

@@TimGrose Ah actually it's 4 dead turns, you have two dead turns in each loop twice so 4 in total! I don't think it's changed from when it started but I've only been doing it for the past 6 months or so.

@@timeonfeet Ah yes that is more like how I remembered it. I recall those turns weren't too bad as plenty of room to swing round. As I mentioned in the video, my last parkrun was at Ifield Mill Pond which had 5 dead turns but all on a very narrow path which meant you basically had to come to a complete stop each time. Well that's my excuse :)

@@TimGrose Yes theyre not too bad at all at Hove Prom the turns as you have the big width of the promenade and usually the field spreads out a bit too so you're not falling over others!

Excellent information. Perth Road Runners are holding a winter challenge based upon four local Parkruns plus a wildcard of our choosing. This will assist me in making that choice. 🙂 Slightly surprised at St Andrews (undulating) being ranked higher than Edinburgh (flat). Probably due to the latter being prone to surface water and strong winds.

Thanks!

Yes I felt exactly the same about Dulwich and indeed I mentioned it in passing in the video. It is however possibly the most like a 5K road race course out of the 80 different ones have done.

Pretty sure nobody else is using “my” system to rank parkruns which only really works given access to all UK parkrun and race results that we have primarily for Power of 10 and runbritainrankings.com That said have seen other attempts at this say ranking by elevation gain but as may have seen a flat one at a beach is the very slowest in the UK by some way! You could probably get a good idea by finding enough parkrunners who are “tourists” and appear to regularly “try”. Indeed I started touring to test out my own theories.on all this.

Yeah Dulwich is always going to be a good bet. I did it when I wasn't that fit and certainly felt the little drag past the start/finish each lap and it was a bit windy that day too. Otherwise it could be very fast as Alex Yee demonstrated one time although he has of course run much faster on the track. As I mentioned the differences in the stats are very minimal. Dulwich perhaps little more than 10 secs from the very top. Looking at the Woolacombe course I presume some parts of it are somewhat "faster" than the Dune of Gloom? Could you use spikes on that course? That would be my choice at Great Yarmouth. Ben Parkes ran it in the Nike Invincible which would probably be my last choice LOL!

Thanks! In case not seen it, I put out an update a few months ago kzfaq.info/get/bejne/hLyPdpyFvpPYXY0.html

Yeah hopefully the algorithm largely minimises those factors by just comparing what each person does on the day to previously and there are numerous filters to ignore runners who don’t go to many different ones or indeed do too many at all. The parkruns with the smaller fields can be a problem due to the smaller sample sizes but averaged over a year it tends to smooth things out reasonably well.

Do you really mean 0.6 miles short ? That would mean Perth is only a 4K ?

Missed out a zero there. I Meant 0.06.@@TimGrose

Course changes could definitely impact things. This is just really about the event over a period and not particular courses which are largely impossible to track as what is used is seemingly never recorded. Sounds like Harwich is only a reasonably quick one. Fastest one I have done out that sort of way is Great Denham although I wasn't particularly fit at the time.

Been meaning to do Hastings at some point as got relatives there. So was the course long before or is it now short? The SSS median/average in this analysis is actually 1.6 and was 1.98 last time but the 1.6 period includes the previous period to give more data. Interestingly though the SSS median for 2021 only is 1.2 so definitely seem to be seeing that 40 metres in these stats. Where are you seeing 1.8 and 1.0 though?

@@TimGrose Ah I only used a sample from 53 events for my own curiosity. From GPS, Hastings seemed pretty spot on for its first 5 years now significantly shorter. The discrepancy is clear from past photos and two GoPro videos on youtube, however the race director seems as likely as Boris to concede an error now. Luckily I was able to go along the coast to Hove Promenade parkrun yesterday and run a 16.49 PB at a quicker pace on Strava than my friend who won Hastings in 16.38 but in my head a 4.95k PB would feel undeserved and be tough to beat elsewhere.

Have done all four of those but Crystal Palace was some years ago with the others more recent. I remember there wasn't really any flat bits in it! Yeah Dulwich is not overly "interesting" but when I am going for it all am thinking about is where is the end. Burgess was certainly pretty quick. Lot more turns but you can take most of them without slowing and just one dead turn as I recall which wasn't too bad. Also had the advantage of being able to get to as a warm up/down run to and from Waterloo station as well.

Yes good points. Have to admit that the data analysis here is largely restricted to those doing down to about 25 mins so hopefully those factors are minimised.

@@TimGrose interesting, so does it only use people who have a run Britain profile? Which if they are running over 30 minutes they are a lot less likely to have one

@@maxafc4695 it is more that the algorithm works by comparing what you do in races. 30 min runners have got vastly more scope for natural improvement than say 20 min ones so I figured it was hard to rely on that data.

Yes well reminded. Once we get the latest results into Power of 10 and runbritainrankings.com can have another look. I did a "preview" few months ago and looks like could be a bit of a shakeup at the top for the fastest ones.

@@TimGrose Great! I look forward to seeing the ordering. Thank you for taking the time to put them together

I just checked using 2022 data (which this video doesn't include) and actually Watergrove is 2 places above Lyme Park but both still in the 12 slowest ones. The difference however between parkruns this close is generally next to nothing. I haven't been to either but maybe Watergrove gets relatively harder in the winter which the 2022 data won't have as much as relative to a full year ? For sure both are very tough/slow - however you want to view it. Whinlatter still slower than both by some way whichever data period I look at.

Thanks for the heads up. These stats were from 2019-2021 so include a fair number post lockdown. Checking the stats for 2022 so far it is still certainly at least one of the fastest and the differences between one parkrun and the next in the list are often next to nothing. I will probably do another update at the end of the year.

Definitely not as fast as it used to be but still pretty fast. The nature of a dock is that it is pancake flat as it is a fixed height above the water. My PB (19:32) was done on the old course and I've had several recent sub 20 times on Storey's Field so pretty blessed as someone who splits their time between Cambridge and London.

@@TimGrose Hi Tim any plans on updating soon?

@@xiuyi yes! I am doing a marathon on Sunday so been largely consumed by that. Also need to check on a few courses that are coming up quick as they appear to be somewhat short of an actual 5K.

@@xiuyi haha i asked the same thing a few hours ago too :D

Thanks!

Ah thanks. I have started looking at an updated list and provisionally it has dropped down a little bit so the overtaking could well a factor. Presume the course itself is still the same ?

@@TimGrose It's still really quick and recorded a 14:34 course record on Christmas Eve 2022 but with 400 runners last week and the numbers seeming to continue to increase there is a heavy amount of traffic on the course which takes awhile to spread out. Course is the same each time I've done it (maybe 3/4 times).

Ah interesting thanks. I will have to see how that impacts the times as we go through 2022. The day I did it was about 40 mph gusts so sort of been meaning to go back on a "better" day. Still ran an OK for me time but remember one way was best part of a min/mile different to the other. I went out really fast that day at Hackney Marshes although it was a windless warm summer day and was somewhat faster in it than that Victoria Dock one. So the weather does play a big part week to week as I tried to explain in the video.

PS thinking about some more then presume it still has 2 dead turns and still pretty much pan flat so perhaps that is why have not seen any obvious different trends in the times as yet?

@@TimGrose I think I was capable of running it faster on the 2 occassions I did try them and was in decent shape but still struggled to get anywhere near my course PB. It's those out and backs which kill me, and there being 3 turns instead of the usual 2 and it doesn't flow as well as the original course.

Ah interesting. In some ways I don't really understand why you would design a really tough parkrun course unless there were no other viable options. The differences between these ones are often very small and I did my best time in recent years at Redcar which isn't that high so any in the Top 100 ought to be quick enough. Berkeley Green course now seems very twisty. I am preparing an updated list so will be interesting to see how it comes out now.

@@TimGrose thanks for the reply’s these rankings are very interesting it’s good to know I do 73rd out of 706, got a time of 17:52 for my pb definitely couldn’t at somerdale 🤣

@@George_b101 yeah there are some courses where doing an outright PB would be nigh on impossible if you have been running a while. These are also average conditions. If you pick a nice day weather wise, feel good and have some similar standard runners around you then every chance of running faster still which is sort of what happened in my Redcar example.

Ah! Which other parkruns do you do? How do the times compare?

@@TimGrose I do quite a few around South / South West London - Burgess, Fulham, Richmond, Bushy, Clapham, etc. They seem to appear in the right order on the list - the only surprise was Dulwich being slightly higher than Burgess, as I’m consistently quicker at the latter

@@petermartinuk Done most of them except Clapham. That comes lower than perhaps meets the eye? I did run relatively well at Burgess but was both lot fitter and probably better conditions than when I did Dulwich. Effectively though those 2 are pretty much the same. 0.1 SSS difference is a few seconds.

Just posted a link to the full list (or watch carefully at the end of the video). It is at 474th so a fairly slow one.

@@TimGrose just spotted it 😂 knew it would be low down 😂

@@BenNotley Yeah I have done it once (in 2012!) and checking back I was somewhat slower on it than most of the other ones I did around that time although it was my first parkrun (or race for that matter) after 7 months off due to a bad accident the year before.

That's good to hear especially as that is an area I have never been to for parkruns. Quick look at the 2 Hs ones and the fastest times do seem very similar but looking at the SSS scores this year Heaton Park definitely has more fairly high scores which would pull down the median/average. Is it perhaps more susceptible to differing weather and/or underfoot conditions? Also, although there is quite a few places between them, there is only a 0.4 SSS difference which is only about 12 secs for say a 20-25 min runner. Might argue such a difference is possibly within the margin of error with this stuff. Chances are then if you turned up for both in consecutive weeks then would most likely run very similar times.

@@TimGrose Yeah, hence why it seems plausible. Huddersfield is hilly enough that I wouldn't expect to see it near the very top, but Heaton has a choke point fairly early on through the wood. For the SSS's to be close feels right.

Ah indeed - that is going to slow it for sure. What’s the thinking there? Safer course, less in way of other users etc etc

@@TimGrose use of the park increased significantly during lockdown, it was a case of move the course to a quieter area or lose the event.

That would be tricky as we only note times for British runners in non UK parkruns. Even if we did record everybody we would have no reference data in “races” for the Irish runners. That said I will have to check if there is enough data to give any meaningful analysis. Which one do think is fastest and slowest?

It seems we have this under Sunderland - was that what it was originally called? parkrun names now are somewhat more "specific" as some towns can have more than one parkrun. Some parkruns have been renamed as a result - presume this is one such one. It is at 402nd with median SSS 2.8 so yeah does look to be a reasonably tough one.

@@TimGrose Oh OK that would make sense, thanks! I've only ran it once and at the time it was called Silksworth but maybe it was called Sunderland before that.

I have that problem nearly time as have now done 80 different ones out of 116 altogether. The last one I went to appeared to have done at all "nearby" so cue some discrete looking for some "cover" :) I usually want to find one "before" rather than after. If I have a bit of a journey sometimes stop at a petrol station or service station on the way.

Do check with the parkrun website for events you want to visit as they mention facilities in the course description. Also, and even better, check to see if Danny or Nicola have profiled the event on their With Me Now podcast and KZfaq channel as they take care to point out facilities including toilets, cafe and parking etc!

@@NickyJones99 yes that is a great suggestion.

I did Kings Lynn in 2015. I remember the park being fairly flat and all on Tarmac so quite fast but a fair few dead turns which certainly took some speed off. I guess it is difficult to fit a 5K within a fairly small park without having dead turns. Ran somewhat quicker than there though than the next week at the now defunct Fritton Lake although that was partly as went off course and took a while to get back on track as it were!

Hello James. Thanks for commenting. These stats include all parkruns from 2019 to 2021 so yes they would include the old and new course. However as I am taking a median of the SSS scores of each instance of each event then the median for this one will now be heavily weighted towards the new course. I figured it would be impossible to do it on just courses as there is nothing in the results API from parkrun we get that says what the course is and AFAIK this is not recorded "centrally". Also parkrun courses aren't like race courses and measured exactly and catalogued as it were. As such this analysis is reflecting the average conditions of each event - whether that is weather etc effected and/or just because of a different course. I did ponder about only including parkruns with a certain number of events especially when Berkeley Green came out top last time after just 5 events pre lockdown but figured it would be better to include every current parkrun rather than some saying "what about Berkeley Green ?" etc etc. At the end of the day this is just a bit of fun with the stats and personally I like to try different ones and not necessarily the very fastest each time. Out of interest, can you share the reasons behind the course change? It still looks a fairly quick course regardless and, as a sanity check, more age category records have been set post lockdown than prior.

@@TimGrose thanks for this. It's my curiosity. Slowing our course was a decision as we didn't want to be somewhere that people felt only fast runners could come to etc. Thanks.

@@jamessage1982 Ah that's interesting and did not expect that reply. I kind of wondered the reverse and whether the hardest/slowest courses might actually put some people as some of those equate to a good mile longer off but equally I think most just see the name parkrun near to them and do it regardless.

Just checked and Hafan Pwllheli has not restarted after lockdown so I excluded it as not run in 2021. Only parkruns "run" in 2021 are included. I also omitted all the "closed" ones (mainly in prisons). Maybe there are also other ones like Hafan Pwllheli in the 730 which technically still exist but on a hiatus?

@@TimGrose ah ok that makes sense, thanks for the clarification

523 Centre Vale 3.2 so a fairly slow one?

@@TimGrose Thank you...I'm officially post covid sight afflicted x

@@TimGrose yes, we're a narrow hilly course that post covid had to introduce a repeated dead stop and additional (repeated) uphill segment.

I've done both Burnley ranked 357 and Centre Vale, I recon Centre Vale to be 20s slower running at 25min,great course though. I really need to do Watergrove sometime......

Centre Vale is surprisingly tough, considering a lot of the park is flat, that stepped hill at the back end takes it out of you. Watergrove is very tough but in stunning surroundings. Also Shroggs Park in Halifax has a lot of short sharp hill climbs that have killed my legs by the last loop.

Good idea so I added an alphabetical list in the link in the pinned post. Have you done Sutcliffe? I am over to the SW (my nearest is Bushy Park) so one I could get to reasonably easily.

@@TimGrose Great & yes done Sutcliffe 4 times plus the test event, & set my PB there :)

It is yes www.parkrun.org.uk/victoria/ The names I am using are the "short" ones. Sounds like I have got those two parkruns at the correct ends of the list!

@@TimGrose Thanks for confirming, impressive work man. Best of luck with your running. 🙂

Well done! Will have to get to that one sometime. Excluding junior, defunct and ones in prisons I make it 762. That's quite a lot of runners for just one race/event !

@@TimGrose Wow, 762 parkruns in the UK. This is truly the 'Mecca' of Saturday morning 5k's in the world.

The names are what parkrun "officially" call them. So here presume the "OG" Edinburgh one www.parkrun.org.uk/edinburgh/ which looks like the one you mention as Cramond

Cheers Tim. Would love to see how Holyrood is ranked one day. Got a solid one-mile up hill right at the start.

I have got it! 34th Hove Promenade 1.4 so a relatively quick one. I did it a few years ago and agree very flat but the wind and dead turns I thought took something off which these stats do seem to illustrate.

Not really in an analysis of this scale. "Faster" runners may occupy a few places but am looking a long way down the fields to produce the stats. Good example is Andy Baddeley's parkrun "world record" at Bushy Park but, on this analysis, that is only the 86th fastest UK course. "Faster" runners must turn up everywhere as otherwise I may have collected more #1 tokens :)

@@TimGrose never heard of Andy Baddeley. Flat is fast as long as it's not muddy grass. Sand and mud and hills is slow. Simples ;-)

@@PoetWithPace Sure he will be upset! 2012 Olympian, 3:49 miler and our best miler for a few years and CEO of The Running Channel - over 400K subs

@@TimGrose nope, none the wiser!

I think there are a few that start/finish on a running track. Stretford springs to mind and that is one of the faster ones. I guess lack of space for a typical parkrun field and the confusion of multiple lapping means doing the whole 5K on a track would not be very practical. Agree though that in a smallish field it ought to be very quick relatively.

Yes hopefully soon. Just got a marathon on Sunday which has been my main focus and got to get some more info about a seemingly short course that makes it “fast” to better understand why it is.

@@TimGrose ok great looking forward to it

Hello Dave I have Bromley as 39th fastest with median/average SSS 1.5. So quick for sure. I did it back in 2011 and was certainly one of my faster times at the time. Where are you seeing a 2?

@@TimGrose at the end, it scrolled by so fast that I most probably mis-read it. Having done quite a few London ones, I was just amazed that Sutcliffe and Dulwich are considered faster than Bromley, as this is known as a very fast course

@@uncledavros There is a link in the pinned post where can study the list more carefully. Or just pause/rewind the video. All of those are relatively very fast. The differences in SSS probably mean no more than about 10 seconds so every chance if you ran one on one day you would go faster than another on another day.

Yeah can be confusing when several parkruns have very similar names. The one in question seems to be the one at Riverside Park, Chester-le-Street. 36th would certainly make it one of the fastest ones. Does that make sense? It is not one that I have done or really know anything about. Just how the numbers come out.

@@TimGrose i was gonna ask is it chester le, ive never been but i think locals here go there to run it, so it must be because its faster than the one here :) thank you !

Things like taking medians not means which minimises outlier effects, not including those who have not done many different events or races, those clearly not trying. But yes there will always an error margin. Any particular ones you are referring to?

@@TimGrose It was just a general thing, I would expect there to be some events listed as unranked (really any under a couple of thousand finishes, but it is hard to tell as the exact methodology is not published). At least confidence level should also be published as a gauge of how likely the score was not due to chance of who happened to have run each event.

@@abeck4391 I generally have a good look at the ones at either ends but otherwise it is easier to list everything as many comments are on the lines “where is parkrun x” ? And I hope nothing is missed - well that happened in 2021. Besides this is a bit of fun really and trying to keep the presentation reasonably understandable even to me. Talking of sampling error bands perhaps best left for university peer reviewed research papers.

@@TimGrose You could list it as "too little data" that is easy to understand or put a "Health" warning say that the calculation are just for fun and not guaranteed to be accurate

@@abeck4391 it's well known that parkrun is a run not a race, also only a tiny minority of courses have permanent start/finish lines or have been professionally measured. If people want super accurate courses then there's loads of races available where the entry fee covers all that.

459th. Median SSS 3.0.

Think that started this year but just missed the cut off which was end of 2021.

@@TimGrose ok thanks Tim, based on your rating system, which btw I think is great, I’d give it a 3.9😆sandy uphill slopes with twists and turns, tree roots hiding under wet leaves. I’ve only run 19 this year, however if Chichester is 3.0 then Rendlesham is definitely at least 30% harder IMHO. Look forward to your official ratings update. Keep up the good work 🏃🏽‍♂️ 👏

@@autodesigner Just checked and good guess. Currently at 3.6 but imagine that could get a bit higher with winter months to come.

@@TimGrose That makes todays 23:38 feel a whole lot better 😆

And I am local to borth Wyre Forest (544 at 3.4) and Sandwell Valley (554 at 3.5) and Wyre is sooooo much harder and has way more elevation. There is a very fast runner that comes along and smashes out 15minute runs there, which is probably skewing the data. After him, you can normally find 1st place at 19:30 times

I have done it once and a lot must be due to the downhill opening section you never come back up. I was at least as fast there as on a much flatter one at the time. No one runner should effect the stats overly much although what the best runners do is often a good guide to the general speed potential.