Problems with the Current Category System
The Zwift racing category system groups racers based on their FTP – or more specifically, 95% of their 20-minute power, converted to watts per kilo (w/kg). The idea is that by joining the appropriate category based on this metric, you are racing riders of a similar ability. This leads to a number of issues which are the source of many community debates and complaints:
- Even if the category system is strictly enforced, arbitrarily grouping riders based on any sort of power profile is open to sandbagging. This is when riders deliberately perform below their actual ability, so that they can compete at the front end of races. For example, as an A rider by FTP, I can manage my power in a race to stay below the 4 w/kg limit of the B cat, but know that I will be strong at the finish and maybe win.
- The human desire to have the chance to win, or podium, or at least compete at the front of the race, means that you are most rewarded when you are at the top of a category. This means that once you reach the top of a category the incentive is to stay there, not move up.
- Riders with certain physiological attributes are able to outperform other riders in their category without being upgraded. Typically this applies to the extremities of weight. Heavy riders may be able to average 350w-400w in a B cat race but be within limits. Unless the race has a significant climb, this is a difficult pace for the majority of riders to stick with. On the flip side, super light riders that average less than the category’s pure watt limit (250w for B) may average 5 w/kg. As long as they can stick in the draft they will perform well on a hilly course, and if there are big climbs they will dominate.
For Zwift racing to work, for it to really become an attractive and rewarding proposition for all, riders need to feel motivated to do the best they can in any race. For progression to be attractive, not avoided. All while meeting the human desire to have the chance to win, or podium, or at least be involved in the shakeup.
ZwiftPower Ranking System
Most popular online multiplayer games uses a matchmaking system – the idea being that you compete against those that are a similar standard as yourself according to your ranking. As you perform well relevant to your peers, you progress up the rankings to face tougher opponents.
ZwiftPower has a rankings system that works on this basis. If you would like the full details of this system and the mathematical model behind it, it is available here (click FAQs). I will try to summarize it here:
- Your top 5 races in the last 90 days determine you rank, between 0-600. Lower is better
- Your ranking is improved by beating riders that have a stronger ranking than you
- 3 key calculations take place:
- Race Quality: how strong is the field (current rankings of participants)?
- Points Per Place: based on the size of the field and race quality
- Rank Points: determines your individual rank score based on the above two calculations
Now that Zwift own and manage ZwiftPower, it makes sense that this ranking system is used as a starting point. Everyone on ZwiftPower already has a ranking, even if you did not realize!
Zwift Matchmaking: A Simple Example
So how could a Zwift matchmaking system work, using the rankings described above?
When a race is created by an organizer, they can specify a field size. Let’s say they choose 50 as the field size. Field size can be chosen based on course profile, race popularity, or a desire to mix up the racing dynamic.
When you join a rankings race, there is only one sign up link. You simply join the race.
240 people join the race. At a specified deadline, the pen calculation takes place and you are placed in one of 5 starting pens (240/50). The pens are split based on rankings – so the top pen has the top 5th of riders based on rankings, the 2nd pen has the 2nd 5th of riders, etc.
The End of Sandbagging
In a matchmaking system, your performance relative to the riders around you updates your rankings, with the goal to move up the rankings. As performance is based on final placing, there is no incentive whatsoever to do badly. Sandbaggers would have to perform consistently poorly to race a lower level than they should, but as soon as they win races they will be upgraded! The end of sandbagging as we know it.
Some days, you may be in a field where you are the strongest rider and have a good chance at winning the race. Other days you may be the lowest rank and get dropped early on. However, doing your best in either of these scenarios is rewarding in terms of ranking points.
Onboarding Newcomers
Managing newcomers is one aspect that needs some attention, otherwise weaker riders (current cat D) might continually face newcomers who have not yet worked their way up the rankings. The way other games manage this is to have new racers complete 3-5 races “open” races before entering the formal cat system.
A Proven System
Rankings-based matchmaking in this style is the crux of why games like Call of Duty or FIFA are so successful. Improvement becomes additive. Sprinting from the 3rd group can pay off if the field is a really strong one. It may sound quite complex, but it is really not. ZwiftPower already does half of it, it just needs a matchmaking system baked into the game.
Will we see this in the future? Is this one of the ‘Systems’ that Eric Min has previously alluded to? I certainly hope so and look forward to reading your comments.
Your Thoughts
I’ve posted a Matchmaking topic on Zwift’s Feature Requests forum – please share your thoughts there and upvote the idea if you support it. Also, feel free to share comments below!
This assumes an even distribution of abilities which is frequently not the case. If distribution of abilities is uneven but categories are of equal size, the spread of some categories will be larger than others, leading to an uncompetitive race for many participants.
Uncompetitive in terms of a win – as really, most races would always be – but competitive at all levels with the riders around you, however many there are. Either way, far superior to the current system!
A good point, though changing the algorithm to help account for this would be fairly trivial compared to the rest of the work required to get this implemented.
I would hope this disparity would encourage those riders to train those deficiencies. I know I’ll be working on 1 and 5 minute power which will theoretically lead to more success in game.
Exactly what I thought. With the cat distribution in the races I currently race (in zwift, not in zwiftpower; higher cat riders more often have a zwiftpower account) I am probably not at the end of B but at the end of current A and B. No matter how hard I work I never will beat the As, I have some hope for the Bs. I think if this happens I will end my racing career.
Well, with so many people racing, wouldn’t the distribution be very narrow? Say I’m level 500 I’d be competing with 50 riders that will most likely be between level 499 and level 501. Curious to hear everyone’s thoughts!
You can’t accommodate everything in the ranking system. IRL rating systems that are based on results have exactly the same issues. People are better at some types of racing (short versus long versus very long, flat versus steep versus rolling, sprinting versus breakaway.)
We also see the solution IRL in that people will decide what races they want to race in based on what they are good at. If you are a climber you don’t expect to do well in long flat races. Sprinters don’t expect to do well in long climbs.
The point is you end up being ranked based on your best events, which for most people tend to be the events they can do best in. And they don’t expect to do well in the events they cannot do well in.
Yeah this is probably true, in particular on the extremes of the distribution (top and bottom categories).
But because ability does not correlate perfectly with ranking, because of the dynamics of the point system, this might not be the biggest issue. Say your ranking points put you in the bottom of the A cat. You will typically get awarded a relatively high number of points (ie… low ranking) when you consistently get dropped, this will reduce your ranking, and after a while, you will get allocated to the B-cat because of it. So, edge riders will be switching back and forth between categories, being very competitive when their ranking drops and they are more likely to be assigned to the B cat, and vice versa when their ranking improves again because of that.
I think this system will be great… it would solve so much of the current issues!
Keep in mind a few things. 1) The proposal is for the categories to be allocated on a per event basis. So two different races the same day could see you racing in two different categories. 2) you won’t drop down in rankings until your oldest top-5 result hits 3 months old. 3) If you finish next to last in a race where you had the lowest ranking, you could still improve your ranking. So if improving your ranking is a goal, even sprinting to stay out of last place may be worthwhile.
Yes this is correct. To avoid confusion it is useful to call them ranking pens instead cats I think.
Simply genius.
makes sense. Sounds familiar, think USA Cycling ranking system. Not perfect, but not bad.
Awesome idea! I like it other than the fact people like to be categorized and in the lower levels you get lots of casual racers who don’t really pay any attention to Zwift when they’re not using it. I feel like sandbagging is starting to fade although Zwift has yet to start pulling obvious sandbaggers from races (last Monday a guy soloed off the front from the gun, averaged 4.1 W/kg in a D race without even a cone of shame.) When you finish a C or D race over half aren’t on Zwift Power so my first recommendation is for Zwift to start talking about Zwift Power and more about racing guidelines in general so the more passive Zwift users are aware of features. Oh and some people (Eric) like to research the competition before the race! 😉
Also apparently only a small proportion of Zwifters are signed up to ZwiftPower, so Zwift would need to properly integrate that as part of the app so sign-up occurs as part of joining Zwift.
Absolutely agree with this system and support it 100%
This works well for individual races. But race series also solve the problem using the existing A, B, C, D system, provided the series includes races that cover the gamut of individual abilities. Imagine a race series that includes flat courses, rolling courses, long climbs (e.g. Epic KOM or longer), and short climbs, spread over multiple days. This levels the playing field according to the W/KG measurement. Sure, the heavier riders will often dominate the flats; and lighter riders will dominate the long climbs. But good all-arounders should do well on rolling courses and courses with shorter climbs. Then, if they can limit losses on the flats and long hills, their combined result should be good. Moreover, a really good heavier rider might do well enough on the flats, rollers, and short climbs to offset expected poor results on longer climbs. Similar with lighter riders for the opposite scenario. Toss in a few ITTs on the same sorts of courses, and your result is exactly what real stage races do to level the playing field for riders of different talents at the professional level.
There are already a few race series that do this, e.g. Herd Beginner Race Series. Moreover, this sort of system can complement a matchmaker system for one-off races. Both can exist together.
Yeah agreed. Race series and team events can be organized in other ways.
The good thing about race series is you don’t have to wait on Zwift to do something. They’ve not done much recently. Waiting for them to implement a desired feature seems like wishful thinking.
I would love to see that happening, but based on previous Zwift releases, it might be not before 2025.
I started doing the math and based on weight and what I think will be an increased FTP when I re-test at the months end I could test out as a class B rider. That would be comical for me to be in a class B race I would be dropped before getting out of the start area. I enjoy racing and typically finish in the bottom 3rd of class C I have the most fun when I am have others to compete against but often times find myself in no mans land 40 seconds behind some riders and 40+seconds ahead of others. I am all for trying the proposed changes.
Thanks
Same boat – I’ve been bumped to B, but I’ve pretty rarely been able to cling to the front group even on the C races I’ve done. The prospect of immediately being dropped in the B races, despite putting out 100-200 wats over FTP for several minutes at the start, doesn’t encourage me to race very much.
200 watts over FTP for several minutes?!?!?! If you can do that you need to re-test!
I totally feel your pain. This happened to me too in real life. Then I got a coach and read a lot. Thankfully there’s a cure.
The likely reason you’re getting dropped is that you have not targeted your 3-5 min power with appropriate interval training.
What’s going on is this: When you are asked to ride above FTP (like at the start of a Zwift race) you’re entering VO2Max land. This is the zone above FTP but below anaerobic. It’s the most challenging area for most of us because it relies solely on how much power you can make from your maximal oxygen uptake.
Generally, riders can only ride in this zone for up to about 8 minutes. Secondly, repeated visits to this zone are what determine selections. The riders that can repeat quality efforts in this zone will shed other riders that can’t. So you need to train your body to repeatedly work in this zone, then rest and then be able to do it again.
This is extremely important: When you see stats like someone has a 5 min power of X watts, that’s just for a once-off effort. Races are generally determined by multiple visits to this zone, so fatigue resistance is just as important as raw power.
There are two ways to improve your VO2Max power and fatigue resistance.
If you choose #2, you will see results but it will be painful. By interval #4, you will start to forget the number you are on! But if you finish and do it twice a week, these will transform you and give you so much more confidence.
There are other intervals as well like ‘over/under’, Tabata and longer VO2Max efforts. But starting with 7 x 3 is a great step up.
Really good idea… And I think with the ZP data if you go over your Cat Threshold you should be popped into the next category up… I’ve had a few Cat C races where I’ve killed myself for 20 minutes to try and stay on the lead and been DQ’d for being over the w/kg threshold. Even though on the overall races I’ve averaged less than my http://FTP...
This makes perfectly good sense. Hope Zwift is listening!
I like it and have been saying this for over a year since I’ve been on zwift – race to the rankings, not the performance metrics…. so many riders do their 20 min power for over an hour, because they know not push too hard.
End of team racing. But good with automatic leveling
Couldn’t Team Racing be some kind of average of the team’s individuals with this system? (realizing getting into some complexity, but not too bad).
It’d be more work, but you could potentially identify a person you’d like to race with and you’ll move up to their category automatically.
Nah, it wouldn’t be the end of team racing. You’d just have to account for that in the system, or have races that are organized differently than the ZP system (which already happens.)
I’m a good example. Top of cat B, 4w/kg, and I enjoy every race. I can deliver a constant 360 watts, but I can’t go above 430w for more than 90 sec, if I move to cat A, it’s simply frustrating, at the first cut I’ll stay behind and not being able to catch the front, no matter what, I simply can’t stay with the peloton… frustrating…I’m a diesel engine…and yes I’m 65.
Same for me. Did some nice top 10 results in B. Now I’m upgraded to A with average 4.01 w/kg and are at the end of the field. Not much fun anymore. If I order the start list in Zwiftpower by ranking I can almost predict perfectly where I would end the race. I support Eric’s suggestion 100%! Futhermore, races aren’t won with FTP but with 30s to 2min power. With only good FTP you are either outsprinted at the finish or dropped at the short climbs.
I think it has to be “… arrange the start list …”
Yes, let’s change the rules and divide classes so finely everyone gets their participation trophy against minimal opponents
The intention is to find a system where you are rewarded for doing better. This is not the case at the moment. In some events they divide it into age groups of 5 year steps. At big events that could be OK if there are enough participants. If only 5 guys start in your group it it’s boring, too that’s right. With the current system it’s better for me to hold back and downgrade to B to have more fun. And that’s why we all do it, to have fun, don’t we?
No, No. Change the format so there are only 3 riders in a race. And everyone gets a real trophy. Problem solved.
Why do you reply if you are not interested in a real discussion, only telling people between the lines they are stupid? A discussion whether or not a existing system can be improved should always be done. If you have a different opinion you are welcome to explain it.
Can’t wait. I think the cat system is broken. I feel unmotivated to race unless it’s with people I know. My idea was to split the cats further. But this ranking version will be better. I look forward to being able to race against strangers and feel like I have a chance.
Categories A, B, C and D are absolute, and this is relative or positional approach. If I understand well, my position in ranking should be defined by percents of people above or below me. Before race the virtual ranking of all registered people in race would be created, and this continuous scale would be divided for instance in the same 5 groups (quintiles). I will compete with people only in my group, then. Yes? The start would be only one, but with different jerseys for individual groups, or each quintile group will have different start time?
My suggestion would be to have only your pen visible for the race (which is an existing option), but in theory it could work in all 3 existing formats.
James, I think this is a step forward with your proposed system. Let me explain a bit more. I competed as a Category 1 cyclist in the 90’s and early 2000’s in the Intermountain region of Colorado…and due to Covid, joined Zwift initially back in late 2019. I have aggressively raced on the platform rather than ride and race outside and I have seen the issues mentioned here first hand. I was rapidly going through the categories with nothing more than the ZwiftPower guidance..to which they recommended I join the A’s after respectfully competing in the C’s and B’s…yes, I adapted more quickly due to my competitive cycling background…but to your point…I got burned off repeatedly and quickly in the A races (mostly by A+ and very high A’s)…from this, ZwiftPower then adjusted me back to the B’s…where I have raced most often since. I sometimes think about the bigger question here…are we engaging the greater community of cyclists to ride more often and train to get better…we all as a community of cyclists need to keep this objective front and center. For me, my livelihood does not require that I “win races”…I would also argue for the majority of Zwift racers/riders..this is the case. It is more “engaging” on the platform when you are a relevant animator of the race…what I mean is that you are in the lead pack with a chance to impact the outcome of the race podium instead of getting jettisoned off the back right out of the pens. Last but not least, I think the ranking system captures the best essence…you will see when you race often on Zwift it is not just the racer that throws the biggest Watts, rather it is about tactically throwing the Watts where it counts on a given course against a given field. You can easily see this if you analyze the results on Zwift Power…notice there is often a difference in the W/Kg and average watts in the top finishers and it isn’t always the top down in the criteria mentioned. This means there is strategy within the race too…being ready to blow watts early to stay in the front group requires training the super anaerobic top end energy system with threshold based recovery follow up capabilities. Then, having knowledge of the course (eg The Bell Lap vs Downtown Dolphin on Crit City) can lead the racer to time the attack for the best result given his/her strengths. I also will close by saying we need to educate riders on the platform b/c most do not understand how the system works…we all benefit by encouraging the dialogue instead of pointing out misunderstandings and/or purposeful sandbagging.
Ranking based matchmaking is surely better than current extremely faulty system but I’d still very much prefer power curve / route type categorization. It allows to group people properly even if they didn’t race previously at all.
But if they didn’t race previously, where would the accurate power curve data come from?
The “knowledge is power” pop up is really annoying.
I would like to see this implemented as it seems that leaders in the C race often put out numbers that would upgrade them into the Bs. I agree with the article that there also needs to be some better tweaking for weight. I was beaten in a TT by someone who’s avg watts was 94 W with a 3.8 w/kg avg in the race. This means they weighed 25 kg or 55 lb. or lighter than my 8-year-old son whom I’m quite sure would not beat me in a flat course 20 km TT. Perhaps it was overlooked by ZwiftPower as it was far down the ranks, but it does affect the experience for those of us looking to improve and race our peers.
As a stand-alone “ranking system” I think it has inherent flaws just as the W/Kg system does currently, I’d imagine a system that combines the existing Cat A-D, and W/Kg, and also the Matchmaking approach but which also ‘hides’ the ZP Ranking from view would work best.
In terms of flaws of a ‘just matchmaking’ approach I can see the following:
1) Even assuming Zwift could update the app to have unlimited numbers of Start Pens for races, whether with or without the existing Cat A-D splits, it would significantly limit the size of the groups you race with and people flock to races with larger numbers and not smaller ones as it’s generally more fun the more other riders there are
2) For riders who actually want the challenge of riding against much stronger riders and really pushing themselves you can’t do it as you’d be forced into rides with people of a very similar ranking to yourself rather than the whole spectrum
3) Zwifters who don’t ride on it year-round would still effectively be able to ‘reset’ their ranking by just using a few races for recovery rides at the start of a season etc., especially if it isn’t done in tandem with our factors such as Categories and/or W/kg (even tweaked versions), even all-year riders could deliberately slack off for a while and effectively get ‘demoted’
3) For people who are ‘not’ sandbagging it doesn’t feel like winning to be put in say Group-4 in the pens and win when you actually finish maybe 151st in the times because the first 3 groups finished ahead of you and you weren’t starting and riding with them.
4) Zwifters even now can see the ZP Ranking and choose races accordingly meaning they can deliberately choose races with lower rankings so they improve less
I think what Zwift needs is an integrated system that combines the existing Cat A-D bandings along with Watts & W/KG (albeit a tweaked version that takes weight into account), in addition to making the Race and Rider rankings hidden so riders can’t attempt to manipulate their own ranking or the races they compete in, and integrating an ‘Anti-sandbagging’ approach to sign-up where riders can’t choose to sign into a lower Category than they are ranked as.
That way if you are ranked Cat-B you’d be required to sign into Cat-B Pen, but the race would rank your result based on the ranks of all riders and your final position and use that as part of the Category Classification alongside an amended W/Kg & Watts calculation (which is scaled for increasing/reducing weight) all as part of a weighted calculation of your Cat.
Thanks for some interesting discussion points.
Brilliant article.
Thanks Simon! Could go down a never ending rabbit hole discussing the finer points, but hopefully this is an overview of how simply this could be implemented, and the huge benefits it would bring to the platform.
All nice and pretty, ok but I get that Zwift is not going after 100% realism but some would be nice. For that it has to implement some type of lane system. The blob cannot work as a huge TTT effort. It is SOOOO!! unrealistic. It has to be harder to overcome other riders so one can stay in the draft. In zwift one gets pulled to the front even without wanting. It cannot happen.
I think it’s a great idea! Crossresults.com uses this formula. It creates a good competitive atmosphere at races and identifies who your “Nemeses” and “Victims” are. You get to know folks that are near your ability faster because you tend to keep and eye out for them.
yes crossresults.com is incredibly accurate, to the point that race predictor tool is almost the same as results in my experience. from what i’ve heard from the creator of the site, usacycling “borrowed without asking” the algorithm because it was so good. i would like to see this implemented in zwift. as the article states, trying to improve your ranking is good enough motivation in many cases, rather than sandbagging a lower cat and going for hollow podiums.
I can and do break the algorithm frequently IRL. The issue is that race results age. And I don’t race often. So I come into a cat 4/5 race IRL with points in the middle 500s. Completely overlooked. Because I’ve raced once in the last year and my other averaged (non-races) are 600’s. And I steal a podium. It’s not deliberate. It’s what my schedule and life allows. Sorry. And IRL it doesn’t matter because I’m racing my category and haven’t downgraded (and am still just short of what I need for upgrade). Zwift points can be gamed exactly the same. Allow my results (or some) to age beyond 90 days, average in only a few results which adds 600s for the missing results and the same thing will happen. There’s no perfect solution. IRL, USAC won’t allow you to downgrade. So I still think you need some sort of “you can’t fall back further than ‘xxx’ once you’ve reached a certain level. That might not be a “cat” per se’ but this will be an issue that will come up. I’ve heard the arguments about injury and stuff for racing down. Sorry, don’t buy it though it’s probably fair when you get into your late 60s 🙂 and a petition system could be added for that. Otherwise, If you were an ‘A’ once, are still under 55, you’ll get back there if you want it. Probably long before I ever scratch out another .25w/kg to my FTP. Otherwise though, points racing definitely does have promise.
I agree with these issues, but they are very easy to solve. I’ll cover them in a follow up post.
I do think this is the weakest part of the current zwiftpower ranking system and the proposal as is (to be clear I’m a huge fan of the proposal James!). I’m interested to hear what the solutions may be. Zwift racing is a little different from IRL because the time/money commitment for a race is so low. The real life and probably not uncommon example that Funkster gives may be rare in zwift.
i think most folks aren’t as concerned about people like you describe, because by definition they are racing relatively infrequently. it’s the known salty sandbaggers with 100+ races that won’t have anywhere to hide.
I live at 5,400 ft above sea level so I am at a 5-10% disadvantage to riders at sea level every time I line up to race. These changes are great but not sure why Zwift refuses to make a very easy normalization change to those at altitude. No, I won’t shut up about this.
Ya, and then for some strange reason, most riders are all from Mt Everest?
i don’t know, no one seems to have any sympathy for us flatlanders when we try and race at altitude…now the tables are turned!
All interesting thoughts but when will Zwift finally DO something about this? Anti-sandbagging testing was done quite some time ago…. hope they start implementing spon.
Please please please start this system of ranking immediately. As racing Zwift for the last year has defined what it is to “game the system”
That is to start extremely hard and then soft pedal after the attacks are over as to lower your watts per kilogram over the duration of the race.
As long as you can game the system successfully you will have Riders at a disadvantage due to their either lack of knowledge of gaming the system or unwillingness to game the system.
Longer more difficult races eliminate gaming the system but those are few and far between. The algorithms are so inaccurate that you can Sprint Coast your way through a criterium and have a watts per kilogram far far lower than your actual watts per kilogram.
I’ve been waiting for this feature for YEARS ! I don’t race anymore because I’m light and near the lower to middle of Cat C. I always get dropped and end up either alone or in a small autobus. Plus you can never tell at the start of the race which group will be DQ’ed and which will have the winner, so knowing who to try to stay with is impossible. If I knew I could be competitive with the other racers in my group I would definitely start racing again. MAKE IT SO !
I think you might be waiting for years to come. I don’t believe Zwift will even do this much.
An alternative (which can still be fun) is to find events with all riders visible and a wide range of abilities. You can still be racing people of similar speed, maybe they will be strong Ds but who cares, it’s still a good challenge so long as there are enough people around you to make it interesting. Of course the solution proposed is better still 🙂
Before you can solve a problem you have to understand why the problem exists and this solution misses the point. IMO riders sandbag not to win the race but to be competitive and stay as long as possible in the front group. This is currently simply not possible for a low end B competing in a B race – the category width is just too wide, 3.2w/kg – 4.0w/kg (+ sandbagging A’s). 0.8+ W/kg is too big a gap to close even with clever drafting. The B rider is quickly dropped and consigned to riding in the B grupetto. The solution therefore is to half the category widths and have C,C+,B,B+. The gap now between a bottom and a top B is now only 0.4w/kg which can be bridged with clever drafting. The motivation to sandbag goes away and the low end B rider now has an achievable goal – an improvement of 0.1 or 0.2 w/kg should see him consistently staying with the lead group. I am a top end C rider, 61yo, if upgraded to B I would never improve enough to be able to stay with the top B’s. I would always want to enter a race where I have a chance of being competitive however the above approach seems to suggest that in some races (50% maybe?) I’ll be uncompetitive and will get dropped early on. Also can I still ‘sandbag’ by pushing the pace for 99% of the race and sitting up in the last km and letting 50 riders go past thereby ending with a very low placing and few ranking points?
There is nothing stopping the above approach from having tightly banded categories. Bike racing is certainly about winning. But with a rankings approach you’d also be rewarded for finishing higher than you should. So there should be a competitive reason to sprint for 15th place.
This is why I like “open” races without categories. If you get a good mix, there is always someone to push yourself against, and, after the race is over, the fun is not looking at your overall positioning but, instead, looking to see if you hung with people you “shouldn’t have”.
And while we’re at it we might as well tune the ranking algorithm to account for time gaps. Riders who finish together in a group should get ranking scores with smaller gaps than the last person in that group and the 1st rider in the next group 1 minute further back. The USAC system can’t do this because not all races have times for each rider, Zwift does.
James – I have been trying to come up with a better system, and this is great. Thank you for taking the time to think about this and write this up. I love it. In theory this also gives you a better chance to improve your race ranking since you wouldn’t be racing a bunch of people with no rank.
Great idea!
Love the idea, and perhaps it could be augmented by quintile calculations based on course-type? I.e., weight being more important than points for the Alpe or Ventoux, or watts being more important on Tempus Fugit. Basically points + watts + w/kg (or whatever math, IDK: it’s late in the afternoon).
I also find it interesting that while so many commenters like the idea, there’s still a fundamental problem with the 1st mile of any Zwift race. I’d love to see some form of neutralization, maybe for an unpredictable percentage of the first kilometer so that a peloton can coalesce out of the pen.
The proposed matchmaking sounds like a good idea. I’m sure it would be possible for Zwift to run a Futureworks trial with a Comissaires car, Scott the squirrel o could even wave the flag out the sunroof. Itd basically roll out and ramp up the pace to 5% over the category (old) or mediated (proposed system) FTP, which would be survivable in the draft but temper the sting that breaks Zwift races up early for many riders.
Not sute how the ranking works with WTRL TTTs as rider placings are not captured in Zwiftpower, being dominated more-so by the roll-out delay each team had to wait for. But then, if Zwift got truly integrated with the likes of Zwiftpower, WTRL, Zwift hacks etc then it would be native and of done right, data is available to use.
Your top 5 races in the last 90 days determine you rank, between 0-600. Lower is better
A lot of effort for very little reward. One tweak I would make to the ZP model, is it should average your ranking points if there are less than 5 races in the period (e.g. 300/305/295 = 300, as opposed to 300/305/295/600/600 = 420) Also I would extend it to 6 months, and maybe weight the ranking of more recent races more heavily. However these details could be worked on at a later date, it doesn’t have to be perfect from the off or it would never get implemented.
If you only win a race here and there, it’s not a problem. Currently, the same group of riders can win every race, every day, without ever getting upgraded. That’s a problem.
Would be great if you only see those that are in your group. I don’t want to be distracted by other riders that are above me, and not sure if they are in the same group as me.
What happens if it’s a smaller event where there is a large points range between riders?
As long as points are only earned for race events (yes, I realize, get more than two cyclists together, and it becomes a “race”). Other “social” events that Zwift and other groups put on, shouldn’t necessarily count towards your ranking
Definitely only races not group rides. Race organiser could configure if you only see your pen or not, as today.
Great commentary here and I appreciate the continued mindshare. One final tidbit, in a recent conversation I had with Scott Barger, one of the co-founders of Zwift, he mentioned it was never the intention originally, nor the goal overall to have the Zwift platform mirror or correlate/translate to the outside riding environment…in other words, Zwift is a virtual platform made to encourage riding and fitness…with racing, group rides and many challenging courses that “simulate” riding dynamics outside…fast forward and yes, we have many subscribers that advocate and race on the platform and many that “game” or garner virtual racing achievements. It is important to keep this in mind as if this were not true…many Zwift virtual platform legends would be riding for money on professional contracts in outside sanctioned races. Please refer to the riding palmares of Rigo Uran…for those of you that don’t know him…you can see his professional results here > https://www.procyclingstats.com/rider/rigoberto-uran/2020 where he finished this year in the TDF in 8th. Please contrast that with his on Zwift platform performance recently here > https://www.zwiftpower.com/profile.php?z=681579…there is no way (if there wasn’t gamification of this Zwift platform) he would have ~7,000 racers on Zwift ranked better than him and an “almost A+” ZP ranking with W/Kg or overall Watts. Hence, I appreciate and understand the need to improve and to make it more “realistic to real life racing” but we are a long way from this reality…the proposed matchmaker mechanism would improve the outcome (also add Jeff Winkler’s time gaps algorithm finish metric on events as well from earlier posted comment here), but it is still a platform for fitness training at the end of the day.
Brilliant. This seems to make so much sense. Zwift to implement in 2030?
Great idea, is the Zwifpower ranking system still in use though? It says rankings are based on last 90 days, but then the last race it says I did was in February, so a lot longer than 90 days!
It doesn’t count group rides, only races. Link to your ZP profile and I can check it out.
Good point. I hardly do races within a 90 days period. Maybe one every 2 months. But I race in the A cat. Would such a system push me back to a lower category when I don’t race for a certain time? That would force me to sandbag when I race only now and then, or am I wrong?
Depends if we measure future in years or decades. But in the end there can only be a single winner and the person should be the strongest one which usually happens to be the person with high w/kg. It’s not that black and white because some just are great in sprints and short distances and others perform better in long rides especially beginners who are otherwise strong. Also even same person can become weaker after a while. But there are even bigger problems in the platform like zpower riders in A races closing gaps after shortlived breakaways and power or weight cheats. Different ranking system does not help those issues. Also ZwiftPower website is becoming slower than 90s internet.
Sometimes even more important than avg w/kg in races is to know when to push those watts because staying in the group’s draft is a must. Many might give up yoo easily during a hard start but it often lasts a few minutes before things calm down so knowing that and being strong mentally is almost as important. Also knowing your 1, 3, 5 min max power helps to keep nice avg power in hill efforts and to keep confidence high as well.
It’s interesting to watch Zwift racing to evolve from casual towards USA Cycling type classifications in order to make racing fair and competitive. In the future when they complain about the rating system being too confining, I hope they reach back to 2020. It’s just a natural evolution of any racing or competitive environment. No right or wrong here, just putting a system in place to make it the best possible experience for everyone.
This totally sounds like it would be better than Zwift’s current system. I’m sure there would still be some hitches but a similar system (that has proven itself worthy) is used in IRL racing all the time to determine a racers category. I definitely think that this would be something worth looking into for Zwift
This is great conversation and thanks for the brainstorming ideas. I want to reiterate my experience with Zwift thus far…there is no other platform from my experience which enables this amount of precise structure for working out the different physiological systems required to pedal with efficiency and power! If used properly, Zwift can provide training that the outside world with all it’s unstructured components (Stop signs/lights, down hills with coasting, safety issues with cars etc) cannot duplicate and we can do it anytime day/night. On top of that, you have the motivating elements of group rides and races to really keep the candle burning and compared to the long, dark roller training days in the basements of our past…this is hard to beat.
The best matchmaking system I’ve seen is on iRacing.com, for which many pages have been written.
Is there system documented anywhere? I can’t seem to find it on the website.
*Their 🤓
The cat system is completely broken. I’ve won 11 out of the last 12 B races I’ve done. I have been racing A’s mostly but my w/kg is 3.9ish ( I’m very short and heavy ish) so no need to be forced to A but when you look at results I should be forced to A. Sometime I just don’t feel like racing A because it’s really hard, so I’ll do a hard but but as hard B race and sandbag it. Racing skill has so much to do with winning races. Zwift should be upgrading people just like real races. You win you move up, especially if you are on a category edge. If this was usac I’d get a mandatory upgrade. It’s kind of nuts people can stay in B and C and just dominate. But that being said there are a lot of races to have a chance at doing well and you can see who is in them via rankings. I agree something needs to be done and I am part of the problem.
The thing that drives me mad in zwift races is when you enter a lower ranked race and 5 people on the front are on zpower and just massacre the field. It sucks for everyone. Generally three high ranked races this does not exist or we let that one guy go.
The point isn’t to give everyone a chance to win. You’re either good enough or you aren’t regardless of your weight, height, and the course profile. Trying to artificially level the playing field so that “everyone has a chance to win” is not real competition. Your heavy guy w/massive watts on the flat or little guy w/high w/kg on a hilly course is no different than real life. It is what it is! To try and alter that is wrong! Like real life, only some people realistically have a shot at winning.
The purpose of this system isn’t to give everyone the chance to win. Lightweight riders will still do well on hilly courses and heavy riders on the flat. The purpose is to group riders roughly by ability for races, but in a manner that gets rid of the ‘gaming’ of the system – notably sandbagging, and adds a level of dynamism that encourages riders to continuously improve. Potentially you could end up in a weak field and have the chance to win that you wouldn’t have with the cat system, but that is a benefit of the system rather than the purpose of it.
It would encourage me to enter races other than the team events. You can always count on a Brit for a good idea that will be implemented and capitalised on by the Americans 😉
I’d love to see someone take up this system and implement it similarly to what WTRL have done with the ZRL. They have refused to wait for Zwift to do something about it and they’ve taken it on them selves and created an incredible team racing series! Similar could be achieved with this I think. Ask riders to sign up in advance, scrape their rank and sort them 2 days before then send them a unique link to their category race.
Good idea in the interim. It would take a bit of manual work and still need Zwift buy-in, but it could definitely work.
In theory it sounds great, however we could see some fear of loosing rankingposition and thereby fewer starts in races, so maybe choose a system, where ranking was based on e.g. a riders top10 performance (if that could work).
I love cycling and races, but the main issue about motivating leisure riders to compete is that, compared to other sports, VERY few actually experience winning. So maybe Zwift could start to think about other ways of competing during the classic races, fx matching you with 5 other riders in a field of 50, that you are evenly matched with (e.g. based on ranking) who you get extra rankingpoint by beating, along with how you perform in the classic race.
You don’t lose ranking by performing badly in races. The ranking is determined by your best 5 performances in a 90 day period.
Let’s break this up in MVP’s, first make a mandatory or automatic selection of the A, B, C or D category for races. This will, even with the current 95% of 20 min power rule, solve a lot of sandbagging. FTP is already available in Zwift itself, so this should be really strait forward. Especially in combination with the roll out of the anti sandbagging functionality already developed (I think it was based upon 5 min power in race?) this would already be a giant leap forward. You can combine this with splitting up the A, B groups for bigger races, since the majority of us is probably ranging from C to low A.
Next step would be to implement some sort of ranking system, because this is indeed a better solution.
However, with the current speed of change at Zwift (did someone say user interface?), the first step will probably be implemented mid 2022 and the ranking system will probably be~2025.
👏🏼
I also think there should be more cats. I’m a C rider. Between the lowest (2,5) and highest (3,2) w/kg that is gap of about 60 watts(at 85kg) That’s huge..
But how simple would it be for zwift to just filter everyone above the set limit out of the rankings.. Very very simple first step. But more steps would be very welcome indeed