The statistics on how to win cat enforcement events

my perspective, as someone who has never considered zwift racing a sport but enjoys it as an interesting way to train

my non statistician, experienced amateur athlete brain is telling me that even if you could convince people who want to get good at zwift racing to race by being seeded against similar competition in every race, with no power ceiling, they are going to be disappointed when they find out their rank will remain broadly wherever they started at. they most likely won’t get better at zwift racing, they most likely won’t get fitter, and they certainly won’t have the opportunity to become anaerobically stronger

it’s not like dota, or chess, or cod, or any other ranked online game you can think of… its physically demanding in every way that a sport can be and unlike offline physical sport, which has a financial cost, travel and other logistical barriers, you can race as much as you like on zwift. just join an event at a time that suits you, maybe look for a course that has a profile that suits efforts you’ve programmed into your training, and line up.

so how do you get good at zwift racing? well, you race every day, or most of them. because if you aren’t doing that, a lot of people are. i race most days on zwift for about 5 months of the year. but at the same time, you can’t ride flat out every day. there’s a time and a place for that kind of riding, but it’s maybe once, maybe twice a month otherwise you’re liable to catch rhabdo, OTS, some kind of eating or sleeping disorder, or an acute stress injury

this is basically how i see a zwift ranking system without power ceilings playing out:

the best ranked guys can race daily, doing whatever ground work out of what for them will be z1 pace, messing around, having fun or testing stuff out or whatever, all of which is crucially important when it comes to getting actually good at something and enjoying the process. everyone else has to ration their race time, and the gap between the skilled and the unskilled gets wider, and the frequency of overall zwift race participation among amateurs is lowered massively

6 Likes

Agree. Most online game revolves around the same skillset from match to match. There is one skillset going into the ranking points.
In Zwift you will be challanged on different “skills” in the different races based on a number of factors. The ranking points you would get are not based on the exact same skills as in another race, but they will end up in the same ranking.

1 Like

Ouch! Touché! You just exposed my weak spot. I’m a disgruntled cat D, former cat C racer who sucks so much due to post-covid that I can’t win. And now I want to blame it on something else… I dunno… the System! Yes! It’s Zwift’s fault that I don’t win. I want to win and win and win again! Why don’t I? It’s not fair! Boohoo!

Wouldn’t that have been neat? No, I’ll tell you my real weak spot. It’s allergies. I have several. Gives me a bit of asthma too, as if the post-covid wasn’t enough. Let me see, there’s the hazel pollen, birch pollen, grass too. And then there’s injustice pollen, inequality pollen… oh, and stupidity pollen too. Well, I normally don’t have a problem with stupidity pollen. Only at certain times when the wind blows in a certain direction as to give stupidity free reign over reason or, even worse, over facts and science. Then I always get really sneezy.

Already long ago I got so sick and tired of seeing the same forum messages over and over from new zwifters and new racers. People who couldn’t understand why someone not in their cat was still racing in their cat. It didn’t make sense. People who asked for help to understand why they were never near a podium even though they had worked on their fitness so hard to rise to the assumed top of their category, only to be told off by wannabe cycling coach know-it-alls that their results weren’t surprising and that they needed to work more on their 2 min or their 5 min or their whatever, as if that was ever the problem.

And just because you are unable to empathize with all these people doesn’t mean others can’t.

My profile never was a secret. (The deliberate UPG’s are from when I practiced cruising while waiting for the downgrade so I could start for real for a bit.)

What exactly do you mean? That if I would run the same test on cat A as above, then I would get same result as I did in the past, right? That there is no weight advantage in cat A since it lacks this artificial performance ceiling? (Which is why I never bothered to as I simply assumed that’s what would happen.)

So the point then with the test above would be to show that even after they introduced the new model NOTHING SUBSTANTIAL HAS HAPPENED. It’s the same crap as before.

Jeez, I really didn’t have time for this but… just in case anyone is in doubt…

THE SAME TEST AS IN THE ORIGINAL POST BUT ON CAT A INSTEAD OF CAT C

INTRODUCTION
So what would be the assumption here. Well, we’re testing the same null hypothesis of course. There should be no statistically significant difference in results based on weight. Weight shouldn’t matter as much as to predict race results. And I showed that in cat C that’s not true. Weight does matter.

But what about cat A then? If cat A shows a similar weight advantage, then we can argue all day long whether this is right or not but we can’t blame it on these performance ceilings that I complain about. Because that is the most important difference between cat A and the other categories, the lack of this performance ceiling. And theory/simple maths says it should matter. So does it?

METHOD
Same method. A slightly smaller sample due to lack of cat A entrants in cat enforced races. But there’s a fair few, more than enough to get significant results from, in either direction. It’s 597 cat A racers.

RESULTS

Weight - Podiums vs All Losers
Avg: 71.2 kg vs 70.8 kg
Median: 70.6 kg vs 71.0 kg
Skewness: Potentially symmetrical/bell shaped (the cat C test was severely skewed)
Significance: p = 0.7574 (NOT statistically significant, null hypothesis NOT rejected)

Takeaway: SURPRISE SURPRISE! There is no significant difference in weight between winners and losers in cat A. And this is because there is no artifical performance ceiling that thwarts some racers and can be exploited by other racers. Just like two years ago. NOTHING HAS CHANGED WITH THE NEW MODEL.

20 min W/kg - Podiums vs Losers
Avg: 4.1 W/kg vs 4.1 W/kg
Median: 4.1 vs 4.1
Significance: p = 0.6956 (not significant)

Takeway: 20 min W/kg is apparently not what decides a winner in cat A. Still this measure is arguably the most important one even in the new model. To what effect? Nothing!

Watt - Podiums vs All Losers
Avg: 325W vs 303W
Median: 318W vs 304W
Significance: p = 0.00001 (significant, null hypothesis rejected)

Takeaway: So raw Watts, that they downplayed in the final model, would apparently matter somewhat in predicting results. But how can that be when neither weight nor W/kg does? Well, it’s because of the next result…

1 min W/kg - Podiums vs All Losers
Avg: 7.5 W/kg vs 6.6 W/kg
Median: 7.4 W/kg vs 6.4 W/kg
Significance: p = 3.405e-12 (highly significant, null hypothesis rejected)

Takeaway: No one is surprised. To win at Zwift in cat A you need to have the burst. It has nothing to do with weight or with your 1 hr or 20 min functional threshold.

Avg-to-Max HR - Podiums vs All Losers
Avg: 0.84 vs 0.87
Median: 0.85 vs 0.87
Significance: p = 0.0000056 (highly significant, null hypothesis rejected)

Takeaway: The difference is smaller than in cat C but it’s there and it’s highly significant. But here we need to understand the difference in a different way. The strongest rider in the cat A field in a race doesn’t have to work as hard to stay with the front group as someone “further down” the starting field. So it’s only natural that we see this difference.

But doesn’t the same argument hold for cat C with regards to avg-to-max HR? No, and you need to understand why. When you race right under an artificial ceiling, the top dogs in the cat, there is the assumption that only those who can barely reach that ceiling should be in that cat. If they could go any harder they would be pushed to the next cat and that’s where they rightfully belong.

So when we see podiums in cat C where they didn’t work quite as hard as the rest of the field and if the race ran right beneath the ceiling (it usually does in the lower cats because it’s how you win when there’s a ceiling and you’re not best in show because those are in the next cat - you push and drop people), then those guys on the podium could theoretically have worked themselves a little harder. But then they would no longer be qualified for cat C. They would get moved to cat B. And that makes their effort at least border territory of cruising. If they could go harder than the ceiling, then why aren’t they in cat B? Something isn’t working here.

And it never will. So give up on these stupid power measures, people…

But it really doesn’t matter what evidence gets thrown in your face, does it? Facts will never convince you anyway. Only authority would. Someone to look up to and take all their views from. No need for thinking for yourselves, just agree with them. Someone like Flint or Eric Schlange maybe?

4 Likes

Nice stats, thanks Andreas.

1 Like

very debateable though. it’s not really possible to know how hard someone worked, especially if the race ends with an effort. example

that’s 15 entire minutes of recovery from a thirty second max sprint out of z1, 15 minutes you wouldn’t get to see in the metrics of a zwift race, because the event ends when you cross the finish line

2 Likes

Amen to that.

6 Likes

Glad to find someone who took as long as I did to finish.

4 Likes

:joy: :joy: :joy:

3 Likes

Some thoughts after nearly two years at the bottom of Cat A:

  • There’s not enough incentive to change the system. Or, there’s incentive to keep it the way it is. I’m starting to think the choice to utilize and maintain a W/Kg system is intentional to make Zwift racing more accessible to a broader audience somehow. Maybe heavier people since it’s a training/fitness platform? IDK.

  • The term “race” or “racing” doesn’t appear very often in the forums relative to other topics, thus it’s not a priority. (yes, some racing topics are contained within the General Discussion topic)

  • Because I can’t compete within my assign Cat, I’d like to see how I fared within my age group. But that column is broken. That seems like an easy fix.

  • Would you be willing to pay more for a “better” racing system? Maybe it runs parallel to the W/Kg categories so you can choose which one best suits your interests. The weight-based system has to be retained (not simply replaced) based on my first bullet point.

  • Simplest fix for me would be the ability/choice to turn off the Cats, i.e. make every race a Cat E race. That’d be an end-user choice, not a hard parameter. A user could leave it as-is if preferred.

  • The existing W/Kg cats could still be tweaked. Why is there a black A cat but no black B/C/D? Why not expand black A, split red A into black A and B, etc.?

The belittling tactics isn’t working on me. I’m immune. So you might as well drop it. Are you racing in A or something? In that case you don’t even know what I’m talking about. What you fail to understand is that there are many guys out there who did everything right and still they are nowhere near a podium. It’s my standard reply amidst all the stupid wanna-be coach replies over at FB after looking at his HR histogram and his competitors’, and their weights.

“Tthe subleties of the drafting algorithm…”. LOL You know, I’ve been a gamer all my life. Zwift’s drafting isn’t complex. It isn’t hard to get at all. Kids playing the standard online games of today typically have to deal with far more complex game mechanics. A monkey can learn Zwift drafting. Pack racing outdoors is a vastly different thing. That’s hard for real. Zwift isn’t. When people get dropped in Zwift it’s rarely because they’re clueless as to drafting. They’re being outgunned. Sometimes legitimately so, it’s in the game. Sometimes less so.

I don’t accuse everyone at the upper of categories of cruising. I never have. But I do know they exist. You’d have to be blind to miss that.

You’re wrong here. The weight difference in lower categories is a direct consequence of the ceilings. And hence, as predicted, it doesn’t show up in cat A. Go back to my earlier posts. Reread them. Think a minute or two about the very simple arithmetics involved. And you’ll see that the heavy weight advantage is not only a direct consequence but also a necessary consequence of W/kg cats. It couldn’t be any other way. That’s why all my tests always show them. And that’s also why the power metrics has to go. It’s crap.

No no no no… The sample sizes are about the same, not a big difference. Yet we spot a 4 kg average difference in cat A that is significant and we don’t in cat A. There’s no way you can talk yourself out of that.

Of course they are informative! They confirm the obvious “theory”, the necessary arithmetics predicting a heavy weight advantage. Punchers win cat A. No one is surprised. But you obviously haven’t raced cat C-D. We are not talking who get first across the finish line and who comes second in a sprint. Who even gets the chance to participate in a final sprint is decided much earlier in the race. And there it is not uncommon that heavies and people with questionable overcapacity (as per cat definitions/philosophy) drop would-be contestants. You’d have to see it with your own eyes. I have. From all ends. As a loser. As a legit podium taker. As an illegitimate podium taker. And this is also why I have always encouraged others to try cruising. It’s an eye-opener, trust me. You see the lower cat races in a different light after that experience.

You’re right. That would be a better test. And to be honest, that measure is crap because it is so very crude. It’s just so accessible. A much better but also much more time consuming way is to consider the HR histograms over at ZHQ. It’s not the test you suggested but it’s something at least (it’s old and on the old categories). Looking at the histograms it is also much easier to separate good and perfectly legit sprinters or even punchers from the cruisers.

This reasoning is why I suspect you race in cat A. Do you? You would think it would be the same in cat C as in cat A. And in cat A it is indeed quite natural that winners show a somewhat lower effort than the runners up. But this really shouldn’t be happening in cat C. Let’s go through it again.

As a top cat A you are the cream of the crop. Few if any are better than you. As a top cat C, you are not best in show and neither are your competitors. There are better guys out there, only they race in A-B. And there is no lack of competition at your exact level within the field either. This will tend to drive the effort in races up towards the ceiling. Not in every race but as a general tendency. Some top C’s don’t have a good sprint and their only chance to win is to wear people down well before the finish. Hence efforts tend to approach 3.2 W/kg. It cat A the avg WKG may vary more and there is also no ceiling cutting efforts off.

Now, if someone wins a cat C race than ran just under 3.2 W/kg and shows a histogram that is mostly yellow whereas the rest of the podium was all orange, then we know for a fact (unless he has some weird heart condition and there are only so many) that he could have gone harder. He didn’t need to in this race, but he could have. His fitness would allow it. Now imagine a race where he did have to go flat out because the competition was so stiff. Then he would break the ceiling and move up to cat B. So should this guy even race in cat C? That can’t be the intention of the WKG cats, because if it was then you’re effectively saying “in order to reach the top here you should cruise”.

And this is a huge problem with power based cats, one that can never go away. They become contradictory.

That’s another huge problem. And they go hand in hand obviously.

And there comes the belittling again, usually a compensation for weak arguments. You know, it’s not rocket science to win a race in Zwift if you happen to have a fitness (because technique/tactics really isn’t very hard in this game) that just so happens to be in the top end of your current cat. Unless of course you are somehow not allowed to win because you are too light or something.

And the cat limits could be anywhere, both in a performance based system and a results based one. Either you are at the top end or you’re not. One day you could be on top. Divide the cats a little different, and the next day have suddenly turned into a loser who gets told he needs to Learn2Play, like the CS:GO would say (although in cycling lingo, just like you did). Most guys on the WT scene could never win a WT race either even with the whole team working for them. There is that span in any cat. And personally, believe it if you will, I don’t give a sh** whether I win myself or not. But I want the races to be fair. In a fair race I can heartily congratulate the winner and be impressed by his performance and fitness. In Zwift, as it stands today, I often can’t because the cat system ruins everything.

I think it would be a shame to drop the categories and go all E. It works so well in other sports in order to develop new talents (I will never be one, too old, don’t care to - get this: I’M NOT DOING THIS FOR ME!). But if I was given the only choice between straight E and keeping the current cat system, then that’s an easy one still. Neither I nor anyone else in B-D would ever win races and personally I don’t need that driver. But the races would be fair.

4 Likes

You get a pretty decent picture when looking at HR histograms. It’s not perfect either but it can often be quite revealing. Such as in a non-A race right under the performance ceiling where one guy who can’t quite manage to win looks like this:

And the winner looks like this:

Remember, the podium was right under the ceiling (4.0 W/kg in this case). The cat definition of B says you should at most be able to put out just below 4 W/kg. Just like the first guy. The second guy, I argue, would do just fine in the bottom of cat A.

The first guy was one of many many people asking in some forum for help on how to improve their racing. And as always they are told a lot of belittling crap that they need to work on their 2 min or 6 min or that they need to learn how to draft (lol? only an idiot doesn’t know how to).

The big question is what the worst problem is: Is it the cat system or all the yes-men protecting it just… because?

3 Likes

Like this post if you didn’t read that.

20 Likes

Wouter
Can’t say, like V.C., that I got anywhere near page 60 but I really did enjoy your Acknowledgments section.

Would be great to think we could all pull together to encourage and help Zwift develop the very best results based ranking system. I think they are listening, but then again I have only been here for a couple of months so possibly still glass half full.

3 Likes

People “cruise” or sandbag by avoiding 20min w/kg efforts that will put them at/above the next category up trigger. So it’s not weight specifically that makes sandbagging/cruising riders win, it’s their power (which doesn’t need to be held back inside of 20min) regardless of weight. There are any number of A racers with race ranks of 250 or better weighing 65kg or less who if sandbagging in B, will beat most if not all 76kg or heavier B sandbaggers/cruisers because they’re more powerful (and maybe aero I guess) for the bursts that matter. Just because the stats can be lined up a certain way doesn’t mean there is a primary causal relationship between weight and winning.

But definitely, why on earth not have a results based system triggering mandatory upgrades to next category up? It makes no sense not having it.

2 Likes

Personally, I think the category system is fine. I was a Cat D (94kg) racer a few months back. Once I got a string of podiums I started racing Cat C. Naturally in C Cat I finish very close to last. But I am learning. And I even managed a Podium Cat C. Now I am closer to 90 kg, getting faster in Zwift and outside on the trails.
My son is Cat B, 60kg. He has even bagged a podium in one of the recent Category enforcement races, and got 4th in a second. Both with fields of 30+ people. His average watts was lowest of all the riders except 3.
We do well because we train hard and go all out.

4 Likes

Congratulations! You have reached an all-time low and have been rewarded with access to the Call of Duty forums. Keep up the good work!

(Seriously, what are you even doing here, in a thread that I started, if you have no interest in actually reading the posts in it? To try make sure nobody else does, bully style? It’s not going to work, you know.)

Your overall point has been discussed and refuted, but you don’t acknowledge any reasonable or rational retort so instead of a good debate, you get a good point from a forum member followed by war and peace that no-one has the inclination to digest, so the conversation completely breaks down.

If there is some value in what you are saying, it is so well hidden in the weeds (there’s a lot of weeds) you should take a step back and have a think about what it is you are trying to say, and what is a more effective way to present so that you can be taken more seriously. If that is not your objective, why post at all?

6 Likes

Here’s the recipe to mix and stir a half-full glass of something sweet:

It’s not going to go smooth with all the yes-men who habitually counter-attack any critique of the performance based cats (the critique is necessary to pave the way for something results based). Zwift will just take it as massive support and justification for power metrics and nothing will happen.

It took two years for something to happen. Right before Christmas there was an interesting development in the forums. With most people you can’t argue with Reason. A much stronger card is Repetition. So at some point, while I was playing the Repetition card in the forum, suddenly people were like “dude, yesterday called and wanted you back, we are already well aware of this and now we’re discussing that”. Success! Finally… Coincidentally, it was around this point Zwift announced the release of a little something to improve racing…

But as I said, it took two years. Two years ago few, if any, were willing to believe there were more subtle ways to cheat or gain unfair advantages in races than sandbagging. Two years ago most people were still convinced light riders had an advantage [sic!], quite possibly an unfair one, in Zwift’s lower cats. Today that’s no longer true. Well, they still refuse to accept that heavy riders have the advantage instead, because they can’t be arsed to think the implications of the W/kg model through. Plus there is their reflex negative response to anything pointing towards possible change (which makes them feel uncomfortable). Still, though, there has been progress.

Right now there is a back-lash since the introduction of the new model. People now believe in power metrics again. They take the validity of the model for granted since it’s coming from Zwift and since it is still so new and underused that they don’t really have a feel for it yet, for whether it is actually working as intended or not. It could also be added that the model itself is, among other things, an attempt from Zwift to address some issues I have brought up again and again. Sure, the model doesn’t work because it can’t by definition, but still. It’s a sign of progress.

Zwift would never adopt a results based model you come up with. They can’t. No company could. It would be PR suicide. Still, it can be worthwhile to show them it can be done. It will mean something. But you’d need to be pretty detailed to be convincing, which is something of a catch 22 because at the same time the details will obstruct your communication since people prefer one-liners, a simple message. Still, you need a certain level of detail. No sketches. Present a full solution.

And you also need to be extremely patient. Create the model, then rehash it and refer to it over and over for the next two(?) years. It will slowly seep into people’s minds, even Zwift’s (they won’t actually read it early on, only much later). Once people you don’t normally interact with start referring to it, that’s when you know are beginning to apply pressure.

Also be wary about the role you take. Addressing Zwift nicely doesn’t work. They will appreciate the tone but dismiss/disregard you, surrounded by the yes-men. So, unfortunately, you need battering rams to break the gate first. But if you are that battering ram yourself, then you can’t be on the model’s credit list. It will reflect negatively on the model, the last thing you need. Thus you must not be in the first line of the assault on W/kg cats. You need to be in the 2nd or, ideally, the 3rd line. Actually, the best thing you could do would be to bond up with authority. People love authority. Have someone with authority endorse the model (bribe them or something!) or, better yet, make it look like they invented it. Ghost write the whole thing. That will have better impact.

Check, check, check… all in all, it’s looking good so far. So go ahead, start working! (Stop talking and make it happen.)

1 Like

Yep, that’s exactly what you should do. Start with the build me up training program.

2 Likes

@Andreas_Traff Think of this like a power point presentation - No more than 3 points per post.

This is a visual discussion, not a lecture.

You can be smartest man in the room, but if no one listens to you what good is it. No one is listening to you at this point. Believe it or not, people are trying to help up there.

5 Likes