I don't know, I'm probably the first person to think of it.

I question your assessment here. It's possible you're just not that great at Zwift pack riding. My experience is though there can be some variation in certain situations due to height/weight, W/kg is still the dominating factor and you should be generally pretty darn close to those around you in W/kg if you're riding efficiently. Certainly not the huge jump you describe above. And Zwift pack riding is a legitimate skill for racing purposes. You can think you're good at it, but not be. As I learned from racing and seeing people who spent the entire race (with similar height/weight) needing a lot less W/kg than me just to "sit in." And don't think you're good at sitting in in Zwift because you're good at it IRL. Racing with @echappist on this forum was informative here. He recommends using a Zwift Pace Partner bot, learning to sit on the bot, and being able to stay on the bot with about 25% less W/kg. (about the drafting effect for being behind a single rider). Learning how to navigate turns (where the "drafting cone" causes effects that don't exist IRL) is important.

And ignore jaretj!!!

FYI the latest update seems to have messed with the algorithm. D grade was raced and won at 3.5w/kg, no coning, and yes it was a ZHC anti-sandbagging event. I just hope they fix it soon. Would be a shame to lose those getting into racing this way.

Fair point. They have seemingly been messing with physics and pack dynamics quite a bit lately. Usually in a positive direction, but not always.

Yeh its been improving over time. Three steps forward, one big step back. I'm sure they will address it quickly.

Aww thanks for the shout out, and I should clarify this a bit, or else I may be giving others false expectations.

It may no longer be possible now, with the updates to the drafting dynamics, but it's still worth a try.

What I can say though, is that as recently as November, one can do ~20% less W/kg than the red bot when she's cruising the flatter routes on Zwift. It was possible to hold her wheel at 3.3 W/kg (myself at 71.5 kg), while she was doing 4.2 W/kg.

The larger the group, the more difficult it is to do this. It's next to impossible to do on Brevet's group (B-group), because it's usually about 15 people or more, and the front of the group gets reduced CdA when approaching other cyclists, while the rear has to crank up the watts to keep up. I've never managed to do less than 3.1 w/kg when cruising on the flats in that group.

Otoh, other riders on the course are effectively "invisible" to Coco's group, and this is where one could possibly learn to get a better sense of when one is in the draft. I could tail-gun at ~2.2 w/kg; however, there are occasions when I need to dial things up just a tad to stay in touch. After a few sessions, one gets a sense of what is safe-ish and what requires a few hard pedal strokes to stay in contention.

As @trail mentioned, this is no different from race craft for IRL racing, where one should learn how to move through a pack without having to resort to side-blasting.

One other thing, bike choice on Zwift matters, but not in an immediately intuitive manner. At least ~last November, a Tron bike, despite being marginally better in terms of aerodynamic performance as compared to something like Tarmac SL6 (2-star aero and 4-star mass) paired with the best all-round wheels, performs a lot better in a group. My sense is that Zwift ranks who gets to follow wheels partially based on the bike + wheel used, and even slight advantage in aero gets one further ahead. So @rich_m, if your mass and height is generally comparable to the others, but you aren't using a very aero bike, that alone could cost you perhaps 0.3-0.4 w/kg and shunt you further back as compared to a hypothetical where you would be on a Tron bike or similar.

________________________________

It's also worth repeating that most Zwift racing places a premium on repeatable 1.5-5 min power output over something steadier. This sort of mirrors real life (where most races are won based on 3-8 min power), but gets skewed toward the anaerobic end of things on Zwift. With most here focusing on longer-duration power output (and rightfully so), it shouldn't be surprising that this focus doesn't give one the predispositions to do well with short-term power output.

after doing a bunch of races you can use zwiftpower and strava to figure out your power profile, specifically what max power you can do for any given time/climb duration. For me, i know that hilly KOM is best 2 min effort, reverse KOM is 4 min effort, epic is 20 min, volcano is 7 min etc etc. you can use strava to look at all your efforts over time and see the trends. you mention getting dropped on 3% climb at 4.7 w/kg, without knowing duration of climb it's hard to say. If we are talking short climbs (1-3 minutes) you will probably need to be way over threshold. For A&B fields probably looking at 6-7w/kg for short climbs, maybe more for the stacked As or A+. if we are talking 5-10 minute climbs, then 4.7w/kg is very good for a B race, you shouldn't be getting dropped. would be a pretty standard A level effort, and sandbagging in C or D.

Hey!!! My plan works :)

A slightly more cynical rendition might be as follows. When they introduced anti-sandbagging, it was intended to be a work in progress. However, they do not appear to have put much work into it. Their stance is that only a minority of users race. They seem to feel that efforts to introduce new features have a better return on their investment. So … maybe they intend to get back to anti-sandbagging at some point (Zoon, as people say). But there are competing demands on their time. I mean, at some point, we’ll get bored of the Makuri islands, and they’ll need a new world. There are a bunch of little bugs to fix as well, e.g. the London elevation graph, Dura Ace 9200’s in-game weight apparently being bugged, various visual bugs, etc that they eventually need to get to. Plus the racing community is clamoring for them to shift towards some sort of achievement-based category system, and in the interim to at least hard enforce the W/kg requirements for category assignment. So they are presumably aware of the issue and they may even intend to act on it, but things keep coming up and it’s just lower down the priority list.

He more cynical version is that they can’t be bothered.

thank you (and trail) for the advice. will look at my equipment choices. i have an SL7 in Zwift, and i think 202 wheels - but will see if i can get anything better at my level and number of drops.

i can see what you mean with getting better with position in the group. i do have those moments where my bot takes a bad line at a corner, and suddenly there is a 2m gap needing to be closed and i have to step on it for 5-10 seconds (cos if it gets any bigger it is impossible to close), the speed and draft carries me through to the front where i don't want to be, but when i back off just a tad the pack seems to catapult by and i am hanging on at the back again at the next corner. i thought it was the same for everyone. Similar things happened when i did C races. i just checked the last one and i averaged 3.8 w/kg to stay with the leading group and get outsprinted by most of them (but no-one else out of the 17 in the leading group are over 3.6, and many are quite a bit lower (3.1-3.2) - and the absolute watts indicate they are mostly not big guys doing a lot of watts but low w/kg). Difference is that in a C i had just enough power to compensate for the lack of pack skills.

Since this topic is back up again.........why don't they just move to a ranking/points system like the rest of the world does for racing or competing in almost any other sport in the world? I mean, can someone give me a reason why? Is it some attempt at fairness? All the needed database values already exist for them to do that.

I thought in the past when talking about IRL racing it was said the craft mattered to the extent you can have grossly overpowered racers who just "don't get it" and don't do well. And then weaker riders who do 'get it' and win. So w/kg does, but doesn't, matter.

In real life this plays out well with time trial. The US doesn't rank by results in TT, but mass start points. Which doesn't work well. The UK does rank TT points.

Zwift........move to points/ranking versus power and weight based. Allow for promotion AND demotion based on the points. You race a lot and learn to do well with the power you have and win a lot, you move up. You lose some power and start always getting dropped, you lose points and get demoted.

Wish I could give my drop away so somebody could use them. I have over 6 million...I don't know if that is a lot.

<rookie numbers meme>

I think that's more susceptible to gaming than W/kg. Train for a few months in races, being sure to just dangle off the back during every finish. Get demoted. Once the ZRL season rolls around, you're in prime sandbagging form. Just finish in the top 3-4 most of the season to avoid making yourself too obvious, then dominate the last race.

If there's a perfect solution to sandbagging (other than people being...good people and unlike jaretj), I don't know what it is.

They made that aspect very non-intuitive. The ideal when chasing back on Zwift is to do ~1.5 w/kg more than what most of the group is doing (call this group mode w/kg), and start gradually dialing things toward the group mode w/kg when you see “close the gap” sign (usually at 6 meters from group). Technically, you are receiving draft benefit even at ~15 meters back, but if you dial it down at that separation (~1 second gap on flat terrain), the gap will only grow. However, the “attractive force” of the blob kicks significantly at small gap distances, which allows you to pedal softer (than your chasing power output) when you are within 6 meters of the group. Once safely within the group, you can start doing the group mode w/kg. That you get shunted to front upon rejoining is merely an artifact (people within the group get shuffled to the front for doing group mode w/kg, this is known as “churn” in zwift parlance); ignore it.

You generally can never do a full-on soft pedal (usually 1.5 w/kg less than the mode), but any time you find yourself on the front, you can dial it back slightly if you feel like you have to do something; or alternatively, if you just maintain power at group mode w/kg, you'll get churned back into the middle. If the group is small enough (say 15 riders), you can actually coast a bit (e.g. coast 2 sec and pedal for 8 sec; 4 sec off and 6 sec on if you get good at it).

Also, that SL7 + 202 set-up is definitely hurting you in the pack. For whatever reason, the group order algorithm, when picking between you vs someone on an aero bike but all else equal, would always place you further back. This effectively magnifies a very small difference into a large one, as riding at the back of a group is often more subjected to surges (and game-ender for most beginning phase of races).

just checked and i can't even tell if it is an SL7 i have. Picked up an aeroad so will see if that helps me keep contact with the B pack.

The generous answer is that it's not that simple to create a points system. You have to determine which one. The Elo system used in chess is based on one vs one matches. I think people on the forum may have pointed Zwift to an alternate that's suitable for group competitions, but I can't remember which one it is. The Elo system's math looks simple enough, but someone at Zwift has to go through it and implement it in programming. All while they have to maintain the app's stability, sort out past bugs, etc. And all for what they perceive is a small percent of the user base.

Also, and I'm not a programmer so I'm open to having this argument disproven, as you increase the number of individual features in a program, the increase in the number of interactions between features is not linear. So, the overall complexity grows more rapidly than the number of discrete features. Basically, if they add complexity to the program, there's a higher chance of bugs developing. Earlier, I made a snide comment about how the London elevation graph has been broken for well over a year, and they have even listed it as a known problem for over a year, but apparently it was always on the back burner. I mean strange little things like that are likely to crop up as they expand the code base even further. Plus they have to make the code current for several operating systems (on different types of computing platform as well). So, they may have a genuine challenge with complexity.

I'm not trying to excuse their inaction, just to give a generous interpretation of why they have not yet moved on this.

Of course, going back to a different thread: this is why I would not be very confident that Zwift can integrate TrainerRoad's software and not cause a lot of short-term pain for both user bases. In that post, I also meant in a bigger organizational sense of being a competent steward of both softwares and their communities, but that's not the issue with this particular question.