Sales before vs after 4 years is the ideal statistic to judge a systems legs imo. Using absolute numbers without converting to percentage doesn't make sense as legs are all about how a product resists decline not how well it sells. It should be proportional. A system that sells 60m in it's first 4 years and then 20m after doesn't have stronger legs than a system that sells 20m in it's first 4 years and then 18m after. One has barely declined, thus strong legs, the other has declined massively, yet is still the more successful system.

As for the point about the year changing it's outcome, I did year 5 as well the results remain similar. "after the successor is out" would be a damning metric to use for Nintendo products.

"If the switch has a similar increase in 2020 and sells a little above 2019 in 2021, it's getting around 100m at the end of 2021."

I expect it will have a mild increase in 2020, nothing major. I don't think there's much chance at all that 2021 can match, let alone beat 2019.

To get 100m by the end of 2021 what are you expecting? Something like this?

2020: 28m
2021: 22m

This seems VERY optimistic, but you're right all we can do is speculate. We'll have a much clearer picture by summer 2021. We'll know by then how high the peak was in 2020 (Assuming it does peak which isn't completely guaranteed) and have some indication of how big the decline in 2021 is.

Nu-13 said:

...

The switch is selling too fast to sell under 100m, it might not have the legs of a PlayStation but it's definitely going to be one of the strongest Nintendo systems when it comes to legs.

I'd say the opposite to you, that the Switch reaching 100m is a near certainty.

The only way it won't is if 2019 was the peak and 2020 see's significant decline, I don't see this happening. 2021 is the year when the decline rears it's head.

My prediction:

2017 - 13.1m 
2018 - 16.5m (29.6m)
2019 - 20.5m (50.1m)
2020 - 21.5m (71.6m)
2021 - 16.5m (88.1m)
2022 - 11m (99.1m)

With years after 2022 bringing the LT total up to roughly 110m.

I don't see no reason at all for the sales to drop off Rick's cliff like that starting 2021 except maybe wishful thinking.

The Wii dropped off so much because of Nintendo stopping making games for it and instead focusing on the 3DS and the upcoming Wii U, and due to this the casual consumers flocked to 360 and Kinect instead while Gamers waited for games that weren't coming anymore. That's something that certainly won't happen with the Switch, as Nintendo repeatedly stated that they want to run the Switch for longer than the usual 5-6 years; because they don't have another console or handheld on the side to support and because neither Microsoft or Sony are producing anything similar to what the Switch delivers.

I could see 2021 going down a bit, but probably stay at around 18-19M. However, the real culprit here is your 2022 prediction, where you're predicting a crash just like the Wii did. That won't happen. Sales will drop of course due to age, but I'm expecting more something like 15-16M and nothing like a crash down to 11M like you're predicting.

You're right, lifetime sales is much more important. Better Legs doesn't mean more successful. 

I did a quick research on long legs... they are very important and nice to look at:

https://www.bing.com/images/search?q=long+legs&FORM=HDRSC2

https://www.google.com/search?q=long+legs

"Third pillar" was said regarding the DS as a way to back out of it in case it didn't sell well; that way Nintendo could have simply continued with the Game Boy line while the DS would have been remembered as a failed experiment that was done in addition to the usual home and handheld consoles. NX was always meant to be a successor to the Wii U and 3DS, but Nintendo's statements that the actual name of NX will neither include "Wii" or "DS" were something that some people took and ran with to mean that Nintendo was planning to support three different platforms side by side; however, that was contradicted by Nintendo's statements that their software output will be consolidated on a single platform in the future. I don't recall anything said about the 3DS.

So much for the history lesson, now for the meat. It's hardly relevant what Nintendo said about the lifespan of Switch. What's clear is that Nintendo won't face another transitional period where up to four consoles see game development concurrently, so Switch will continue to receive multiple big Nintendo IPs after year 4 which is a key difference to all of the Nintendo consoles you've had on your list today.

Fall 2021 is when I expect a mid-gen refresh to happen for Switch. Not something along the lines of a Pro model, but something similar to what the New 3DS was to the 3DS. Price cuts and revisions continue to be glanced over in Switch sales predictions, just like the presence or absence of similarities between Switch and previous Nintendo consoles. Since it's already clear that Switch will be backloaded in terms of price cuts and revisions, the only remaining Nintendo consoles on your list that are useful for comparison are the DS and 3DS. If you narrow it down further by looking at overall software support, the DS is much closer to Switch than the 3DS is. If you expect Switch to sell 70m by the end of 2020, 120m is a conservative estimate for lifetime sales based on your table.

Unsurprisingly, I expect more than 120m lifetime, because unlike the DS, Switch won't be limited to Pokémon in regards to big Nintendo sellers after year 4.

"Say console A sold 100m in 5 years and 130m lifetime vs console B 50m in 5 years and 75m lifetime. Who had the better legs?"

Console B, legs is all about sales curve. B had a less sharp decline, thus it had stronger legs.

"And does it even matter when total sales are more important? A system selling 100m shortly is still more successful than one that reaches 70m steadily over a decade."

You're right, lifetime sales is much more important. Better Legs doesn't mean more successful. Looking at the legs of previous consoles is just a way to surmise what the lifetime sales could be. Regardless your original point was "I don't see these ps4 legs you're talking about." so I provided data showing you those legs.

"26-27m in 2020 and 22m in 2021."

I think closer to 22m in 2020, but if you are right and the Switch sells 26-27m in 2020 then beating the ps4 seems likely.

For measuring the legs of a console I like to use the percentage of total sales it sold in it's first 4 years. E.G if a console sells 70% of it's lifetime sales in it's first 4 years it has much worse legs than a console that sells just 55% and has another 45% still to come. The PS4 sold 67m in it's first 4 years so let's go with the PS4 selling 120m LT to put it in the table:

You're right it has neither bad or good legs... for a PlayStation. But it exceeds the legs of every Nintendo system that's been released since they joined the market.

"At that point the switch will be at around 100m. It will keep dropping in the following years, ending it's life at 125-130m."

(...)

As it turns out Nu-13 is exactly right.  PS4 has average legs, for a leading console.  The main factor that makes the PS4 have good legs is that it is the leading console for the generation.  Once you compare it to other leading consoles, then it has average legs.