$68 million changed hands in online auctions last week, across just over 1300 vehicles.

The typical sale was $23,750, almost unchanged from $23,646 the week before. At the top end, two cars alone accounted for $7.2 million.

One quick note on the comparison with last week. I included a few live-auction houses in that total. Going forward I’m going to separate those out. They’re still very much part of the same market, but the auctions themselves are more sporadic and behave a little differently, so I think they’re better looked at separately. We’re still monitoring them, of course, and I’ll come back to them when there’s something worth digging into.

A one-owner 2005 Porsche Carrera GT with 5,600 miles made $3,824,000 on duPont Registry Live on Monday. Three days later, a 2017 Ferrari F12 TDF with about 1,600 miles made $3,425,000 on the same site. (the premium, which I haven’t show on DuPont is unlimited, so buyer pays quite a bit more than other online auction sites)

Bring a Trailer’s biggest sale was a 2014 Mercedes-Benz SLS AMG at $1.7 million. It had 576 miles. Twelve years old and it has barely left the garage.

Five Ferraris, two Porsches, a Mercedes, a Lamborghini and a Rimac. Between them, those ten cars accounted for $13.8 million, roughly one-fifth of all the money that changed hands last week.

That tells you quite a lot about where the money was concentrated.

This is the bit I was most interested in.

Last week I said I wanted to start following the cars where a seller turned down a bid and then came back later. Not just whether they eventually got more, but what actually changed. Did they move sites, try the same site again, or simply wait?

26 of last week’s sold cars had previously been listed without selling. 18 of the 26 eventually beat the bid their owner had turned down.

Of the 12 that changed sites, 10 beat the old bid. Of the 14 that came back on the same site, 8 did.

Waiting didn’t seem to separate them much. The typical gap between the no-sale and eventual sale was about three months, but the cars that came back inside five weeks did about as well as the ones that waited half a year.

The size of the difference is the part worth remembering. The 18 winners got roughly 20% more than the bid they had previously rejected. The eight that didn’t beat it gave back about 4%.

A few examples stood out. A 1986 Porsche 911 Turbo was bid to $255,000 on Bring a Trailer in April. The seller said no. Six months later, on the same site, it made $310,000.

A 1972 Corvette was bid to $19,750 on Bring a Trailer in September. Four weeks later it sold on Hagerty for $38,259. (Hagerty show premium included)

A 1967 Shelby GT500 turned down $263k in May and sold for $272k on Saturday, while a 1962 Corvette turned down $71,500 in April and later made $84,500.

It didn’t always work. A 1969 Camaro turned down $150,000 in July and sold for $132,000 on Friday. A 1985 Porsche 911 Carrera rejected $125,000 in April and took $122,021 last Wednesday.

There are often some significant factors that affect second listing, but 18 of 26 is enough to make me want to keep watching it, and 10 of 12 cars that changed sites beating the old bid is probably the number I find most interesting.

A failed auction clearly isn’t always the market giving you its final answer.

There was another group I wanted to watch separately: cars that had actually sold before and then came back on a different platform.

There were 24 of them last week. Twelve sold for more than the previous time, 11 for less and one was basically flat. The busiest route was Cars & Bids to Bring a Trailer, ten cars in total, with six higher and four lower.

Some of the moves were substantial.

A 2006 Mercedes-Benz G55 AMG went from $41,500 on Cars & Bids in August to $77,777 on Bring a Trailer last Friday, up 87% in six weeks.

It didn’t all go one way. A 1992 Toyota MR2 made $12,900 on Cars & Bids in June and $10,000 on Bring a Trailer last week.

Changing platforms can matter. It is definitely not a guarantee.

123 of last week’s sold cars had sold at auction before. 55 came back higher, 64 lower and four were flat.

The typical one was about 3% down, 18 months after its previous sale. That’s actually better than the week before, when the typical drop was 8%.

But this is where weekly data can make you look silly if you declare a rule too quickly.

Last week, the cars that came back fastest held their money best. This week, they were the worst group. Cars that returned inside six months were about 9% down on their previous sale, while cars that waited 19 to 36 months were 4% up.

Two weeks, two different answers.

That’s exactly why I want to keep running this rather than turn one week of data into a theory. If the same pattern is still there after four or five weeks, then we may actually have something.

Two 2016 Chevrolet SS sedans sold on Cars & Bids two days apart.

One had 118,800 miles, an automatic and an extensively modified 800-whp turbo setup. It made $23,750. Two days later, a mostly stock 6-speed manual with 108,900 miles made $30,500.

That’s only about 10,000 miles apart, but $6,750 in price.

The mileage difference alone clearly doesn’t explain it. One was heavily modified and automatic, the other mostly stock and manual. I wouldn’t pretend we can separate exactly how much each of those things mattered from two sales, but it’s a good reminder that two cars can look almost identical in the headline data and still be very different to buyers. (I’ll take the stock and manual every day of the week!!).

The top end did the work again, with two cars alone accounting for $7.2 million, while the typical sale basically went nowhere.

But the number I keep coming back to is 18 of 26. These were sellers who had already turned down a bid, and most eventually got more. Even more interesting, 10 of the 12 that changed sites beat the old bid.

That doesn’t mean saying no is always the right move. Some times it’s just about timings and sometimes its one of the million other factors.

But it does suggest a failed auction isn’t necessarily the market giving you its final answer. What’s great about the online market there are plenty of places to go for a second time and more coming….

We’ll keep tracking the numbers and sharing the raw data as well as a few insights each Monday.

I like this one, mainly because I’m curious where the market is on decent E39 Tourings now.

It’s a California car from new, 86,600 miles, and currently at $7,999 with more than 600 people watching. It’s not the engine or gearbox everyone chases, but it has the right look with the Style 37s and M Sport bits. Pasted text Pasted text

There are a few things to get comfortable with, including the damage history, but at around $8k I’m less interested in picking the car apart and more interested in the final number. If this gets pushed well into five figures today, that tells me people are starting to pay up just for a good E39 wagon, regardless of which one it is.

Project cars are definitely not for everyone, and this one is a long way from being finished. No engine, no gearbox, no interior, floors still need work, and the whole thing has been turned into a 993 Turbo/Speedster-style widebody that is going to divide opinion.

But I’m still surprised nobody has put in a bid. It’s a real 1969 912 Targa underneath it all, the heavy bodywork is already done, it comes with replacement floor panels, and the seller says the reserve is reasonable.

Maybe everyone is looking at the amount of work left and doing the sensible thing. But these are also the auctions where, if the reserve really is low enough, somebody with the right shop and the right imagination can end up with something pretty wild for not much money. I’ll be interested to see whether one bidder finally blinks before it closes.

I’m not actually sure when this one ends, which is part of the fun with Dutch Auctions. The price is currently dropping from its $22,000 opening figure and is sitting just under $20,000, but it also shows two proxy bids, so I’m half expecting somebody to suddenly jump in through one of those and grab it.

The car itself is interesting enough without the format. It’s a Euro-spec 280SL with the 2.8-liter straight-six and a 5-speed manual, dark blue over tan, 82,000 miles, and a bit of a life story having spent time in Germany, Hawaii and now Kansas. It isn’t perfect, there’s prior damage, some cosmetic stuff and the AC doesn’t blow cold, but at this sort of money I can see the appeal.

More broadly, Dutch have had some genuinely interesting cars lately, and I hope that continues. The format still takes a bit of getting used to, but it does make you watch the screen differently when you know the price is moving the other way.