Gagosian scores a Rembrandt cheaply; AI gets a little better at predicting prices

This commentary by Marion Maneker is available to AMMpro subscribers. (The first month of AMMpro is free and subscribers are welcome to sign up for the first month and cancel before they are billed.)

Christie’s London PWC Evening = £79.2m

With David Hockney’s double portrait of Henry Geldzahler and Christopher Scott selling for just under $50m, Christie’s was able to realize almost $375m for the entire Ebsworth collection. That sale is significant because it moves the collection from a most likely loss against the massive guarantee Christie’s is said to have offered to a probable gain for the auction house. Such a high-risk endeavor is rarely taken. But the Ebsworth gamble came just after Christie’s much larger bet on the Rockefeller estate also paid off by showing a profit for the house.

The question is whether the performance against such priced-to-perfection guarantees will whet either auction house for more risk in the seasons to come as the business shifts into a phase driven by estates and collections than by deal-making.

It should not be lost on most that almost half of the sale’s total came from that one lot. The next most valuable lot in the sale came in at 1/8th the final price of the Hockney. Nicolas de Staël’s painting of bottles took a run at doubling the high estimate. Two other works by the artist also performed well within the estimate range making de Staël a surprise force in this sale.

Cecily Brown’s Night Passage sold just above the high estimate and fitted this work into the broad range of others selling this week and over the last year and more with little sign of a waning appetite.

But the overall diminution of the market was demonstrated by the presence of Gerhard Richter’s AB Tower in the sale at the number 4 spot of the top ten works. Purchased 16 months ago in New York for $3.8m, it was offered by an Asian seller without a guarantee. The £2.6m hammer price meant a loss of $400k loss for the seller but a $700k commission for Christie’s.

Elsewhere in the sale there were impressive prices for Barry Flanagan’s Nijinsky Hare which made a record at £1.45m and Jordan Casteel, whose work François Pinault was observed studying intently, made £299k for Patrick and Omari.

Gagosian Scores Rembrandt Coup for a Few Sticks

Gagosian has gotten a coup by arranging for Rembrandt’s late self-portrait to be loaned to the gallery for a show of selfies by a range of heavyweight artists. The Financial Times has some of the details of the show:

• “Jenny Saville, who last year became the most expensive living female artist at auction when an early self-portrait sold for £9.5m at Sotheby’s, will produce a work directly in response to the Rembrandt. It will be one of three new self-portraits for sale in the show, alongside commissions from American artist Richard Prince and German painter Albert Oehlen.”

Gagosian are paying to restore the frame on the Rembrandt picture, Self Portrait with Two Circles from 1665, as well as boosting awareness of the picture and its presence at an English country home:

• The show, “Visions of the Self: Rembrandt and Now”, will run at Gagosian Grosvenor Hill from April 12 to May 18 and entry will be free. Richard Calvocoressi, director and senior curator at Gagosian London, said: “We’re in central London and we get a large number of visitors. When the Rembrandt goes back to Kenwood after the exhibition for its celebratory show it will be higher in the public consciousness, one hopes, because of it having been here.”

In the FT, art historian Bendor Grosvenor raises some objection to the show mixing a few works for sale into the non-selling show. But on Twitter, he was more direct:

English Heritage are renting out a Rembrandt to help Larry Gagosian sell pictures – and at just £30k (to restore the frame), the rate is way too cheap. https://t.co/XOkM6qcubf

— Bendor Grosvenor PhD (@arthistorynews) March 6, 2019

AI Gets Better at Predicting Art Prices, Sort Of

Christophe Spaenjers announced on Twitter the publication of a new paper on using neural nets to predict art prices.

My new paper on the power of machine learning to predict prices of illiquid and heterogeneous assets such as art is now available here: https://t.co/Wclbk8aD8M In short: human experts + machines > human experts > machines.

— Christophe Spaenjers (@CSpaenjers) March 5, 2019

If you’re an auction house specialist, don’t worry. The machines aren’t ready to replace you at pricing works. But they are getting better and there’s reason to believe that specialists using technology will be better than specialists relying on their experience and intuition:

• “we show that machine-learning valuations can help overcome human biases in expectations formation. We present evidence that auctioneers have a tendency to ignore negative information—leading to systematic patterns in their prediction errors. More specifically, pre-sale estimates are on average too high for works by artists that are associated with relatively low prices or returns in recent years, suggesting that auctioneers are reluctant to adjust their valuations downwards. In these subsets of assets, machine-learning valuations are particularly likely to be able to improve prediction accuracy”

Then, again, the problem might not be the auctioneers ignoring “negative information” as much as the consignors not wanting to hear that negative information.