In December 2008, a Contax with an ivory finish sold in Vienna for €11,400. It came back to the same rooms sixteen years later and made €9,600.
A Tower 46 bought for €840 in 2007 returned a decade later and made €480.
A Steinheil Casca II outfit bought for €2,400 in 2009 sold in 2015 for €2,400. Six and a half years, to the euro.
A Hugo Meyer Trioplan bought for €360 in 2006 made €900 eight years later.
And a Taylor-Hobson Cooke Anastigmat, bought in 2014 for €1,320, came back a decade later at €10,800.
Every one of those was held for years. Between them they cover minus forty-three per cent and plus seven hundred. Any of the five would make a convincing anecdote, and each would prove something different.
Which is why we counted.
The belief being tested
Underneath most camera collecting sits an assumption that mostly goes unexamined: hold the thing long enough and it will be worth more.
It is not a mad assumption. Prices for collectable cameras have risen over the past twenty years, sometimes steeply. But "prices have risen" and "your camera will make you money if you wait" are different claims, and only the second one matters when you are deciding whether to sell.
Comparable sales cannot settle it. Two Leica M3s in the same condition fetch different sums for reasons nobody records. What you want is the same physical object, sold twice, with everything else held constant.
Leitz Photographica publishes its results in full. Reading them back to 2005 gives 113 ordinary cameras that came back to the same rooms and sold both times — same serial number, same description, years apart.
This is an unusual way to look at a collectables market, and as far as we can tell nobody has done it here before. Price guides average across different examples of the same model. Indices track what a category did. Neither can tell you what happened to an object. Repeat sales of the identical thing are how housing economists measure house prices, for exactly this reason — and an auction house that publishes serial numbers makes the same method possible for cameras.
It also produces a small sample, which is the trade. There are only 113 of them.
The typical outcome barely moves
Sort those 113 by how long the owner held on:
| Held | Cameras | Median result |
|---|---|---|
| Under a year | 30 | +6% |
| One to three years | 29 | −8% |
| Three to six years | 26 | −12% |
| Six years or more | 28 | +28% |
Read that column top to bottom and it does not behave. Up, down, down, then up. If waiting reliably raised the typical outcome, this is where it would show, and it does not.
We could stop there and report that patience does nothing. Several things in the same table say otherwise.
The odds behave very differently from the middle
Alongside each median, count how many cameras gained heavily against how many fell heavily:
| Held | Gained 50%+ | Lost 20%+ | Large gains per large loss |
|---|---|---|---|
| Under a year | 3 | 9 | 0.33 |
| One to three years | 4 | 11 | 0.36 |
| Three to six years | 10 | 8 | 1.25 |
| Six years or more | 8 | 5 | 1.60 |
Large gains for every large loss, by holding period. Above the dashed line, good outcomes outnumber bad ones.
That climbs steadily, and it is the only thing in this data that does.
Sold inside a year, a camera was three times more likely to fall by a fifth than to gain by half. Held past six years, it was more likely to gain than to lose. The odds swing by about five to one across the range.
What is actually happening
The shape underneath explains why the middle stays still while the odds move.
The floor barely shifts. In every band, the worst quarter of outcomes lands between 71% and 85% of what was paid. Waiting did not make bad results less bad.
The ceiling rises. The best quarter climbs from 1.34 in the first year to 1.63 past six.
Put formally: the chance of a large gain rises with holding time, and the chance of a large loss does not fall. Fitting each against holding time separately, the gain relationship is the stronger of the two by a wide margin, and it is the only one that reaches conventional significance.
So the honest sentence is not "waiting pays". It is:
Waiting does not protect you from losing. It buys you a chance of winning.
Which is a smaller claim than collectors usually make, and a more useful one.
What we cannot say
We are going to be blunt about the limits, because they are real.
The sample is small. One hundred and thirteen objects is enough to see a large effect and not enough to settle a modest one. Everything below follows from that.
The overall correlation is weak. Across all 113 objects, the relationship between holding time and outcome has a rank correlation of 0.15 — about 2% of what happened to these cameras. Ninety-eight per cent is something else: which camera, which year, who was in the room.
Its confidence interval touches zero. Resampling the 113 objects ten thousand times puts that correlation between about −0.03 and +0.32. Around 95% of those resamples came out positive, which is suggestive and is not proof.
The strongest single result is marginal. The rise in the chance of a large gain has a z of 2.2. That clears the usual bar, but only just, and we chose the "large gain" threshold ourselves.
What holds up best is the consistency. We tested every possible place to cut the data — 73 of them, from a few months to sixteen years. In all 73, the longer-held group had better odds than the shorter-held group. Not most. All. And no single object drives it: removing any one of the 113 moves the correlation by less than 0.03.
A weak effect that points the same way every time you look at it is not nothing. It is also not settled.
What a collector should take from this
If you are buying a camera and telling yourself it is an investment, this data does not support that in the way you would like. The middle of this market is noisy, and your camera's fate is mostly about your camera.
What the numbers suggest — gently, without insisting — is that time changes the shape of the bet rather than its centre. It does not make the downside smaller. It makes the upside more likely to happen to you.
The Contax that lost sixteen per cent over sixteen years and the Cooke that made seven hundred over ten were both held patiently. Only one was rewarded, and nothing in the catalogue entry would have told you which.
So, as plainly as the evidence allows:
Over 113 cameras and twenty-one years, holding longer did not raise the typical result. It did raise the chance of an unusually good one, while leaving the chance of a bad one about where it was. The effect is modest, the sample is small, and the direction has been consistent every way we have cut it.
We will check again
This question answers itself with time. Leitz holds two sales a year, and each adds a handful of cameras that have been through these rooms before. The traceable population grows by ten or fifteen a year.
At around two hundred objects, a correlation of 0.15 becomes something you can settle rather than argue about. That is roughly six years away — which is, pleasingly, about the holding period this piece is uncertain about.
Other records would help sooner. Houses elsewhere publish results the same way, and a second archive of comparable depth would nearly double the sample overnight. We are looking.
Until then, this is what the evidence says, including the parts that do not cooperate.