So, I'm a doctoral student with access to some fancy statistical analysis software and always looking for ways to procrastinate. After spending a great deal of time looking at 50s/60s Omegas on ebay I noticed a great deal of variation in prices for seemingly comparable watches. The question is, what is it that makes a steal versus a steep price? My guess is that there is something involved other than chance. It occurred to me that it would be possible to run some regressions in Stata and figure out scientifically what factors contribute to prices other than qualities collectors look for.

For instance, if the end day of the bidding is on a saturday or after working hours in certain places it might contribute to higher prices because more people who aren't using last minute bidding services are available to engage in bidding war. Or possibly, if the seller starts out the bidding at a fixed amount versus 0 does that contribute to final price? Number of pictures? Etc.

The problem is I don't know enough to pinpoint all of the relevant variables or to judge how different watches compare on those parameters. However, with a little bit of cooperative effort I could fairly accurately figure out if there are any objective reasons why comparable items sell for different amounts and post the results for the benefit of the community. The corollary of this is that members on the forum would have concrete information as to both how to get the highest amount possible for watches they are selling and how to find the best possible deals with less effort.

I haven't fleshed out exactly how to organize such a collective action effort just yet but my best guess is that it would take on the order of 10 minutes a day of looking at watches for anybody involved and we would probably need to continue for three to five weeks to gather enough data. If there's any interest I'll devote some more time to figuring out the details of how we would go about this.

 

The first few mentioned is would be what we would be looking for. I couldn't agree more that what seems non-quantifiable that are (presumably) going to have the largest effect. But say it accounts for 90% of the price. With a $1k watch that's 100 bucks. Even for less expensive ones, over time that adds up. The field I'm in (political science) is often about quantifying what might at first appear subjective. The variable "quality of watch" isn't knowable. However, that's not exactly what drives the price. What is primarily driving the price is "community perception of quality of watch". That we can make valid inferences about! Simply asking what is the quality of the watch is problematic. But instead of using "community perception of quality of watch" as a control variable we break it down into "community perception of condition of movement", "community perception of condition of dial" etc. That would be easier to come to consensus on. So the idea is that if we have a few people coming to consensus on each one we get an alright idea of what "community perception of x" actually is. The statistical significance of using three coders for each observation might not be high enough to get published but the idea is to give a firm footing for how to get better deals/maximize income.

 

Agreed. Breaking it down in such a way that everyone only spends a minimal amount of time is the key part. I think if we had a handful of people who would be rating the watches it would be feasible for me to divvy up the listings though. We don't need to rate every watch on ebay, just a large enough sample to make the findings statistically significant. Technically the minimum is 29 + number of variables, but that puts you on pretty thin ice. Once you hit 120 + number of variables you're gold but we can get away with less than that.

Once one's nerd quotient hits a certain level you start avoiding studying statistics by doing more statistics!

 

Community perception will be a hard thing to really gauge. When it comes to watches, there are a number of communities you'd need to look at. There's the uninformed bloke who saw his mates PO and wants one too because he thinks its a great watch. He's the same bloke who saw a cracking watch on Don Drapers wrist in Mad Men and wants one of those too.There's the first time buyer who has done a lot of research and wants to pay less than MRSP for his PO. Then there is us, a dedicated community who know what is and isn't a fake/redial/Franken. All three have a very different mindset.
One only has to hang around forums like Omega Fake Busters on WUS to see what happens. I've used the PO as my example for the same reason.
You may have to restrict the analysis to the WIS community because the previous two groups will slant the data if that makes sense.
If I'm barking up the wrong tree here be gentle, I know nothing of statistical analysis, I'm just an EOD tech! Oh and I note that analysis has the word anal in it. Pretty much sums it up

 

where do you place passion? tension of an auction? there are so many variables in human psychology... go to an auction house and absorb the mayhem of high price items....but, why not? if time is no factor, why not try it? what could be the benefit? if i want a certain speedmaster, i set my auctionsniper to your average data and will never get one? because guys with passion for the same piece or deep pockets want it? i do not really get it ..... but would be interested in the results anyway ! good luck ! achim

 

There is no pattern where auction prices are concerned. Things like phase of the moon, a bad or good day at work, arguing with the wife, funds available at the time, etc. can all affect how a person bids. People are too much of a variable to make any equation useful.

 

You're probably right. My process is simple:

1) See shiny watch
2) Get googly eyed for shiny watch
3) See price
4) Check account balance
5) Buy or not buy shiny watch

That's it.

 

Forgot about that.

 

Yeah LouS, that pretty much hits the nail on the head. If nobody is up for it no biggie. I thought it would be a neat little thing that might be helpful for everyone. The best way to do it would be to have a running thread that had the randomly selected auctions and people could vote on the specific aspects of the watches that we deemed to be relevant variables.

I would still be interested in kicking around what specific aspects of the watches and auctions the forum thinks contribute to the overall quality. My initial thought was

movement shown [y/n]
refurbished [y/n]
visible problems to expert [y/n]
visible problems to layman [y/n]
condition of movement [1,2,3]
condition of case [1,2,3]
condition of dial [1,2,3]
rarity of movement [1,2,3]

...and then of course the quantifiable aspects to the auction that you mentioned before but do not require rating. I'm sure there's more to add to the list, though. As to desirability, that is pretty subjective. I have some thoughts on how to approach it but they would likely be too much work on my end.

@ Wheels



That's a really good point. Our sample (whoever is rating on this forum) is not necessarily representative of the population (everyone bidding on ebay). My own misguided first purchase is indicative of what you're saying. I dismissed that initially as I haven't really looked at the other forums. I will put some more thought into that theoretically.

 

I think it would be an interesting exercise and a good thread if posted here. It would contribute to our learning resources and in some cases track price rises or falls for a particular model

 

The issue of defining "community" and whether the the sample of raters here is representative of the "community" of ebay bidders would probably the biggest hindrance and might be the deal killer. I might have to ask some of my colleagues about that one.

A lot of what seems insurmountable is actually taken care of with some statistical magic that I would be a better political scientist if I could get a firmer grasp on. Essentially, it isn't necessary to rate every model because if there are enough of them (>120) the variations in how specific model affects price should be normally distributed. It's counter-intuitive but I've seen it explained with a lot of greek letters that made me reconsider a PhD program.

I'm going to dig into that second link a bit later, it looks to be right up my alley. ATM work has me swamped and my attempts to avoid it all day long have been too successful.