Monday, March 18, 2012

After Hurricane Sandy, at the end of last October, we were without power and internet access at home for over two weeks.  During that time Kay was able to keep up with the eBay listings during our daily trips to a nearby Internet cafe.  However, since the Bidstracker database is housed on my home desktop computer, I was unable to update Bidstracker at all during that time.  When Internet access was finally restored to our home, I found that I had a huge backlog of eBay listings to enter into the database.  Entering all of them into Bidstracker was simply not feasible given both the volume of listings as well as many other demands on our time.  I therefore took a very selective approach and, while I viewed almost every listing Kay had looked at, I entered data into Bidstracker for only the less common lamps.  I skipped entering data for those extremely common lamps since for those lamps since the database already had way more data than necessary to determine a statistically stable average value; entering additional data, from a statistical point of view, would not any significant information to the database.  Although I finally caught up with all of the old listings which I felt were unusual enough (i.e. for which we didn't already have tons of data), I decided that I would continue entering data for only the more interesting and somewhat less common lamps.  I am continuing to skip lamps such as the "Poppy" (S1-172), the "Daisy" (S1-112), the several Greek Keys as well as a number of other more common, less interesting and relatively low value lamps.  I am also skipping entering data for the more common bases found by themselves.  If you are interested in just when the last entry was made for a given variation or condition of a lamp, just scan the details page for that lamp.  The end dates of the auctions are shown for every entry on the detail page.  This new approach is saving me a great deal of time and is not, in my view, degrading the overall value of or results contained in the database.  And, frankly, the alternative to this new approach would be to stop collecting data for and publishing bidstracker...something I think about from time to time.

Monday, October 22, 2012

It has been over two years since we have updated this "Message of the Day" page.  That's mostly because has been nothing of significance to report.  During that time we have continued to update the Bidstracker database on a more or less weekly basis (although we continue to look at and record data on eBay listings almost every day).  Improvements in our software tools have quickened the process of maintaining the database for us (we now have an automated tool that checks and records the data that we used to enter after the eBay auctions ended).  The process of maintaining the database is still quite time consuming and so we've also stopped recording data on a few more of the most commonly found and generally less interesting lamps.  The database has so much data for most of these lamps that adding more data would not significantly change the calculated values.  The details page for each variation of each lamp shows the date on which each listing was recorded.  By looking at this page you can get a feeling for whether or not we're still collecting data on that particular lamp/variation.

Although we've entered some additional data for live auctions, there just haven't been that many significant or sizable such auctions over the past couple of years.  We did just enter data from the Whalen Auction held on October 11, 2012 in Neapolis/Grand Rapids, Ohio.  There we probably less than 75 listed miniatures offered in this auction, but some of them were ones that just haven't been seen that often.  The general impression we, and other attendees, had was that prices at this auction were quite high relative to other recent sales.  Some folks felt that this might be a sign that the market for miniature lamps was picking up; some of us felt, however, that this auction was probably just an outlier.  Nonetheless the data are recorded and our users can judge for themselves whether this sale has had an undue impact on a particular lamps value.  Items sold at the Whalen Auction are listed on the details page of any particular lamp/variation that was sold there.

Monday, August 9, 2010

We have just finished adding the data from two recent live auctions.  On June 22, James D. Julia in Fairfield, Maine, had an auction that included a very few (but mostly high-end) miniature oil lamps.  We have entered data for seven items from that auction into the data base.  On July 31, Jeffrey S. Evans and Associates had a major lighting auction including a large number of miniatures from the collections of Fred and Shirley Reesbeck and Jo Ann Dreyer.  We have also entered the relevant data for that auction in the data base; as is always the case, we included data only for miniature lamps listed in the four standard reference books and only for lots that contain a single item.  Prices entered in the data base do not include the buyer's premium (15% at both auctions for those paying by cash or check, 17% for those paying by credit card and 20% for those bidding via the Internet).  As of this writing, Bidstracker has data on 45,592 listed miniature lamps offered on eBay and 4,777 lamps sold at selected live auctions.

Monday, February 15, 2010

We have made some minor changes and have added some new features to bidstracker which we hope will improve the utility of the database and provide you with some hopefully interesting new information.  These changes/additions are as follows:

• First, in our continuing effort to show how market prices change over time, in addition to the trend line charts added a while back to the details pages, we have added a column to the summary page for each lamp.  That column shows the average high bid amount for the last ten lamps of a given book/figure, variation and condition.  The choice of the number 10 was totally arbitrary.  If a given item has had less than 10 examples offered, the value in this new column (on the far right of the page) will match the overall average value.  As an example, looking at a fairly common lamp, the green Nutmeg, S1-029, you can see that the overall average for the 98 examples of this lamp that we have tracked since June of 2002 is $113.86 while the most recent 10 examples offered had a high bid average of $106.26.  That represents a decrease in theoretical market value of slightly over 9%.  Note that, with 98 examples overall, the average high bid amount is likely to change only slightly with each new example added making the overall average a more stable estimate of the lamp's market value.  Since the new column only contains data on the most recent 10 examples offered, it is quite likely to fluctuate much more as new examples are offered and older examples drop off the list.  This new column, however, may provide a better estimate of current market value than the overall average. 

• Second, on the "FACTS" page we have added a new item to the "Did you know?" box on the left hand side of the page.  The "facts" in this box appear randomly each time the page is accessed or the refresh button is pushed.  This new item is the year to date total dollar amount of listed and tracked miniature lamps offered on eBay.  This value will change each time we update the bidstracker database (which is generally about once a week, on Mondays) and by the end of they year should reflect an pretty good estimate of the total eBay market for listed miniature lamps.  The figure may not be particularly useful, but we think it will be interesting.  As of this writing, that value is $35,324.14.

• Third, in a similar vein and also on the "FACTS" page, we have added an item to the pull down menu which is just to the right of the "Did you know?" box.  This new item has a perhaps grandiose title of "The total size of the lamp market by year".  Obviously the values returned include just the totals of the lamps tracked in bidstracker--the lamps we find and track from eBay and those we track from selected live auctions.  There are, clearly, many auctions which we both don't know about or from which we do not have reliable data.  Further, we have no way of knowing about or tracking sales of lamps in shops, at antique shows or lamps sold privately among collectors.  Nonetheless, we believe that the data returned when you click on this item, provide a glimpse at and an estimate of the total market for listed miniature lamps.  The most recent figure, as of this writing, for the year 2010 is a little over $148,000.  This is significantly higher than the eBay only market because of the approximately $113,000 worth of lamps sold at the January 30, 2010 Jeffrey S. Evans auction and tracked in the bidstracker database (see the entry below for Sunday, February 7, 2010).  The annual estimate of the total lamp market varies quite significantly.  Ignoring for the moment 2002 (during which we tracked lamps beginning in June) and this year which is just barely started, the smallest annual sales figure was approximately $270,335 in 2009 and the largest was about $715, 556 in 2006.  The table that you see when you click on this new item also shows the total number of lamps actually sold in each year and the average value of each sale.  These values also show quite a bit of variability, (excluding for the moment 2010) ranging from about $86 to about $130.  We believe the value for 2010 ($193) is artificially high and will come down quite a bit over the year.  The high value appears to be due to the large number of very expensive lamps sold at the Evans auction.

• Finally, we have added two columns to the "Best Day of the Week" and the "Best Month of the Year" queries on the "FACTS" page.  These columns are the number of lamps actually sold (we previously showed just the number of lamps offered) and the average selling price of those lamps that did actually sell.  While Saturday and Sunday are still the two best days of the week to sell a lamp on eBay, based on the percentage of lamps offered which actually sell, Sunday yields slightly better average prices ($84.66 vs. $78.75).  These difference may not be statistically significant (we have not tested for statistical significance and have no current plans to do so); nonetheless, it is interesting to see that the average selling prices vary from $68.49 on Thursday to $84.66 on Sunday.  While we previously thought that Monday was a somewhat middling day on which to have an auction end (only about 65% of lamps offered on Monday sell, while over 68% of those offered on Sunday do so), Monday has the second highest average sale price ($80.29).  Similar observations can be made about the "Best Month" for eBay auctions data.  Note that while September and December have middle of the range sales percentages, they have the two highest average sale values.

We hope you find these minor additions to bidstracker both interesting and useful.  We are working on a couple of other enhancements but have no realistic estimate of when the might come to fruition.  Your suggestions and comments are always welcomed.

Sunday, February 7, 2010

We have just added 352 lots worth of data from last weekend's (January 30) Jeffrey S. Evans auction in Mount Crawford, VA (formerly Green Valley Auctions).  There were 441 lots of miniature lamps at this auction  However, as noted previously,  we only include lots of correct listed lamps and exclude unlisted lamps, marriages, and mixed lots containing two or more non-identical lamps.  As of today (February 7, 2010), the Bidstracker data base contains data for 43,100 lamps offered on eBay and 4,612 lamps sold at various live auctions. 

Monday, November 3, 2008

For some time now we've been looking for ways to detect any long term trends which might be occurring in the auction prices of miniature lamps and for better ways to display the data contained in bidstracker.  The recent turn in the economy and our own perception (not backed up by any data) that sales of miniature lamps on eBay were declining in price and in volume, increased the priority of the effort to detect trends.  Some time ago, we attempted to look at trends by computing and displaying in tabular form at the bottom of the details page for each lamp/variation/condition, six month high bid averages.  That table did not turn out to be particularly informative and certainly did not reveal any trends at all.  Recently we became aware of some free software available from Microsoft which would allow us to display bidstracker data in graphic form.  We have now added to graphs to the bottom of the details page again for any given lamp/variation/condition.  When looking at the summary page (the first page that comes up when you search for data on a given lamp) just click on any single line to go to the details page.  Once on the details page you will most likely see two lines which both say "You do not have Office Web Components installed.  Please install it to view these charts."  Just click on the underlined words (Office Web Components) in either line to install this free software from Microsoft and follow the instructions which will appear on your screen.  After the installation is complete, this should only take a few minutes, go back to bidstracker (you may have to close your browser and reopen it) and where those lines were, you will now see two graphs.

The first of these is a graph of the high bid amounts for that lamp over time.  As currently configured, the high bid amount has been averaged for every month for which bidstracker has data and that data has been plotted.  This will show up as a blue, somewhat jagged line.  Superimposed on the blue line is a "best fit" trend line (shown in black).  This best fit trend line will always be a straight line... either straight across the graph (no change in high bid amount over time) or heading up (positive change over time) or down (negative change over time).  At the right end of the line will be some numbers.  The one that is of interest is the number which follows "R*2 =".  That number is the "regression co-efficient" or a measure of how good a fit that line is to the data presented.  The number should always be a decimal fraction between 0 and 1.  The closer the number is to 1, the better the fit of the line to the data.  Conversely, the closer the number is to 0, the worse the fit of the line is to the data.  Most of these regression co-efficients will be low and closer to 0 than to 1.  Thus it is wise to look at these trend lines with a grain of salt.  It is also important to realize that for a great many lamps in the bidstracker database there really isn't very much data.  It is very hard to detect a trend when there are only a few data points.  Nonetheless, we feel that this chart presents the bidstracker data in an interesting and informative way.  The bad news is that looking at quite a few of these trend line charts, we see a rather consistent downward trend in high bid amounts over the last 6+ years.  There are a few lamps whose value hasn't changed and a few whose value seems to have risen, but over all, chart after chart tends to show a downward trend.  Note that the exact format of this chart may change slightly as we experiment with the best way to depict the data.

The second graph is a histogram, or bar chart, which shows the number of examples of each lamp/variation/condition which have sold within different $10 price ranges.  This graph is intended to supplement the averages, modes and medians shown on the summary page.  You can see much more detail in this graph than can be depicted in a single number.  For lamps for which we have very little data, this histogram may not be of great value.  We've seen a number of these graphs which show that the distribution of high bid amounts for a given lamp is flat.  That is, one example of the lamp sold at each of the values shown on the graph.  While the average value computed for a lamp with a distribution like this is still a true and valid average, the reality is that the probability of a new example selling at any given amount shown on the graph is equally likely.  For lamps with a lot of data, the histogram does give a good picture of the distribution of high bid amounts and may be more helpful in predicting the amount a new example of that lamp will sell for than just the average, median and mode.  We will probably modify the labels on the x-axis of this chart.  Right now they are a little confusing.  As they stand now a "0" means any amount between $0 and $9.99, a "10". any amount between $10 and $19.99 and so on.  We're probably going to correct this so that the label reads as the midpoint of the indicated range (for the two examples cited above, $5 and $15).

We hope you find these enhancements to bidstracker interesting and useful.  We welcome any comments you might have about them.

Tuesday, September 16, 2008

September has been a busy month for miniature lamp sales with two major auctions--one at Roan's in Williamsport, PA on September 4th and 5th and one at Green Valley in Mt. Crawford, VA this past Saturday, September 13th.  The data for most of the listed miniature lamps offered at these two auctions have now been entered into the Bidstracker database.  We record this data in real time at the auction as the bidding on each item is completed.  We do occasionally miss the final bid amount on a few items (no more than 3 or 4 in each auction), so that a couple of items may be missing from the database.  Also, remember that we only track lots that have either a single lamp or two or more identical lamps in them.  We don't track mixed lots because it is impossible to know how much of the amount bid was intended for each item in the lot.  Still these two auctions added data on 419 items to the database.

Monday, July 28, 2008

We have entered data from Saturday's auction at Dotta Auctions in Nazareth, Pa.  The data are for only 25 lots from this auction, mostly skater's lamps with colored glass globes.  However, we found this auction interesting because all of the lots were initially listed on eBay Live Auctions, but, disappeared from eBay just a few days before the auction.  We asked about this at the auction house.  They were quite unhappy with eBay.  Apparently, because of poor feedback (which according to the auctioneer came from a single dissatisfied buyer), their account was suspended for 30 days and all of their items were removed from eBay.  They claimed that they could not reach anyone at eBay to attempt to resolve the problem or to restore their account.  As a result, they said they were through using eBay.  [eBay announced in April of this year that they would be discontinuing support for Live Auctions at the end of 2008.  This has led to continued rumors that eBay is attempting to abandon the auction format listings altogether.  According to eBay's April announcement, that is not true; eBay has no plans to discontinue auction format listings which has been the most central aspect of their business.]  In any case, because the lamps from the Dotta Auction appeared on eBay long enough for many collectors to see them and then disappeared, we thought bidstracker users would be interested to know that the results from this auction have been incorporated into the database.

Tuesday, July 1, 2008

We have now entered some data from two recent auctions.  First, an auction of a large number of miniature lamps was held in Cedar Falls, Iowa at Jackson Auctions on March 25.  Most of the lots in that auction consisted of groups of similar but not identical lamps.  Only a few lots consisted of single, identifiable lamps and those are the only lots that have been entered into the database.  It is really not possible to figure out or infer the selling price of individual lamps in a multi-lamp lot and that is why we don't track those.  Second, we have entered the date from the most recent (June 20) Julia auction.  While there were quite a few single lamp lots in this auction (and that's the data that's been entered) there were also a large number of multi-lamp lots which we skipped.

Maintaining the Bidstracker database is a time-consuming effort.  We generally spend a couple of hours/day doing it.  In order to save time we have decided not to track the sale of a number of listed lamps.  We use three criteria in deciding not to track the sale of a particular lamp; First if a lamp is not particularly old or if it is not primarily a lamp used for lighting we will consider not tracking it.  Second, if we have a great deal of data about that "not old" or "not really a lamp" lamp, we will decide to no longer track it.  Once you have a lot of data, adding more does not necessarily change or otherwise improve the value estimate derived from the data (for example, if there is a lamp that we've seen 100 times and it has averaged a high bid of $100 and then one of these lamps sells for, say, 3 times the average or $300, adding that lamp to the data only changes the overall average to $102--not significantly different from the previous average.  And, of course, the likelihood of something that averages $100 across 100 examples of selling for three times the average is pretty low).  Third, generally speaking, the items we've stopped tracking have a relatively low average value.

Here's a list of the lamps we've stopped tracking and the reasons why:

S1-094: Tin lamp painted in various colors.  Dates from the 1930s/1940s.  354 examples currently in the database.  Average high bid amount about $12.

S1-119: Clear swirled glass perfume lamp.  Dates from the 1920s/1930s or later. 417 examples currently in the database.  Average high bid amount about $12.

S1-143: Lincoln Drape.  Dates from 1930s/1940s or later.  Was a perfume lamp.  435 examples currently in the database.  Average high bid amount about $30.

S1-286: Clear glass Cosmos.  Later reproduction of an authentic lamp marketed by Rubicon as a perfume lamp. 677 examples currently in the database.  Average high bid amount about $25.  Note that we still track the frosted clear glass, painted white glass, authentic old white milk glass and the pink and yellow cased glass versions.  We are considering dropping the frosted and painted glass versions but have not done so yet.  We will continue to track the white milk glass and the colored cased glass versions.

S1-630: Vapo-cresolene.  Dates from 1879-1930.  Not really a lamp.  Among the most common of all.  1399 examples currently in the database.  Average high bid amount about $35.

S2-014 and S2-015: Jeweler's or barber's singeing lamp.  Not really a lamp.  About 204 examples currently in the database.  Average high bid amount about $50, but have sold for as low as $5 and as high as $210.

S2-106: Jeweler's or barber's singeing lamp.  Not really a lamp. About 109 examples currently in the database.  Average high bid amount about $25.

S2-182: Hobnail perfume lamp.  Dates from 1940s/1950s.  544 examples currently in the database.  Average high bid amount about $12.

S2-183: Same as S1-119 but with swirled glass shade.

S2-184: Squat swirled perfume lamp with reflector or with metal finger handle.  Dates from 1940s/1950s.  412 examples currently in the database.  Average high bid amount about $14.

S2-185: Squat swirled or floral embossed perfume lamp. Dates from 1940s/1950s.  98 examples currently in the database.  Average high bid amount about $12.

So-191: Colored glass quilted lamps.  Known to be not old.  192 examples currently in the database.  Average high bid amount about $40.

There are a number of other lamps that we are considering dropping from further tracking, but have not yet done so.  We will post information here on any others we decide to drop when and if we do actually stop tracking them.

Tuesday, October 23, 2007

We have incorporated into the database the results of two recent live auctions.  The first of these was the auction of about 60 listed miniature lamps at the annual Night Light Club convention the weekend of September 30 in Reading, PA.  The second was the auction of about 300 listed miniature lamps at Roan's in Williamsport, PA on Saturday, October 22.

Wednesday, January 24, 2007

You may have noticed some new information appearing on the bidstracker details pages (if not, go ahead and take a look at one of those pages--after you've gotten to the summary page for any given Book/Figure Number combination, click on any of the line items which shows the summary for each variation by condition.  This gives you a listing of the raw data that went into that summary.  Now, about the bottom of that listing there will be two new tables.  These are what we're talking about).

As we collected the data for bidstracker and as we've bought and sold lamps on eBay, we began to wonder about a couple of things.  Among those were a) Does the starting bid amount make any difference in the percent of lamps sold or the high bid amount? and, b) has there been any trend of prices for a given lamp over time?

The two tables now appearing at the end of the details listing page are an attempt to address those two questions.  A note of caution before we describe these two tables in more detail.  The data in some of the cells can be quite sparse (frequently only 1 or 2 lamps, even for lamps which are very common).  Thus any conclusions should be drawn very carefully and taken with several grains of salt.  Nonetheless, it is our hope that the new information provided will be of some use and some help.

The first of the two tables shows the percentage of lamps sold and the average high bid amount as a function of starting bid amount for lamps with and without a reserve.  Starting bid amounts are grouped into $10 ranges (without such grouping there would practically never be enough data in any one cell to make any sense).  For each $10 range, we report first, for lamps without a reserve, the number offered in that starting price range, the per cent which actually sold and the average high bid amount (lamps which received no bids are not counted in this average).  Then we show the same data for lamps which started in that price range but which had a reserve.  The average high bid amount for lamps in a given price range which did have a reserve includes the high bid amount for lamps which did not meet the reserve, but does not include lamps which received no bids.

We think that this table may be useful to eBay sellers in helping them to set starting bid amounts and to decide on whether or not put a reserve on a lamp.  So far, however, in the few cases at which we looked, no clear cut decisions jumped out at us--only some suggestions of what to do.

The second table attempts to provide some insight into the trend of prices for a given lamp over time.  The best way to present this information would be as a scatter plot of high bid amounts by sale date.  However, we don't have appropriate graphing software which will work with the data base we use (Microsoft Access) and which is supported on our hosting service.  Thus, we've developed an expedient alternative.  We've arbitrarily split time into six month segments beginning with June of 2002 (when we first started collecting bidstracker data).  For each six month segment, we've calculated the average high bid amount for a given lamp/variation/condition.  Despite a subjective feeling that the prices for some lamps have been increasing, we have yet to find any evidence of that.  We think that there may be some cyclical effect in operation, but we haven't yet attempted to study that systematically.  The data is there if you wish to look at it and perhaps it may prove useful.

We still have some additional ideas of things we'd like to implement in bidstracker...but it will be a while before they can get properly defined and implemented.  We have not yet updated the help pages to describe these two new tables, but will leave this message up on the site until we do that.