[daniela]

Sorry for late reply Mark. I agree with your summary (and with the fact there exists many countries with random train arrival, regardless of the tidy and nice timetable which they print), one last thing I would like to add, is that making *available* -1s / -2s bids means exactly that, an extra option, which people may or may not use. If these last second bids were made available, every user will weight the advantages vs negative consequences, will reach a decision for the auction they are about to snipe, and will either use this hypotetical new feature, or ignore it. If their snipe is rejected or falls under increment rule, it's their bid which fails, but effect upon general Gixen community and eBay community is negligible. As you point out it may be even riskier during high traffic times, but the risk is upon the user (OK, "WARNING!!! Late bids may fail" on top of "Use at your own risk") and it won't generate extra traffic or extra load, not on Gixen servers and not on eBay servers e.g. causing blacklisting threats and the like.

[Cupid]

OK, not a random arrival time for the trains... I was obviously talking about British trains there ... you clearly have not experienced our public transport system ;-)... An even distribution is what our ordered brains perceive in the absence of definite information to the contrary.

I agree with your second paragraph, but would add the affect of currency conversion to the mixture for some items where international shipping costs are not significant with respect to the value/maximum bid.

Actually I get nearly all my items for less than half the long term value I estimate, but since I'm not willing to pay more than that on the vast majority of items I take an interest in that is not really surprising, at least not to me.

Right, we don't really seem to be moving forward on what I think we should concentrate on as the main thrust of this thread... that is, bidding later to improve outcomes... and, perhaps because I did not explain it clearly enough, I note that you have not addressed the advantage over placing both your bids at the same time that placing the true maximum later has for any auction.

So, I'll try to summarise what I think we are able to agree on and then allow you to continue the discussion with others.. after all, anyone with enough resilience to read this far through this thread should know what both our opinions are, and that they are both unlikely to change:

There is agreement that something can be gained when we outbid ourselves and the next highest bid is less than one bid increment below one of our lower bids... we also agree that it is possible to place manual bids later than 6 seconds before the auction ends provided that we are sitting there prepared to do so, with a figure in mind that we would be prepared to bid (which I contend is a waste of time, and see very little evidence that people actually do... but that is open to further debate if others claim to be doing it often too)...

I'll leave it to you to try to convince others that the:

Justification> Likelihood of others deciding only to place the bid, that they have previously decided is their true maximum, after seeing your bid at 8 or 6 seconds often enough to make a difference, justifies having a 1 second snipe...

as opposed to:

Negative Consequences> Those late bids more often hit the bid increment rule and are thus rejected entirely, (or, less often, the first bid rule) and a slightly increased possibility of the bid not getting placed at all, (not because I think Gixen servers would be more inaccurate in that time interval but that when ebay experiences heavy traffic it impairs ebays' ability to recognise that bids were placed before the auction finished... which is not something improvements on the Gixen side could ever address).

[daniela]

Of course the arrival of the train is not random, everyone knows the timetable is not a random distribution. Did you expect trains to arrive at random times, in the story and/or in real life? of course not. What we expect is distribution to be uniform and identical for desired outcome / unwanted outcome, and indeed both directions have a train every half-hour throughout the day.

Similarly, ebayers don't place random bids. They (we) do place bids whose expectations (there are usually not enough bids in a given auction to say the same of actual bids) that are roughly uniform within a certain interval, centered around one or more "perceived price" peaks. Thus, if most people think something is worth 20$ to them, we may expect them to bid some random number between 19$ and 21$, or 19.50 and 21.50 (shipping costs are a good way to estimate how large the peak will be). We may also expect the "perceived worth distribution" to have peaks at round numbers and at round numbers minus shipping cost, because people will estimate, say, 20$ and then add a few pennies, we don't normally estimate 21.12 and then add random amount. While we can't control their bid, we can bias ours.

Examine the situation at -5s. The other bidder has no clue I am willing to bid and if necessary, shell out 27.54 and instead I know he is not willing to bid even 25$. How do I know? Because he is bidding at -8 and not at -1. The very fact he is bidding at -8 gives it away. He may well be playing the same game and be prepared to bidding against himself and everyone else the amount 27.53 or 27.56 or 27.54 at -1s, in which case we are on equal ground and the highest biddest and/or the luckiest wins, but I can tell for sure that, unless he is placing another bid at -1, his current bid is less than 25$. This is good for me, because I want to snipe the item at <25$ and do it right now. Then it is him who has to deal with the unlucky tail of the distribution and go off to another auction and compete with bidders that are more determined than both of us and deal with sellers who are starting to realize that prices are slowly but surely rising and decide increasing reserve / BIN.

If, instead, I were to bid 27.54 at -8s, while he is bidding 23.54 at the same time, he sees the price jump at 24.04 and might right away respond at 24.54 (minimum increment) or 24.99 (just below BIN) and then would give up or the auction would end (unless he is indeed determined to pay 30$, of course) and I win the auction, but a couple increments higher than I would have by placing second bid at -1s. I am making it very unlikely that I have to pay this extra amount, which I commit to, by withholding this extra bid until it's too late for anyone to react. I get almost all the benefits of nuclear war bidding but at a low risk.

Obviously as I have already stated, for many items this is stupid and/or counterproductive. You have mentioned items you got at half of the value you estimate, this means there are not enough bidders for the item to establish a "market value" at this time and the item is up for grabs for anyone who makes a valid offer without giving away to other people and potential bidders that the item is valuable (say with an early offer), in this scenario 6s before the deadline are perfect.

Again I am not stating that sniping at -1s or -2s is magically and automatically going to result in better outcomes. However I believe the users should be allowed the option to snipe late - at their risk - if so they wish.

[Cupid]

Not at all, I recognise that prices vary considerably from day to day, and of course value is regarded differently by every person so they reach very different conclusions.

My own methods usually result in my valuing the item at less than half the highest value it might obtain over a year or so.

The value of an item at any particular time is only what people are wiling to pay for it... a few minutes after the auction ends the value might be higher if the under bidder recognises that they can not get the item for less anywhere, soon... but by the time another is listed at a lower price the value may have dropped once more to less than was paid for the auction I started this paragraph with. Such is the nature of markets.

What my argument relies on is that the market fluctuations and randomness of the process are all irrelevant to the choice of successful sniping strategy, and thus it is not worth complicating my argument by introducing many more complexities which though they undoubtedly exist, have little to no bearing on the current discussion.

I do agree that your method of two bids is not as resource intensive as having many more bids between the current auction price and the maximum that you are willing to spend, but it also yields less favourable results than that option, so in moving in that direction the principle is the same.

The only way your method reduces the price paid is if the lower bid is less than one bid increment more than the next highest bidder at the end of the auction.

So, since this thread is focused primarily on later bidding, what is the advantage of your method, with a later bid over a method in which both your bids of differing amounts are placed at the same time, for arguments sake 8 seconds before the auction ends?

I don't see confirmation codes being an issue really, if ebay wanted to introduce those they could easily do so for bids originating from sniping sites like Gixen, though I guess they would be more likely to take such action of they got more traffic but that did not correspondingly provide them extra revenue as it does now.

The story relies on an element of the situation not being random, i.e. the arrival of the trains... so that begs the question, what element of eBay auctions that most of us assume to be random actually isn't and provides an opportunity to exploit by bidding later than Gixen currently provides the option of doing?

[daniela]

Your reasoning is founded upon the statement there exist a "TRUE" value of the item, which people, once we remove all disturbances, irrational bids etc. would agree with, and which is stable enough (for short periods at least). When your assumption is not reliable, for example there are two sorts of people, a group which thinks true value is about X, and another group whose members think true value is Y, that's when the strategy of b2 pays a lot, because it will bias bids towards the group of the low bidders and keep winning prices lower.

The story is a classical example in probability distributions. No, there is no cheating involved, Lenin does board the first train whatever that is, and arrives at the station at a random time. Actually in the fictional story he is presented having agreed this "solution" with the ladies who found it extraordinarily "fair", only later did the older lady realize how cunningly unfair it is.
Lenin (the story has explicitly chosen a character who has the power to do so, even though this detail may be overlooked at first sight) has signed a service order and rearranged the subway schedule. Now, trains in the direction of the young lover's apartment are passing at x:00 and x:30 every hour, while those in the opposite directions, the old wife's, pass at x:01 and x:31 every hour. So, even though there is exactly the same frequency of trains bound for either direction and even though the timing of his arrival in the station is truly random, probability in one direction is much larger than in the opposite, because if he arrives between x:00 and x:01 or between x:30 and x:31 he will see his wife for the evening, but if he arrives any time within the other 58min of the hour, he will be sleeping at his lover's. There is a small drawback, the expectation of the time he's waiting at the platform is larger than with the previous schedule x:00 and x:30 in one direction, x:15 and x:45 in the other one; he'll read the newspapers while he waits.

Shooting multiple bids in large numbers is risky because of confirmation codes, as you say it also creates much load on the server, too. OTOH strategies such as those I am discussing would not be a burden for the server.

[Cupid]

Well, even you are acknowledging that b2 is not bidding rationally.

The worst case rational bid when aware of 25 BINs available is 24.99.

In reality auction prices always lag below BIN prices (that remain active for more than about 24 hours), because BINs have other advantages to buyers, they are easier to purchase, arrive much sooner than waiting for an auction to finish and are a guaranteed win as opposed to a potential loss... these things are worth extra expenditure to the vast majority of buyers, hence prices are generally higher, though a seller runs the risk of underselling, not selling at all or in the best case (for the seller) having to wait longer for an eventual sale.

So b2 is better off buying up all the BINs available at 25 before considering any strategy that involves bidding more than $24.99.

Any form of irrational high bidding is going to win auctions, but doesn't really help the case for later bidding.

What you are tending towards, in my opinion, is a strategy where multiple bids all less than a bid increment apart are submitted all at once close to the end of the auction, up to and including the maximum bid... I have already agreed that such a strategy will lead to savings (of less than one bid increment), but the computing power required to achieve it can't be justified from the potential savings involved (the cost would, I think, be hundreds a year, per user, not, unfortunately, $20).

No, I have not heard the Lenin story before... would you care to explain why your hypothesis that the younger is favoured by random events holds true, because I would expect other unstated (non random) factors to be having more of an influence than you imply... like him choosing which platform to walk on to ?

[daniela]

Hi Mark, please notice (I have deliberately picked a very worst-case scenario) b2 is bidding 27.54$ for an item which is available BIN at 25.00$. If b2 was to repeatedly bid 27.54$ at -8s, the chance that he actually has to pay >25$ occasionally is greatly increased, especially if we take into account that the manual sniper has few seconds to react and lots of adrenaline and no time to check the ongoing price, but he may recall having seen a BIN at 25.00$ so the manual sniper inserts 24.99$ and b2 does win the auction, conveniently, but at 25.49$ every time this happens.

The strategy of b2 is meant to shape the probability distribution to maximise the outcomes we are interested in, which are winning many bids well below our max price and winning the rest of the bids for an amount close to our max price, which we do at the expense of a very occasional small loss which we control both in regards to amount (we have not bidded 100$ with the remote, but existent, risk we have to honor it) and in regards to probability (and here it is crucial to have a -1s bid that the opponent can hardly react to). If you like, it is like the story of Lenin and the subway, if you ever heard that one, Lenin has a wife and a young lover, they have an apartment each, every day Lenin after work takes the subway and boards the first train that stops at the platform: one direction goes to his old wife's home, the other direction to his young lover's. He arrives at the station at a random time. Yet he spends one night a month, on average, with the old lady, and is usually in the girl's bed. How comes? :D

Again, it is fine that you don't care for this strategy, but I'd be happy to pay Mario 20$ a year to automate it, besides bidding with my own money.

[Cupid]

daniela,

My contention is that b2 will win all the auctions if they place their 27.54 bid at -8s... Since that is what they are ultimately willing to pay that is the best result for b2, if 27.54 is more than they are willing to pay then they wouldn't construct a strategy under which they could end up paying that much.

What happens in reality is that if b1 needs the items and can't find cheaper alternatives they raise what they are prepared to pay until they start winning.

This is basic to the supply and demand model, where demand exceeds supply prices rise, where supply exceeds demand prices fall, it applies to an environment where there are snipers just as much to any other.

If items are also available as BINs and auction prices exceed the BIN price more people start buying the BINs until that supply runs low again ... and so the market continues.

The double bid strategy is always sub optimal in that there is a potential for a response (which could even be automated if someone wanted to be silly enough to construct such a service) but does not reveal the true amount that a bidder is prepared to pay.

I also contend that b3 has wasted their time and money by bidding manually at the last second, and that any rational being will realise this quite quickly after employing such a strategy on a few auctions... the last person that pointed out that it is possible, on this forum, only claimed to have done it once... so they proved it possible, but also proved to themselves that it wasn't worth doing, at the first attempt.

[daniela]

Mark
Your assumption is that we have bidder1 who follows your strategy and bidder2 who follows mine, and they happen to want to buy exactly same things and happen to pick same auctions, they happen to be the only serious bidders too. Also they give exactly the same value to the item, up to the random cents we add in (if we do). This assumption is not fully realistic in our scenario. However, even so, it is not as black-and-white as you are stating.

Imagine for simplicity an artificial example, both b1 and b2 always care for only one type of item, widely available, starting price 10$ buy-it-now 25$ and both of them have set their max they are willing to pay to 23.54. Also they are good mathematicians but not very bright, they don't notice a pattern in the auctions and that there is a rational adversary with repeated, and predictable, behaviour (ebay hides the name, but not completely ie it tells us which sellers this person bids with and other info, and if they keep bidding against each others they should realize quickly). Anyway they keep shooting according to plans:

b1 bids at -8s for 23.54
b2 bids at -3s for 23.54 and at -1s for 27.54

Every time they bid against each others, and no one else makes significant offers, b2 wins at 24.04 and is not highly delighted having spent 0.50 more than he wanted to, but such is life. He will be skipping starbucks and have coffee at home today. In real life b2 is going to win some auctions at 10.00 when neither b1 nor anyone else has placed a bid, but let us put this aside. What we should not overlook is that b2 is going to win some auctions at 24.04, auctions in which b3 has initially bid 15$ under the belief he was the only one interested in the item, then at -3s he saw price go up at 24.04 and he shoots 24.55 but is outbid by b2 last second snipe to which he can't react. Now b3 is annoyed at not having won, and will bid higher for the next auctions, but we don't care; on the other hand b1, who still needs a dozen copies of the item, has now to bid against b3, who by now has decided his price is 24.55 and so b1 will keep losing auctions or has to raise his own price. Instead, b2 keeps getting what he wants, delivered at his door tomorrow (I assumed item is widely available; what if it weren't?), at 24.04 apiece even in this worst-case scenario, which is almost as convenient as buy-it-now but it's cheaper than the buy-it-now price of 25.00. The strategy of b1 and the strategy of b2 are different compromises which are suited to different situations, I would however not dismiss outright b2 nor state outright it'd be a bad idea for the Gixen community. Like every tool it may be misused, or it may not work.

[Cupid]

I will have to agree there has to be some guess work as I also agree no one has access to all the data and variables involved.

So what we have to fall back on is our own experience and a good imagination for the scenarios that can occur, with an ability to think them through to their logical conclusion... now I don't claim I know them all, I was hoping, from this discussion, to gain some where the better strategy is always to bid later... unfortunately that has not proven to be the case, but, no matter, we have gone to other places instead.

What I can say is that for every strategy that involves bidding in the last second (that has so far been postulated), I have one that involves bidding earlier than 6 seconds before the auction ends that will result in winning the auction for at most the same amount... often less, without anyone being able to offer the possibility of any true reaction that had not already been built into their later bidding strategy.

However, for the strategy that involves bidding at 8 seconds before the auction ends I can't find a way of beating it, at the same price, with a later bid. The only way is to decide before that 8 second bid is placed that you are going to bid a greater amount.... and that is an amount the rational sniper should not be willing to pay.

[daniela]

It is hard to really compare strategies and determine the hypotetical outcome of an eBay style auction with another strategy as opposed to the one we actually used. In fact it is impossible for any of us, because we don't have access to data, namely, to the winner's max bid (except in very rare cases when someone bids just below maxbid+increment). But, even eBay would not be able to make an analysis which is 100.000% reliable, because there are many factors which are invisible to their data (and which we can only know if, say, we later become friends with the other bidder, go for a drink with them, and ask "what really happened in that ebay auction?" and they feel like answering). Example: John sees an item he is interested in, starting price 11.00$, Dick sees the same item because he is bored and spends time on eBay, Harry is also interested in same item. John decides he wants to snipe the item and so goes over to Gixen and inserts 31.76$ which is slightly above his perceived item value. Dick closes the webpage and goes to the coffee machine, and he is uninterested in buying. Harry sees the counter "00003" of visitors to the item, no offers, and decides the other two must be uninterested visitors, or the same one doing reload from different computers / IPs, or the seller checking that the listing looks good, and, one hour before end, goes over to Gixen and snipes 28.54$. John wins the auction at 29.04$ and is slightly annoyed, Dick loses the auction and mentally curses the other bidder, because had he seen another offer even at 15$ for this item, he would have not bid at all, given that other similar items regularly expire without selling at 12$, he had the bad luck of picking the wrong number and now the hassle of dealing with another auction, and the other guy had the same bad luck + he also has to pay. The seller is happy the item finally sold and off to packing it he goes. Now you may ask, but what really is the true value of item? If it is around 25-30$, why do identical items at 12$ not sell? There can be various answers to that, perhaps the item is worth 30$ to those who need it, and hardly anything to anyone else (suppose an obsolete spare?) perhaps everyone is buying at the same time and not buying at other times but someone does need them right now (seasonal items?) or whatever else, but, this can happen and I am sure it was experienced by (and has frustrated) most ebayers, buyers and sellers.

We could run simulations, we could even set up "sandboxbay" for people to make fake offers on nonexistent items and do data analysis on those, but nothing fully captures reality. There's people out there (again, factual story) who, in order to win an item they dont want to miss, set up new account and bid 5000. They are well aware they will never have to pay it, in fact if they don't like the amount they receive in the email, they can write eBay and ask for retraction: it was "obviously" a typo, as it's absurd with respect to average buy-it-now price. These one-off are not investigated by eBay, they don't even take care of the "regular" scammers!! Also, think about it, eBay does not ask for ID. Then there are "auxiliary" accounts and different sorts of scams which border on the criminal and I will not discuss either.

Math tells us my method would win in the limit of infinite number of bids, because we can't guess a distribution not even in the infinite limit, but we know what a distribution of distributions looks like. However with a number of bids of the order of hundreds per year or less (sure professional ebayers bid more than that, but I definitely don't) it's basically guesswork.

[Cupid]

No, I wasn't talking about shill bidding, that is a different illegal activity which sniping helps you to avoid but, unfortunately, still occurs occasionally even on sniped auctions... I won't ever accept any second chance offers as a result... rather I was talking about the use of multiple accounts under the control of a buyer... but it's a complicated scenario and, as I said, it's off topic so I won't explain further here, I wouldn't wish to explain how to implement criminal algorithms that could ultimately end up with people that utilise them getting (rightly) prosecuted.

Yes, I've experienced crazy bidding activity... it's not prevalent because, I imagine, it doesn't work long term... as with many if not most of the activities some people believe will get them an edge in ebay auctions. Nuclear bidding, as you call it, would be even less successful of people employed Nuclear Automated Sniping by placing crazy large amounts into automated snipers like Gixen... I certainly wouldn't recommend it as many sellers are quite prepared to play hard-ball with auction finishing prices and ebay rules about buyers being responsible for paying them... I would suggest more than are willing to roll over and take a lower offer... particularly as poor feedback is not an issue with non paying bidders.

I have also seen sellers conned into selling things far too cheaply by ending their auctions early.

With my strategy I often do get things at a much lower price than I would be wiling to pay... but I am always looking for a bargain, even at my maximum price... I also do a lot of research, so when I get a good price I know it.

Interesting though it is, none of this discussion gets us any closer to an argument that enabling the ability to enter later bids benefits the Gixen community as a whole, or even any particular members of that community... other than, and I will accept this... some would 'feel' better about it... but the evidence of my own experience is that though they may 'feel' better they would actually be doing worse in so far as they would win less auctions at the price they were willing to pay... not very many less, for sure, perhaps not enough for most to even perceive, but none the less .... less.

Now is that better 'feeling' actually worth the consequence of a worse outcome even if many people don't realise it is worse?

On that I think I will have to rest the case.

[daniela]

Because nuclear bidding wars are not under my control and there are bidders who are at the computer, one window shows current price (years ago one had to refresh, now is automated) the other window is ready to shoot their atomic bid. Some do bid crazy amounts to make "SURE" they grab the item, counting they won't meet another thermonuclear war bidder (I know because I've talked to some). To these people I am willing to leave the auction if I must, but also, I try my best to protect my bid, my very reasonable bid, from these people who just shoot, and, you know what? Sometimes there's a nuclear war with two crazies that shoot high, someone wins the auction at unreasonable amounts, the winner writes to the seller and negotiates, and the seller gives in and gives a large discount, why? of course, because they don't want to upset the crazies as they will buy again in other auctions of similar items, or will sell an item of interest and send an email to friends before listing on ebay, etc. This is not made up, it is factual experience.

I think the difference between your strategy and mine is that you always determine for every object a "not a penny more" price, while I try to determine an expectation and don't mind to occasionally pay something more on the single item because I am convinced that in the long term I get more items at discounted price with respect to "not a penny more" strategy. Both strategies are perfectly respectable and perfectly reasonable, I guess which one is better depends on the details of the bids, one might run a few computations but I don't have time, so, it's just a feeling and it may be wrong, plus it may not work well for the items a particular person is interested in bidding on.

If you are referring to a seller who has another account to bid on their own items, this can work only if there are high-value bids and they are early enough, which I don't do (even when I want to remove buy-it-now, I bid low). There are such people around, if they play that game regularly it can be noticeable (e.g. when their operation fails, we see in their history some item sold for a handsome amount, and yet it was relisted, either by them or by the fake winner) and IMHO the best policy is not to do business with them.

[Cupid]

daniela,

To be honest I don't think I completely understand the whole of your last post... I have read it through a number of times and I don't feel comfortable that I grasp exactly what you want to do, or why.

At this point I will freely admit my posts on this subject are perhaps not as clear, on all occasions, as I would like... and on others are overly expansive to the point of confusing the casual reader.

I can't take issue with any of your suggested technique for reaching the figure that you regard as the maximum you are willing to bid... I take all the aspects that you mention (and some more of my own) into consideration when calculating this also... Whatever it is we all do to reach that figure it invariably comes down to 'a penny more than that and I would rather someone else has it'...

Now, as I said, without fully understanding what you are proposing, tell me why a later snipe is, on average (or even in particular if you like), better than submitting that amount as your one and only final bid at 8 seconds before the auction ends? What other persons strategy are you trying to out manoeuvre to win the auction at the amount you are willing to pay or less?

Try as I might, and I have thought about it for many years, with many thought experiments about what others might be thinking/doing, and many thousands of observations of auction bidding, after the event, many that I have lost, but many also that I have won... I can't find a strategy whether involving multiple bids, or not, at an early or later stage in an auction that saves anything more than a small proportion of a bid increment, occasionally, over the strategy of using Gixen with an 8 second offset.

The strategy that occasionally saves that portion of a bid increment, involves submitting many hundreds of bids each less than a bid increment apart in the last few seconds of an auction... and is not something I would ask Mario to implement due to the huge costs in computing power required for such a small gain.

There is another strategy involving the use of multiple ebay accounts and not biding very late at all, which I won't go into further, but is, I think, illegal (termed 'auction manipulation') and could not be implemented by any legitimate process and would be relatively easy for ebay to detect in any case... but this last paragraph is going off the subject of later snipes which I think I should probably stick to on this thread.

[daniela]

Hi Mark,
normally people's usage of eBay is not that we bid on an auction, then forget about eBay for years, then bid again etc. Users bid on many auctions, not on a single one occasionally, and this is especially true for us, who use a sniping service because we dont want to waste time at the computer. Then IMHO in some cases it may make sense to think statistically.

The double snipe I use in some cases (I already remarked in many cases it makes no sense) as a sort of, if you want, "reverse insurance". I get an extra risk of paying an extra amount for the item with a small probability, which I want to minimize by not allowing anyone to be able to react. So far I do this only manually, because I am unaware of services that offer for free or at reasonable prices a late snipe and good reliability. In fact the sniping service landscape outside of Gixen is, if I may say, pretty dismal. The only other service I consider above decency, which is older than Gixen, allows -8s for free users (and -5s for paid service, which I never had). It is reliable at -8s and I never lost a snipe, I would not trust it at -2s. OTOH Gixen servers I would trust.

Here is what I do. First of all I set a "max I am willing to pay" which is the amount I am inserting in Gixen and I would continue to insert at -6s or -3s which means almost always (I have paid Gixen mirror service) it will go through. I compute it like this (formula is not mine, I found it years ago and am very happy with it):

Max value I give to the item + X + Y + Z

where
X= an odd amount to break ties, which I feel will be negligible in a month when I've got the item (as opposed to having lost the auction for a few pennies or dollars)
Y= (bid increment)*(expected other snipers +1)
Z=if I expect to have to wait n weeks to find the item again, I add n%

This is the "max I am willing to pay" that I find very good in minimizing the sorrows "I paid too much" or "I should have paid more and gotten the item".

If it is possible to me to waste time at the computer, or if it were possible I would do it via Gixen, I would do, for select auctions (I don't want to overdo, because Gixen already submits two bids, and the risk is that ebay requires confirmation codes and automatic snipes fail) that is for auctions on items that don't come up regularly, the "reverse insurance". Just before end of the auction, when no one can react, shoot a significant increase in the offer, not a crazy one, not one I would sorely regret, but an amount 5-20 times the "negligible" X amount above. In most cases this would change nothing at all. In a few cases one has to pay part of the extra amount one has bid or all of it - this in my experience happens rarely, in a few percent of cases, which means I have to shell out some money I did not want to, but, that money comes from a pile of money I saved thanks to sniping, and that money that I have to pay once in a while is my guarantee that I win almost all snipes and thus save time with searches etc etc. If of course I lose the auction, fair enough, another person was willing to pay well above my price, let him enjoy the item. I do not use last-second bids as only bids unless there is an abundance of the same item buy-it-now at an amount not far from "max I am willing to pay", in almost all cases I use them within a double bid strategy.

This is why I would welcome a button which allows another different offer to be sent very late in the game. Again I can understand you don't use and don't care for it, but I think there should not be apriori refusal to implementing - again, the work which is needed to implement is very little - those who don't find it useful, don't use. My opinion is that, like a supermarket has entire shelves of products I don't need or want and which other customers buy, surely Gixen can accommodate different approaches to bidding on eBay. I don't think I have "THE" perfect method or some genial breakthrough, only that this is how I do and I am very satisfied with it, if others find this post useful they are welcome to do the same, otherwise they can use their own strategy, with or without last-second bidding.

[Cupid]

daniela,

I don't think it is impossible to 'react' to a 6 second or even 3 second snipe as you suggest, I think it is pointless.

What you are describing is not actually a true reaction, in order to do it you must be prepared to do so... so this 'reaction' is actually prior to the 6 or 3 second snipe.

So in order to do what you describe you have to sit there at the end of the auction without having your true maximum registered with ebay (or even with a sniping service like Gixen) waiting for someone else to bid... does that really sound sensible to you?

If you think about it you risk losing the auction, to a bid that is evidently less than your maximum bid, far more often than placing your true maximum bid earlier than 6 seconds before the auction ends.

... and that is why I do not think last second snipes are in the interest of any Gixen user... no matter how reliable the service could be at placing such snipes.

Actually it would be better for me as a buyer if more people did use last second snipes (while I continue to use earlier ones)... I would win more auctions... so my self interest would be to agree with you... I don't for the good of the Gixen community as a whole.

It is a fallacy for people to believe that late is good therefore later must be better... please provide some case studies or even scenarios where I am wrong and I'll gladly be convinced, or continue the discussion... but no one has done so yet.

[daniela]

Hi Cupid
To me, the key question in deciding the delay is, what is the shortest time which still gives good reliability from my connection (if manually) or the server(s) (if sniping service) that the bid goes through. In fact I may even be willing to compromise on last second reliability, by implementing a double snipe.

You think it is impossible to react to a -3s bid? You are entitled to your opinion but it actually is (open new window and bid, but don't confirm until -2s or -1s) You think it is stupid to use very-last-minute bids? No one is insisting you do, of course. And for certain auctions it indeed is stupid, but not always and in every case. My modest opinion is that the Gixen servers are well able to deal with (a small percentage of) -2s or -1s snipes, at least if there is no flood on ebay at that time; that Mario would hardly have to waste any work on that; and that no one is forcing use of this option, it would simply be one extra option for those who want to use it when they want to use it. I have sniped manually, or double sniped manually, at -1s and even at a time ebay records as auction end time.

[Cupid]

Mario already bowed to pressure from users to provide a 3 second option with the safeguard that a Mirror server is also required to back up the snipe... not because he thought it gave an advantage over other sniping sites.

So if an even later bid is not an advantage to the users of Gixen why would he bother to bow to pressure to enable a worse service to the community?

In my view, a service that takes the easy option of providing everything people ask for without explaining what is and what is not to their advantage, is not treating its' users with the respect for their intelligence that they deserve.

[jkanel]

Another sniping tool is entering the bid 2 seconds before the end - gixen is 5 seconds before the end. Will Gixen adjust? Who is the the other company?

[Cupid]

[1argo]

The later bid in last second does not give a time for other sniper to outbid my bid.

[Cupid]

I'm always quite calm when posting on here 1argo.

I'm just not convinced that the provision of even later bids would improve the Gixen service.

I firmly believe that it would cause more problems than the advantages people perceive it may have.

If sniping with a 3 second offset works best for you, I am pleased for you... for the auctions I take an interest in it doesn't, unfortunately... otherwise I would also use it.

I know why it doesn't work for me, it is because I bid on reasonably high value items where there are quite a few other snipers prepared to bid, but we all agree quite closely what the real value to us is... and if the price goes above that lowest market rate non of us are interested at all. In that environment sniping is essential but the earliest sniper wins more often than the later one.

So, in a similar fashion, can you explain why a later bidding time works better for you than it does for me?

[1argo]

If this makes you calm I will tell you. When I was bidding with free version 6 seconds I had plenty of loosed bids.
Now with 3 seconds I cant complain :wink: And it is not about the mirror server.

[Cupid]

There isn't a technical problem with providing 2 or even 1 second snipes... it is just pointless to do so as it does not increase your chances of wining.. on fact I think the evidence points towards it making it less likely that you will win with 2 second snipes than with 6 second ones.

As I've said, if you can provide any evidence that can support your case I would be interested to hear it... ?

The fact that many bids are placed at the end of an auction is evidence that many people employ a sniping technique, not that they react to other bids that are also placed in the last few seconds.

Multiple bids of the same amount can be placed in the same second, this is much more common in the last seconds if an auction and happens often when you have a mirror account with Gixen and schedule snipes with the same offset on both servers, as I do. There are other reasons why two bids of the same amount can be placed at different times, by different users, all quite normal and to be expected from the ebay bidding system, but not evidence that sniping in the last second is better than sniping a dew seconds earlier.

Yes, sniping is a more successful technique than regular bidding on ebay because people can and do react to bids placed hours before the auction ends... they can only react in the last 6 seconds if they already had the intention of doing so, which isn't really a reaction at all, it is the same as scheduling a higher snipe with an automated service like Gixen.

Price jumps in the top two bids of the history are usually the result of the sellers having reserve prices on their auctions... at lower levels on the bidding they are quite normal and to be expected when people bid rationally and place their maximum bid only once.

[1argo]

I have been trying the regular auto bidding on ebay with big amounts and actually I had won 0 auctions. So I am not so familiar how it works exactly but IMO the regular auto bid is pointless on ebay. Almost on every auction it can be seen for the last 3-4 seconds how the price changes 3 times ! How can be explained this ?
And also I have seen on my bidding history often 2 times same price appear with 2 bids, how is possible ?? And sometimes if the price step for bid is 0.50$ and the moment bid is 5$ the next final winning bid is 9$, how it jumped from 5$ to 9$ ?

For evidence the only way is to try other bid sniper with less seconds but most of them are too expensive. Also I didint understand where is the problem to make 2 seconds bid for testing ?

[Cupid]

1argo,

I refer you back to the earlier posts in this thread... There is no evidence that later bids than 6 seconds before the end of the auction increases your likelihood of wining.

Remember that the only thing that sniping provides an advantage over is people that change their mind about the amount they are prepared to bid after you have placed your bid and as a result of your bid appearing on ebay.

People that have already decided to bid more than you, at whatever time their bid is placed will invariably win... in fact, the only slight advantage that you can gain over them is to bid earlier, not later, because that way the bid increment rule might just work in your favour.

As I've already said, if anyone has any evidence that supports later bidding I'm happy to continue the debate... but once again you haven't provided anything other than anecdotal evidence, which might seem convincing to you but has been stated as if it proves the fact for so many years that people start to believe it... when it isn't actually supported by any real evidence.

[1argo]

Hi, Congrats for the gixen. I currently using it online from 2 years with free version and after I paid for mirror version is really nice and works very good. For so long time Only 2 times gixen server didnt work and just didnt bid.
The problem is that recently I see more oftenI loose some items because there is last second bid from someonelse... It is frustrating for me.
I suggest better making an option with 1-2 seconds with indicate popup menu alerting is possible to miss the bid. This is much better in my case.

Also I was looking for other soft. snipers with less seconds but didnt find any good. I am pretty happy with all of gixen and dont want to change it.

[shunshanmao]

Hi, I have used GIXEN since one week ago, congratulations and thank you for an excellent job!

For me a snipe from five seconds before auction ends is fine, I have seen winning bids placed at the last second, but there is a big risk to lose the auction for timing reason.

Another aspect is about bid range, most of bidders place bids according to shipping fee, so if actual bid is $13.50 and shipping is $5, you can think the highest bidder placed $15 because he wants to pay $20 for the total, then bidding $15.01 you will save $0.49 instead bidding $18, because your winning bid would be $15.50. Other bidders place rare amount like $15.72 to prevent this type of thought. Personally in this example, I like to bid higher than $15, like $23.99, I prefer to pay more than lose the item, so placing an high bid on five seconds before auction ends, you will be the winner, of course, you have to search previously what is the average price on ended auctions, to set your maximum bid. :D

[Cupid]

I can't argue with that Amazed, but I would like to assure everyone that my intention when sniping auctions is to win them at the lowest achievable cost.

Further, my advise on best practise is aimed at enabling others to do the same.

[Amazed]

Mario, Mark
I'm convinced you fellas are looking at this all wrong.

I know with absolute certainty.. that if the bid was timed one second AFTER the end of the auction... I'd save myself a lot of money that I otherwise spend on EBay

[Cupid]

I will add another couple of thoughts to the debate, in a way I have not done so in the past:

In the real world if you were a sniper with a gun, pitted against another with the same weapon, would it be better to shoot first or second?

Bearing in mind that an enforced cease fire is imminent, if the other sniper does not know you are there how long does it take him to load his weapon and fire back at you after you have taken the shot, whether or not you even hit them?

[Cupid]

Yes, it is a little tiresome to constantly have this debate... can I suggest to those that wish to continue this discussion that they search this forum and familiarise themselves with the various arguments involved.

After all these years both Mario and myself remain of the same opinion that snipes placed later than 4 seconds before the end of the auction are of no benefit to winning those auctions... in fact, I go further and say that it is counter productive to do so, due to the bid increment and first bid rules.

I am happy to respond to well reasoned argument on the subject... but no longer to blanket statements that things would be 'better' if later snipes were possible.

I and Mario have refuted all such arguments for many years now.

[mario]

For the millionth time, anything less than three seconds is completely pointless.

[rh56-de]

Its nice but a little too slow. 2 seconds before end may better. :( :(