Cut back to 3 digits

Even a thousand stocks at 0.0001 only comes to 10 cents, so it makes no sense at all to have that fourth digit; it doesn't make the market any more accurate. All that fourth digit does is enable one of the most infuriating traders on ipredict to ruin other people's enjoyment of the site by queue-jumping them (outbidding by 0.0001).

I suggest running a few stocks with 3 digits and seeking feedback from ipredict users. If users prefer 3 digits, make that the norm. This would probably mean doing away with the markers at 0.0001 and 0.9999, but are they really necessary anyway?

admin in the past has indicated their intention to do this. But nothing ever happened.

I don't think dropping the last digit is going to stop the pond-scum behaviour anyway. They'll just place bids that are 0.001 cent above/below instead.

I agree with Lanthanide, it's not going to stop the behaviour. But it would at least make people feel better knowing that they were out bid by 0.0010 and not 0.0001. So I support the change to 3 digits.

MaxTheCreator Wrote:
-------------------------------------------------------
> I agree with Lanthanide, it's not going to stop
> the behaviour. But it would at least make people
> feel better knowing that they were out bid by
> 0.0010 and not 0.0001. So I support the change to
> 3 digits.

And 2 digits would make them feel better than 3 digits, and 1 digit would make them feel better than 2 digits.

I know this is reductio ad absurdum, but I don't think the argument has been made as to why 3 not 4 or 5 or 2 digits is the correct level.

Changing the bid increment feels like the wrong solution to an ill-defined problem.

2 digits is going to far in my mind. With 2 digits there would be only 99 possible prices to trade at. And you would have situations like this: You want to place a buy order at 0.60, but there are already at least 20 bids on each price from .55 to .63. So you either have to get behind someone at one of these prices or place your order at .54 or .64. Sure it would work ok, but it could get quite crowded at certain prices making it frustrating to get the price you want. I think 3 digits is the perfect balance.

The problem has been well defined, anyone who has ever placed a sitting order has run into it. I also gave clear reasons why 4 digits is unnecessary. What exactly is the downside to reducing to 3 digits?

Why argue?

I suggest splitting the difference and going for 3.5 digits.

We could then apply Swedish rounding for cash transactions (unless the trader's computer is in the Northern Hemisphere in which case, it will come up the plughole rather than down of course).



Alan.

@Alan, I don't whether to tell you to get a grip. Or, ask if you've already got a grip.

So 3.5 digits would effectively be a 4 digit display, but the final digit could only take the values of 0, 2, 4, 6 or 8?

Mr Oracle: I hash a very firm grip on the bottle in my handsh thank very you much!



Mr Lathnadine Thannalide Landthine

Mr L: 3.5 digits would mean it could end in any of the first five letters of the alphabet, as long as they aren't vowels. On a Friday.


Alan.

Alan3285 Wrote:
-------------------------------------------------------
> 3.5 digits would mean it could end in any
> of the first five letters of the alphabet, as long
> as they aren't vowels. On a Friday.

WTF...

I think I'd rather see 2.5 than 3.5, assuming it worked in the manner descirbed by Lanthanide.

MaxTheCreator Wrote:
-------------------------------------------------------
> Alan3285 Wrote:
> --------------------------------------------------
> -----
> > 3.5 digits would mean it could end in any
> > of the first five letters of the alphabet, as
> long
> > as they aren't vowels. On a Friday.
>
> WTF...

Only a Friday mind you. Its gets really messy when people don't pay attention to that.... vowels in all directions, spelling all shot to pieces. War, hunger,... thirst... speaking of which, whose round is it?

Alan.

MaxTheCreator Wrote:
-------------------------------------------------------
> 2 digits is going to far in my mind. With 2 digits
> there would be only 99 possible prices to trade
> at. And you would have situations like this: You
> want to place a buy order at 0.60, but there are
> already at least 20 bids on each price from .55 to
> .63. So you either have to get behind someone at
> one of these prices or place your order at .54 or
> .64. Sure it would work ok, but it could get quite
> crowded at certain prices making it frustrating to
> get the price you want. I think 3 digits is the
> perfect balance.

Do you see your own inherent contradiction now?

You are essentially unhappy about people cutting in line when trading is at 4dp, which is fair enough. However if we moved to 2dp you would be unhappy having to get in line behind someone else at any given price because they happened to put an order in before you.

So aren't you effectively saying "we should move to 3dp so I can cut in line".

TraderV Wrote:
-------------------------------------------------------
> You are essentially unhappy about people cutting
> in line when trading is at 4dp, which is fair
> enough. However if we moved to 2dp you would be
> unhappy having to get in line behind someone else
> at any given price because they happened to put an
> order in before you.
>
> So aren't you effectively saying "we should move
> to 3dp so I can cut in line".

You have a point. I't may be a little contradictory, but it's how I feel.

TraderV Wrote:
-------------------------------------------------------
> MaxTheCreator Wrote:
> --------------------------------------------------
> -----
> > 2 digits is going to far in my mind. With 2
> digits
> > there would be only 99 possible prices to trade
> > at. And you would have situations like this:
> You
> > want to place a buy order at 0.60, but there
> are
> > already at least 20 bids on each price from .55
> to
> > .63. So you either have to get behind someone
> at
> > one of these prices or place your order at .54
> or
> > .64. Sure it would work ok, but it could get
> quite
> > crowded at certain prices making it frustrating
> to
> > get the price you want. I think 3 digits is the
> > perfect balance.
>
> Do you see your own inherent contradiction now?
>
> You are essentially unhappy about people cutting
> in line when trading is at 4dp, which is fair
> enough. However if we moved to 2dp you would be
> unhappy having to get in line behind someone else
> at any given price because they happened to put an
> order in before you.
>
> So aren't you effectively saying "we should move
> to 3dp so I can cut in line".


Talking of contradictions, one doesn't have to get in line behind anyone else who has put in an order before them, one can move to the front of the line even if by as little as .0001. If fact moving to the front of the line could be said to be fundamental to the whole process.

I think I *would* prefer 2 digits to 4 actually... but 3 is about the right balance.

2 digits would actually work if all orders were put into a pot and fulfilled serially, eg if 4 people have buy orders for 10 each, and someone else came to sell 20 stocks, each person would get 5 orders filled.

Thinking about this in more detail....

Scenario:

Albert puts in the top bid for 100 units at 0.5000, best sell is 0.8000.

Berty also wants some of those units, thinks 0.8000 is too high, but 0.5000 is a bargain, what is the rational price for them to bid? Let's assume they aren't sure, so they play it safe and bid 0.5001

Albert doesn't want to be outbid by such a paltry amount, so he goes to 0.5002 and thus begins a sub-penny bidding war.

Question 1: What is the rationale for Albert not putting in the bid (substantively) higher in the first place?

Question 2: If Albert did put in a bid at his highest possible purchase price, would that be irrational (from a profit optimisation persepctive)?


Thanks,

Alan.

2 digits would actually work if all orders were put into a pot and fulfilled serially, eg if 4 people have buy orders for 10 each, and someone else came to sell 20 stocks, each person would get 5 orders filled.

Why is a 'pot' fulfilled serially better than a you snooze you lose scenario?



Edited 1 time(s). Last edit at 02/10/2012 09:20PM by TraderV.

Lanthanide Wrote:
-------------------------------------------------------
> 2 digits would actually work if all orders were
> put into a pot and fulfilled serially, eg if 4
> people have buy orders for 10 each, and someone
> else came to sell 20 stocks, each person would get
> 5 orders filled.

Interesting, I quite like that.

TraderV Wrote:
-------------------------------------------------------
> 2 digits would actually work if all orders were
> put into a pot and fulfilled serially, eg if 4
> people have buy orders for 10 each, and someone
> else came to sell 20 stocks, each person would get
> 5 orders filled.
>
> Why is a 'pot' fulfilled serially better than a
> you snooze you loose scenario?

Because at the moment someone who puts a bid in can completely lose out due to some jonny-come-lately that overbids them by a functionally-worthless amount. At least this way the jonny-come-lately doesn't completely deprive the original bidder from a transaction.

The tricky part of this comes in dealing with bids that don't evenly line up:
Say one person places an order for 50 and another places an order for 10. If someone comes along and places a matching order for 20, should the 2nd person have all 10 filled, or should it be split 5:1 ratio so the 2nd person buys 3.33 and the first person buys 16.66? Do you do handle the rounding?

If we assume trading at 2bp

If someone has a buy of [email protected] in the order book and then another person comes along and orders another [email protected].

Then a sell comes in to offer [email protected]

In my view the person who put in the first bid should be completely filled (as it is now) before the latter bidder. I don't see any grounds whatsoever to serve 25 to the 'latecomer'. Indeed your system would encourage bidding at the last minute, which seems to be one of the things you are railing against?



Edited 1 time(s). Last edit at 02/10/2012 05:42PM by TraderV.

TraderV Wrote:
-------------------------------------------------------
> If we assume trading at 2bp
>
> If someone has a buy of [email protected] in the order book
> and then another person comes along and orders
> another [email protected].
>
> Then a sell comes in to offer [email protected]
>
> In my view the person who put in the first bid
> should be completely filled (as it is now) before
> the latter bidder. I don't see any grounds
> whatsoever to serve 25 to the 'latecomer'. Indeed
> your system would encourage bidding at the last
> minute,

Although I haven't agreed with everything TV has said here, I do agree with this. At equal values, first come, first served.

Yes, I totally agree with this, else you'll quickly lose liquidity as people pull bids, and attempt to 'snipe' at the last minute.

Alan.

I used to support the status quo of four digits but I now support 3.

You want a balance of the increments being sizeable enough that there isn't ridiculously small outbids, while at the same time ensuring it's granular enough for a smooth market.

3 digits is good.

PM.2014.LABOUR (as of writing) would then look like:

Buy Orders
10 $0.517
20 $0.500
10 $0.496
35 $0.490

Sell Orders
1 $0.525
50 $0.530
20 $0.533
40 $0.548
10 $0.550

Outbidding by 0.001 is sizeable enough to 'deserve' to go first in my opinion.

--

Another point which I don't think has been brought up is the bid-ask spread. In the forex market, it's so liquid and tight that going to 5 digits is necessary. Here, the spread is often 1 to 2 cents, cutting the bid-ask spread by half a percentage (buy outbidding by 0.0001) point offers next to no liquidity improvement. Cutting the spread by 5% does (by outbidding by 0.001).

Yes, we would still have idiots always outbidding by 0.001, but at least now we are extracting some tangible liquidity from them whereas before we weren't. In fact, these people would probably start being encouraged!

Given current liquidity/volume, 3 digits is the right amount.

Also, there is no material difference between 51.01% and 51.02%, no one things down to that level of detail when they estimate the odds of an event happening.

This may be a dead topic, but I think Eeyore said it best:

"I know this is reductio ad absurdum, but I don't think the argument has been made as to why 3 not 4 or 5 or 2 digits is the correct level.

Changing the bid increment feels like the wrong solution to an ill-defined problem."