|
| 25 Nov 2012 04:26 PM |
I'm not posting this in LMaD or Scripting Helpers because this is more of a computer science question than anything else.
On Roblox's TC (Trade Currency), the deals that are least profitable for the person offering the deal are always at the top of the pile of offers. Only the top offer is ever actually fulfilled, and once a deal is fulfilled, the next-least profitable trade arrives at the top.
In other words, we have to deal with a trade-off: deals that are less profitable will sell faster, but by definition are less profitable than more "aggressive" deals, which don't get fulfilled very quickly (if at all).
Is there an optimized way to find the middle ground? I imagine such an algorithm would need to account for the rate at which people are trading, the rates of other offers, and the "size" of other offers.
I've been looking around online and I haven't found anything useful yet. |
|
|
| Report Abuse |
|
|
|
| 25 Nov 2012 04:30 PM |
| I'm not sure I completely understand what you mean by "middle ground". If it is what I think it is, its a bad idea. |
|
|
| Report Abuse |
|
|
|
| 25 Nov 2012 04:33 PM |
What I'm saying is that if I make a bunch of deals that are not very profitable, I sell them very quickly but I make very little money on each deal. If I make fewer deals that are more profitable each, I make more money on each, with the consequence being that I make these deals slower since they're going to be lower on the pile.
I assume that there's a way to find a "perfect" algorithm that balances profit per trade with speed of trades for a maximized profit over a long span of time. |
|
|
| Report Abuse |
|
|
|
| 25 Nov 2012 04:50 PM |
| I've found that if your deal isn't the best one on the list, it's never going to be accepted. Good deals are quick deals. It's better to make good money fast than to wait for a miracle. |
|
|
| Report Abuse |
|
|
lah30303
|
  |
| Joined: 15 Feb 2008 |
| Total Posts: 10027 |
|
|
| 25 Nov 2012 06:51 PM |
I've found it depends on the size of what the first deal is. But it's rare that it would be as far back as the third deal.
However, a theory of mine is to find a time when a lot of TC profiteers are trading, then put up a large number of very small trades, which would cause other TCers to post new trades ahead of yours, and drive the value up of the other currency without you having to make too many trades. |
|
|
| Report Abuse |
|
|
|
| 26 Nov 2012 05:52 AM |
You need to be at the top to get anything.
What you want is an algorithm to trade at times when the rate is good.
Like give it x money and y time and it tries to get max profit based on recorded data about trade rate fluctuations. |
|
|
| Report Abuse |
|
|
lah30303
|
  |
| Joined: 15 Feb 2008 |
| Total Posts: 10027 |
|
|
| 26 Nov 2012 04:12 PM |
| You don't need to be at the top. There are many scenarios where being at the top is a bad idea and sometimes it can lose you money. I thought this was obvious. |
|
|
| Report Abuse |
|
|
|
| 26 Nov 2012 04:34 PM |
| It was obvious to the point where no one needed to say it. |
|
|
| Report Abuse |
|
|
|
| 26 Nov 2012 06:11 PM |
It's a race to the bottom.
The optimal strategy is to put top trades on both sides of the book at very high frequency.
This obviously drives the spread to 0. Detect and backoff when the spread is negative.
Whoever can do this fastest wins.
Optimal means "the best thing you can do assuming everyone else is as smart as you". |
|
|
| Report Abuse |
|
|
|
| 26 Nov 2012 06:12 PM |
| http://en.wikipedia.org/wiki/High-frequency_trading |
|
|
| Report Abuse |
|
|
|
| 26 Nov 2012 06:26 PM |
| So it's basically a mix of chicken and gambling. I'm up for some of that. |
|
|
| Report Abuse |
|
|